Results

Summary of Results

For our 2024 project, we conducted both Dry Lab and Wet Lab experiments, and achieved results in two major categories: (1) Discovery and (2) Application of satellite phage. 

Our results for the discovery of novel satellite phage are split into two categories: (1a) bioinformatic discovery and (1b) physical discovery. 

  • In discovering novel satellites bioinformatically
    • We collected data from our runs of our novel satellite phage identification software, SaPhARI.
  • In discovering novel satellites physically we screened for satellites in E. coli, M. smegmatis, and M. aichiense, and sequenced phages discovered. 

Our results for Applications of Satellite Phage are separated into three categories: (2a) gene delivery, (2b) overcoming prophage immunity, and (2c) host expansion. In these categories, we conducted experiments at the in silico, in vitro, and simulated in situ levels in our model colon and soil microcosms. 

At the in silico level our results include:

  • The predictions of the mathematical model to describe the functionality of phage satellites in simulated real-world environments.
  • We collected results from a phage-host interaction predictive model with an extensive phage-host interaction database.

At the in vitro level our results include:

  • Results of gene delivery:
    • Quantifying transducing units in both nonreplicative and replicative systems.
    • Testing killing effect of kanamycin targeting cosmid.
    • Testing silencing effect of dCas9 cosmid.
    • Testing fluorescence output of RFP cosmid.
  • Results of ability to overcome prophage immunity:
    • Resurrecting Mycobacterium aichiense phage satellites, termed “phagelets”, from six-year-old phage lysate and characterized these satellites.
  • Results of phage host range expansion:
    • Plating phage lysate to test the infection range of phages HerbertWM and a novel mycobacteria phage.
    • Assembling chimeric tail fibers that can theoretically expand host range of HerbertWM via the phagelets.

At the simulated in situ level our results include: 

  • Quantifying effects of transducing units:
    • Sampled the effluent of the model colon to test for a killing effect model colon with the CRISPR transducing units.
    • We inoculated soil microcosms 5-8 with the RFP transducing units and conducted bacterial plating to test for fluorescent output.
  • Results of overcoming prophage immunity:
    • Conducted phage plating to test for a killing effect and persistence over time after inoculating soil microcosms with Mycobacterium phage satellites

For detailed information about our experiments, please visit our experiments page.

Discovery

We discovered new phage satellites with two methods, bioinformatic and physical, and developed novel protocols for discovery which can be used by other laboratories. 

Bioinformatic Discovery

Experiment: In order to gain deeper insight into the microbial dynamics of satellite phage in natural environments, we took advantage of the wealth of available metagenomic sequences to categorize the extremely diverse and ubiquitous satellite phage families, while simultaneously deepening our understanding of satellite genetic structure We searched through metagenomic samples for satellite phage protein clusters using SaPhARI, assessing its ability to find and tag protein clusters.

Results: Identified Protein Clusters

Using SaPhARI at varied protein parameter stringencies, we searched through 7,659 metagenomic contigs, which were less than 20,000bp, representing 37 metagenomic samples. We identified 87 putative satellite protein phage clusters representing 5 out of the 6 developed satellite family models:

  • PICMIs - 41
  • cfPICI - 21
  • PICI - 2
  • P4like - 2
  • Phagelet - 1

The only family not identified were PLE’s, which are only categorized in Vibrio cholerae.

Summary

For each satellite model, the parameters are defined in the “proteins” variable, which is a Python list of potential proteins that each respective satellite family is expected to contain. For example, the above PICI model has 5 protein parameters —primase, integrase, decor, either “major head” or “major capsid”, and either 'alpa' or 'icd-like'—. To define stringency in our experiment, we changed the “minnumber” variable for each model, which defines the minimum number of these proteins that need to be present in order for SaPhARI to tag the cluster as that satellite family. For the satellite group defined in photo above, the protein cluster will be identified as a PICI if it has 3 of the 5 protein groups per defined in the model. A stringency score of 0 would define this minnumber variable as 5 and a stringency score of -1 would define this minnumber variable as 4.

We ran five iterations of stringency, decreasing the minimum required proteins by one in each run until the models reached or exceeded half of the total proteins. The stringency scores were: 0, -1, -2, -3, -4, and -5. Not every model was run with the higher stringency scores, as the highest stringency score used for each model never exceeded the equivalent of ½ of the number of protein parameters defined in the model. For example, with the PICI model, the highest stringency model used was -2, thus requiring 3 out of 5 proteins to be present.

PICMI had the most lenient model, with three potential protein parameters—primase, integrase, and either 'alpa' or 'icd-like'—all located within 15,000 bp of each other. 4 putative PICMIs strictly adhered to the requirement of all three protein parameters, while the remaining 37 were found with a stringency score of -1, requiring only two of the parameters.

2 putative PICIs were detected at a stringency score of -2, with three of the five possible protein parameters to be present within 15,000 bp of each other, making it the second least stringent model. The P4-like family required the highest stringency for detection, at -5, due to having twelve possible protein parameters within 12,000 bp, making it the strictest satellite family model. The 2 putative P4-like satellites were detected at a stringency score of -5.

A single cf-PICI was found at a stringency score of -3, with seven of the ten protein parameters to be present within 15,000 bp of each other, making it the second most stringent model. The remaining 20 cf-PICI’s were found at a stringency score of -4. Lastly, the sole Mycobacterium phage satellite, "phagelet," was found at a stringency score of -3, requiring three of the six protein parameters to be present within 12,000 bp of each other.

Data Analysis:

To assess whether putative satellites could be full phages, we analyzed all contigs containing identified satellites using PHASTEST to infer prophage status, as well as analyzed each contig using BLAST to assess similarity to known species. The results were compiled into a CSV file accessible below:

Click here to download CSV

Within the 82 contigs analyzed with PHASTEST, 44 prophage/phage regions were identified. Of these, 20 were classified as "complete" by the PHASTEST algorithm. Notably, 12 of the 20 "complete" prophage regions were identified as cfPICIs. Additionally, 14 regions were categorized as "questionable," spanning various satellite families—5 were cfPICIs, 8 were PICMIs, and 1 was a “phagelet”. All 9 regions classified as "incomplete" were identified as PICMIs.

Interestingly, 37 of the 43 satellites not detected by PHASTEST were PICMIs. This was expected, as 29 of the 41 PICMIs had a query coverage of less than 10% for their top hit in a nucleotide BLAST, highlighting how poorly characterized these phage satellite sequences are. Furthermore, only 18 of the 82 metagenomic contigs analyzed by BLAST had both query coverage and identity greater than 90%, further demonstrating the lack of characterization in satellite phages.

Reliance solely on nucleotide similarity can yield limited results and overlook numerous satellite phages hidden within bacterial genomes. SaPhARI addresses the urgent need for novel bioinformatic approaches that move beyond nucleotide similarity.

All contig fasta files, original SaPhARI outputs, contig BLAST results, and PHASTEST results are available in our GitLab within the folder titled “METAGENOMIC_SATELLITES”

Intepretation:

The discovery of these 87 putative satellite phage clusters only scratches the surface of the capabilities that SaPhARI has to offer through the identification of satellite protein clusters within metagenomic data. Our team continues to analyze more metagenomic data daily, seeking to characterize more satellite phage systems for future applications within synthetic biology.

Physical Discovery

Phage satellites, though incredibly abundant at an estimated 3.2 x 1026, are widely uncharacterized (Eppley et al., 2022). Furthermore, comprehensive protocols to screen for phage satellites have not been developed. We developed procedures to screen for phage satellites using bacterial lysogens and applied these in Mycobacterium and E. coli. To discover phage satellites, we used bacterial lysogens, to screen for satellites that can exploit the prophage’s packaging machinery. The general steps to discover a phage satellite include 1) obtaining/creating a lysogen, 2) collecting a soil sample, 3) enrichment plating, 4) purifying the potential satellite by plaque assays and 5) Phage DNA extraction by PCI. 

Phage screening does not typically use bacterial lysogens; however, to screen for satellite phages that can exploit a prophage’s packaging machinery, we used bacterial lysogens. The general steps to discover a phage satellite include 1) obtaining/creating a lysogen, 2) collecting a soil sample, 3) enrichment plating, 4) purifying the potential satellite by plaque assays and 5) Phage DNA extraction by PCI.

Experiment: Applying our phage satellite screening procedure, we discovered two novel bacteriophages that infect E. coli strain EMG2 K-12, a tailed bacteriophage and filamentous bacteriophage. These discoveries exemplify how our detailed screening protocols can lead to new, unexpected discoveries, particularly by our discovery of new E. coli bacteriophages despite how well-studied E. coli is.

Summary

Figure. Experimental process from collection of soil sample 1, which showed plaquing on enrichment plate, to final phage genomes obtained from sequencing. 

Summary

Figure. Experimental process from collection of soil sample 4, which showed plaquing on enrichment plate, to final phage genomes obtained from sequencing. 

Raw Data:

DNA isolation of phages were performed after a series of steps to purify the phage from the original enrichment plates which showed plaques (see Figures 1 and 2). Double stranded phage DNA was obtained by performing a DNA extraction of this lysate with phenol/chloroform/isoamyl alcohol. More detail on the experimental process can be found on the Experiments page.

DNA samples were sent to the Oklahoma Medical Research Foundation for Illumina Sequencing. Raw sequencing data was assembled using the Shovill pipeline for de novo assembly of phage genomes (Turner et al., 2021). To look for phage, the raw sequencing data was then assembled into contiguous sequences using the Shovill assembly pipeline for The Shovill pipeline allowed assembly of our phage genomes by allowing us to specify contig length and implementing error correction. DNA sample 1 resulted in 12 contigs of various sizes. DNA sample 4 resulted in 2 contigs of 7,432 bp and 31,664 bp.

Putative E. coli filamentous phage raw sequence
Putative E. coli tailed phage raw sequence

Data Analysis/Interpretation:

Novel Putative Enterobacteria Filamentous Phage

We identified a novel 7kb filamentous phage that infects E. coli strain EMG2 K-12 by isolation from soil samples on W&M campus. 

Unexpectedly, this putative phage was likely present in both initial soil samples collected. Both DNA samples showed nearly-identical contiguous sequences of about 7,000 bp (7432 and 7420) after assembly of the reads. Also, both samples showed identical bands around 9kb on a 1% agarose gel. Though the size appears to differ, filamentous phage genomes are single-stranded and circular, and therefore, may appear higher on a gel than expected by size (Hay 2019). Together, these results suggest that both soil samples contained this 7kb phage.

Summary

Figure. E. coli filamentous phage Benchling linear map with Prokka protein annotations. 

Prokka genome annotation of the 7kb contig suggests that the putative phage contains key functions for filamentous phage function from Enterobacteria filamentous phages: f1 phage, M13 phage, and IF1 phage. The virion export protein gene, involved in filamentous phage extrusion, and gene 1 proteins share homology with that of f1 phage; attachment protein G3P and head virion protein genes share homology with that of phage M13; and the capsid protein G8P and DNA-binding protein G5P share homology with that of IF1 phage. This discovered phage appears similar to the Ff filamentous phage family, which have M13 and f1, which share 98.5% genetic similarity (Mai-Prochnow 2015, Hay 2019). The M13 phage has important applications: half of the 2018 Chemistry Nobel prize was awarded for “phage display of peptides and antibodies” which primarily uses filamentous phage M13 (Barderas 2019).

The plaque morphology of this phage during purification further supports our hypothesis that we have discovered a filamentous phage. When undergoing plaque purification, the phage from sample 1 showed decreasing numbers of plaques, which are transparent areas in a bacterial lawn indicating phage lysis of bacteria. Initially, assuming that we were working with a tailed phage, we hypothesized that the phage was degrading in lysate form. However, while tailed phage lyse bacteria and can be detected by plaques, filamentous phage do not lyse their bacterial hosts. Filamentous phages instead extrude from their hosts through channels in the bacterial membrane, and as such, filamentous phages plaque at varying degrees as a result of reduced or slower bacterial growth (Gavric 2022). If this phage is a filamentous phage, a possibility is that the phage titer decreased considerably by the third round of purification, resulting in no observation of plaque, despite the phage infecting E. coli at this titer.

Summary

Figure. At dilutions of 10-1 to 10-11, phage plating of dilutions of the phage from soil sample 1 showed no plaques in the third round of purification. Since discovering that this phage is a filamentous phage, we may question if the discovered filamentous phage was present on these plates but simply did not show plaques.

Novel Putative Enterobacteria Tailed Bacteriophage

Additionally, we identified a novel tailed bacteriophage that infects E. coli strain EMG2 K-12. 

Summary

Figure. E. coli tailed phage Benchling linear map with Prokka protein annotations. 

We identified a 31kb contiguous sequence present in the DNA isolated from soil sample 4 as a bacteriophage based on the presence of a capsid protein, virion proteins, and tail proteins. This bacteriophage has not been previously isolated–using NCBI Blast did not yield any phage with high similarity. Yet, this phage may be present in multiple E. coli strains as a prophage. A NCBI Nucleotide blast of the phage genome resulted in a 97.95% identity match and 84% query cover with E. coli strain ETEC1721, and high similarity to other Escherichia species and E. coli strains including Escherichia albertii Strain sample 168, E. coli strain RHB39-C23, and E. coli strain E2528-C1.

Experiments: We created lysogens in Mycobacterium smegmatis in order to screen for satellite phages. To test lysates potential to form stable lysogens, we performed a lengthy and robust procedure, testing the titer, phage release, and superinfection efficiency.

Raw Data:

Lysogen Data Table. Cells highlighted in blue represent phage names while gray cells are the isolators' initials. If a lysate fails one of the tests, it can be repeated up to 3 times, with yellow representing 1 failed attempt, orange representing 2 failed attempts, and red representing 3 failed attempts. If a lysate passes a second or third attempt, the lysate continues to the next experiment.

More detail on the experimental process can be found on the experiments page.

Data Analysis/Interpretation:

Out of the 55 tested lysates, five passed all of the lysogeny tests and resulted in a stable lysogen, ready to be used for satellite screening. We successfully created 5 stable lysogens of phages JollyBabz, StubbySaigey, KingChadwick, and DocMcStuffins, and SucetteDeLMort in M. smegmatis.

Experiment: We screened for novel satellites in Mycobacterium aichiense, a strain of bacteria that conveys strong immunity to other phages due to the presence of a prophage called HerbertWM.

Raw Data:

3 phages, including 2 potential satellite phages, one discovered by Phage Lab and one of the novel “phagelets”, and the standard phage were sent for sequencing to better characterize these phage. We received Illumina short, paired-end reads for each sample, which were evaluated and trimmed to ensure the highest possible quality for further assembly. Quality and adapter trimming was performed utilizing FASTQC and trim_galore, while assembly was performed utilizing the Shovill assembly pipeline.

Data Analysis:

Using the phage satellite screening procedure that our iGEM team developed, W&M Phage lab successfully discovered a putative phage satellite that infects M. aichiense. However, characterizing this putative satellite has proved difficult thus far. Several attempts at sequencing show that a large part of the putative satellite’s genome aligns with a pathogenic bacteria called Giardia, which is not only a different species but also a different phyla than M. aichiense.

Importantly, our team has characterized a group of putative satellite phages that infect M. aichiense, which we term “phagelets”, by increasing their titer and sequencing one of these phage satellites for further analysis. These putative satellites were discovered over 6 years ago by Jamestown High School students, but the phage lysates were thought to be degraded and were not initially characterized.

We have additionally characterized a novel M. aichiense standard phage, also discovered previously by Jamestown High School students. This phage bypasses homo- and hetero- immunity in M. aichiense and lyses the bacteria without excising the prophage HerbertWM.

In our soil microcosms, we utilized both the novel M. aichiense “phagelets” and novel M. aichiense phage, to compare spread and killing efficiency of satellite phages to phages. These results are described in detail below.

Interpretation:

Two samples of a novel M. aichiense phage were sequenced, which resulted in assemblies of about 60,000bp each. The two largest contigs (23,000bp each) from the assemblies showed 72% query coverage and 88% identities to HerbertWM, and 99.99% identity to each other. The similarity to HerbertWM supports our theory that the novel mycobacterium phage is a close relative HerbertWM. We hypothesize further that this similarity, yet few differences, plays a key role in this novel phage’s ability to excise the temperate phage HerbertWM.

The other Mycobacterium phage satellites also showed significant similarity to HerbertWM, as well as other Mycobacterium phage such as Caroline.

Applications

Using Satellites for Gene Delivery 

Experiment: The P4 cosmid system was originally designed by Dr. Fa-arun at the University of Edinburgh. The original purpose of this system was to induce Cas9 mediated cell death for disease-causing strains of bacteria such as E. coli O157:H7 and Shigella flexneri. Aspects of the P4 packaging mechanism were used to create transducing units carrying a plasmid with a Cas9 cassette. In order to create a versatile tool that could be used for both E. coli and Shigella, Dr. Fa-arun also created chimeric tail fibers to expand the host range of these transducing units. The Cas9 used in the original study targeted shiga toxins in the bacterial genomes, causing a double stranded break in the genome, thus killing the bacteria.

Raw Data: 

Summary Summary

Data Analysis: 

After the first attempt to create transducing units we observed their degradation over a week-long period. The titer decreased approximately 10-fold each day of quantification. The second attempt at creating transducing units did not degrade over a two week period, and the titer was consistently around 2.6*109.

Interpretation: 

During the initial trial, the lysate degraded within a week period, decreasing the titer 10-fold per day. However, after another replicate, the titer remained stable after a two week period at about 3.4*109, which is higher than that of a standard phage. That supports the argument that satellite phage have a higher titer than other phage and thus a higher efficiency.

Experiment:  Satellites have the unique quality to be nonreplicative in the absence of their helper phage, providing enhanced biosafety compared to standard phages. However, if there is degradation of these satellites, their inability to replicate can prevent them from having the desired effect on the system. Therefore, developing a way to maintain the high titer of the P4 transducing units can make them more efficient long-term. Since P4 requires the P2 packaging mechanism to lyse, infecting a P2 lysogen transformed with the necessary tail fibers with the P4 transducing unit and triggering lysis should produce a higher titer of transducing units. The result was a 100-fold increase in transducing units with an average titer of 1.9*1011.

Raw Data: 

Summary

Data Analysis: 

After one round of replication, we observed a large increase in titer. The titer of the original transducing units was about 109, but after infection with only 100uL of 10x diluted transducing units, the resulting titer was about 1011. Therefore, there was a 100 fold increase in the titer of transducing units.

Interpretation: 

If after one round of replication the titer increased 100-fold, then the P4 transducing units and the replicative bacteria could be utilized as a system in vivo to increase the effectiveness of the transducing units due to constant production of new transducing units.

Experiment: After producing transducing agents containing the Kanamycin Resistance (KanR) targeting P4 cosmid, as well as transducing units containing the original P4 cosmid lacking a targeting crRNA as a negative control, we measured the titer of each on an E. Coli indicator strain lacking kanamycin resistance. We used these data to equalize the titer of each transducing agent filtrate we produced for these experiments to 109 transducing agents per milliliter.

We measured the killing effect of each type of transducing agent filtrate by incubating 100uL of dilute transducing agent filtrate with 100uL HL 713 bacterial culture diluted in SM buffer to a concentration of 107 cells per milliliter. Thus, all experiments were conducted at a multiplicity of infection of 100 transducing agents per target cell. HL 713 was grown up to the end of log phase as indicated by an OD600 of between 0.5-0.6 before being diluted to the desired concentration.

To determine absolute killing effect post-treatment cells were serially diluted in LB and plated on nonselective LB agar, then incubated at 37 C for 16 hours. Colonies were then counted at each dilution level to estimate post treatment bacterial titer.

Raw Data: 

Summary Summary

Table. HL713 titer was 107 per mL, transducing agent titer was 109 per mL for all measurements

Data Analysis: 

A two sample t-test was used with the alternative hypothesis that the true difference in means between KanR transducing agent treated and control groups is less than zero.

Interpretation: 

Our experimental data was sufficient to reject the null hypothesis at a significance level of 0.04 with an effect size of 6.26 fold difference between post treatment colony forming units of each group.

Experiment: By repeating the same procedure used to quantify KanR cosmid killing effect on HL 713 and plating on chloramphenicol containing LB we can screen for cosmid transductants only. Since the same titer of transducing agents and target bacteria was used between measurements, a similar number of transductants should be generated in each round of measurement, thus any differences in the number of chloramphenicol resistant colonies between KanR targeting and non-targeting treatments arises from the killing effect of Cas9 mediated cleavage induced by the KanR targeting cosmid.

Summary

Data Analysis: 

From three trials, we concluded that there was a killing efficiency between 0.50 and 0.99. Meaning in cases of successful transduction the KanR targeting cosmid kills its host between 50% to 99% percent of the time. Cases of HL 713 colonies surviving transduction with the KanR targeting P4 cosmid could be due to a variety of factors, including the possibility of HL 713 escaping the effects of sequence specific killing through a mutation to the protospacer adjacent motif in its kanamycin resistance marker, which would effectively prevent Cas9 from cleaving HL 713’s chromosome.

Interpretation: 

We chose the target site specified by the crRNA spacer in the KanR targeting cosmid in specifically so that silent mutations (mutations which do not change how a protein is translated) in the relevant protospacer adjacent motif were impossible, thus immunity to the sequence specific killing effect of the KanR targeting P4 cosmid and full kanamycin resistance should be mutually exclusive. Additional analysis of the same measurements indicate this design strategy may have been successful, as when screened for cosmid transductants which have functional kanamycin resistance genes (by screening on LB containing both kanamycin and chloramphenicol) the killing efficiency recovers to high levels (96-99%).

Experiment: To characterize the silencing ability of the dCas9 cosmid we inserted the neomycin phosphotransferase targeting crRNA into the dCas9 cosmid via a golden gate reaction with BsaI, transformed into E. coli DH5-alpha, harvested the construct and confirmed its sequence with nanopore sequencing.

We then followed the same procedure as with previous experiments to produce transducing agent filtrate containing the dCas9 P4 cosmid (with added KanR targeting crRNA), as well as original P4 cosmid filtrate as a negative control. We treated cultures of HL 713 with each, serially diluted in LB, and plated on LB agar plates with chloramphenicol, kanamycin and chloramphenicol, and no antibiotics. We incubated the plates for 16 hours at 37 C and counted colonies at each dilution level.

Raw Data: 

Summary

Data Analysis: 

Overall treatments with the dCas9 cosmid with a KanR targeting crRNA spacer insert produced a smaller percentage of transductants with resistance to both chloramphenicol and kanamycin, while producing a comparable number of chloramphenicol resistant colonies. First, this indicates that the dCas9 cosmid can be packaged into transducing units as effectively as the original P4 cosmid. Additionally, it suggests some silencing effect on HL 713’s kanamycin resistance among transductants.

Interpretation: 

Though this experiment was not sufficiently powered to draw any firm quantitative conclusions about the potency of the dCas9 cosmid’s silencing effect on this target, we wanted to characterize it somewhat, and this result is encouraging enough to motivate further characterization of this device with a different experimental design.

Experiment: We deployed the kanamycin targeting P4 Cosmid system into the model colon in order to determine the effect and persistence of the system in conditions simulating a human colon.

Raw Data: 

Data Analysis: 

Summary

Interpretation: 

The Cas9 kanamycin targeting killing effect is significant with a p-value of 0.04 in the 3-chamber colon (EF1) and has an effect with a p-value of 0.15 in the single chamber colon (termed “Mini”). This suggests that the P4 cosmid system can induce sequence specific killing at the bacterial population level, even subject to the conditions of our simulated colon.

Experiment: We engineered the P4 cosmid to include the red fluorescence protein, and afterwards quantified the fluorescence of E. coli strain HL 713 transduced with the RFP cosmid.

Raw Data: 

Summary

Table. Fluorescence measurement of E. coli RFP results from the soil microcosms. Sample 1: Negative Control - LB, Sample 2: Positive Control - E. coli HL 713 with RFP

Data Analysis: 

The E. coli HL 713 transduced with the RFP cosmid was significantly more fluorescent than the negative control of LB.

Interpretation: 

The E. coli HL 713 was successfully infected with the P4 RFP cosmid and showed a fluorescent phenotype.

Experiment: We deployed the RFP cosmid into soil microcosms 5-8 in order to determine the efficacy and persistence of the RFP cosmid in the natural environment of the soil.

Raw Data: 

Ten samples were collected from distinct colonies from soil microcosm plating on LB with kanamycin and chloramphenicol.

Sample Image       

A1: Negative control - uninfected E. coli HL 713, A2: Positive control - E. coli HL 713 infected with RFP, A3: Negative control - water. A4: Sample 1, A5: Sample 2, A6: Sample 3, A7: Sample 4, A8: Sample 5, A9: Sample 6, A10: Sample 7, A11: Sample 8, A12: Sample 9, B1-11: Empty wells, B12: Sample 10. 

Data Analysis: 

The above table is a red fluorescence protein measurement for the 10 colonies from soil samples. Sample 6 showed a significantly higher measurement than the negative control showing that this sample was fluorescing, but not to its fullest extent.

Interpretation: 

This, in combination with the results from in-vitro testing, indicate that many RFP cosmid transductants do not show a fluorescent phenotype after re-isolation from soil. A likely explanation is that the stressors of the soil environment cause E. coli HL 713 to express the RFP at lower levels than when observed in more favorable conditions.

Overcoming Prophage Immunity

Experiment: We plated novel mycobacteria phage satellites, which we call “phagelets”, that were stored in lysates for 6 years, to increase their titer.

Raw Data: 

Out of the 14 “phagelets” that we attempted to use to infect M. aichiense by phage plating, four formed plaques.

Originally, these phage satellites showed one or two plaques after the initial plating of large quantities of old lysate, but were grown out to a titer of 107 to 108. The non-satellite phage reached a titer of 1 x 109.

Data Analysis: 

Our team managed to increase the titer of four “phagelets,” two of which were not sequenced or characterized before. What we found was one was in fact not a phage satellite and was a standard phage that shared large similarities with the prophage in M. aichiense called HerbertWM (more on it in discovery). While the satellites worked to excise HerbertWM, there was no evidence of HerbertWM when the phage lysed the cell.

Interpretation: 

Although all mycobacteria satellites were not revived, the fact that four were still viable shows the ability of our developed procedures to revive denatured lysate.

Experiment: M. aichiense and the “phagelets”, novel mycobacteria satellites, were deployed into soil microcosms 13-16. M. aichiense and a novel mycobacteria phage were deployed into microcosms 9-12. This experiment was to determine the efficacy of the phage satellites and their relative effectiveness to an M. aichiense phage.

Raw Data: 

Data Analysis: 

Summary

These results indicate that neither the mycobacteria phage satellite nor the non-satellite mycobacteria phage had a significant killing effect on M. aichiense.

Interpretation: 

There was no significant effect of the “phagelets” on the population of M. aichiense in the soil microcosms. However, the phagelets continued to survive in the soil due to their confirmed presence via phage plating.

Host Expansion

Experiment: We wanted to determine whether the “phagelets” have circular or linear DNA. We amplified a large section of the phagelet genome to confirm the expected size, as well as a smaller overlapping section to cover the rest of the genome from the other direction.

Raw Data: 

Summary

Figure. Electrophoresis gel from long range PCR. Ladder is Generuler 1kb+. Lane of interest is labeled “ChrisB + Will G Long”

Summary

Figure. Electrophoresis gel from short range PCR. Ladder is Generuler 1kb+. Lane of interest is labeled “ChrisB.”

Data Analysis: 

On gel 1, we expected a band for “Chris + Will G Long” at around 11000bp, and we observed a band right around 11kb. On gel 2, we expected a band for ChrisB at 2200 bp, and observed a band at around 2200bp.

Interpretation: 

We observed amplicons of the expected size, confirming that the “phagelet” genomes to be circular.

Experiment: In order to expand the host range of HerbertWM, we confirmed that HerbertWM does not infect M. smegmatis to create chimeric tail fibers that would be compatible with M. smeg.

Raw Data: 

No plaques formed on the M. smeg plates with each mycobacteria satellite, but each formed plaques on the plates containing M. aichiense, meaning that the phage satellites were present but were not compatible with M. smeg.

Summary

Figure. Novel satellite phages, termed “phagelets”, plated with M. aichiense. Clear plaques visible in the image indicate cell lysis.

Data Analysis: 

Plaques are visible for found with both “phagelets” for M. aichiense infection.

Interpretation: 

Both HerbertWM and our novel phage were capable of M. aichiense infection, and negative for interaction with M. smeg.

Experiment: We wanted to confirm phagelet infection was associated with excision and packaging of HerbertWM.

Raw Data: 

Summary

Figure. Electrophoresis gel. Lane 1 contains a high MW ladder, and lanes 2-4 contain phagelet lysate extracted DNA (labeled ChrisB) or novel M. aichiense phage (labeled EllieK).

Data Analysis: 

We observed bands at 50kb and 11kb for the HerbertWM and “phagelet” lysate. After PCR, we observed clear bands for each of these amplicons.

Interpretation: 

The amplicons match the genome sizes of HerbertWM and the phagelets, confirming the presence of both of the constructs in the lysate.

Experiment: In order for HerbertWM to use our chimeric tail fibers, it would need to be excised and packaged. We wanted to determine if extracted and purified DNA taken from “phagelet” lysate could be transformed into competent M. aichiense and have a killing effect; this would act as a positive control to show that HerbertWM would lyse the cell regardless of the presence of the chimeric tail fiber plasmid.

Raw Data: 

No plaques formed on the plate. There was a full M. aichiense lawn on the plate but no plaques formed after three days.

Data Analysis/Interpretation: 

Electroporation of “phagelet” DNA was unsuccessful and the phage satellite is unable to lyse the bacteria by electroporation.

Experiment: We wanted to test whether a machine-learning model could accurately predict interactions between bacteria and phage using unbiased and balanced datasets.

Raw Data: 

Below is an excerpt of interactions predicted by the model.

Summary

Figure. Top 20 phage-bacterium pairs predicted as able to interact by the machine-learning model trained with our novel interaction database.

Data Analysis: 

After 20 full runs with different allocations of training and test set data each time (test set size = 220, training set size = 800), the model was found to have an accuracy of 68% with p < 0.001, using 50% as the expected prediction accuracy for a totally random model where our distribution of true and false interaction pairs in the test and training sets was even.

Interpretation: 

The model’s predictions are not random, indicating an ability to pick up on patterns in phage-bacteria interaction using genomic information.

PDE Model

For more information about our modeling, please visit our modeling page.

  • Both replicative and non-replicative of each system significant impact the indicator strain population showing an infected population on the order of 1e5 for the RFP and non-targeting CRISPR system and a similarly effective killing effect for the kanamycin targeting CRISPR system. 
  • A preliminary result indicated that the non-replicative system would have less of an effect than the replicative system which informed our experimental design to test the replicative system. 
  • The replicative system has more of an effect than the non-replicative system. 
  • The effects of each system in the model human colon did not disperse to the end of the colon which aligns with some of our experimental results. 

We developed a novel partial differential equation (PDE) model for our satellite phage systems in the human colon. This model uses advection-diffusion fluid dynamics, Monod growth kinematics, and population dynamics to describe the interactions and efficacy of our satellite phage systems. Our results demonstrate that both our replicative and non-replicative systems have a measurable effect and our replicative systems have more of an effect than the non-replicative satellite part systems.

Influence on Experimental Design

An important early result that we found from a preliminary model of the colon was that given the titer and efficiency of the transducing units, the non-replicative systems would not have a significant effect. Therefore, we tested both the non-replicative and replicative systems to confirm this result and show a more significant effect. 

Summary

Figure. Early result of the PDE model for the non-replicative kanamycin targeting CRISPR system in the model colon. 

The models of the colon, particularly the RFP systems, clearly demonstrate significant infectivity reaching an infected population on the order of 1e5 CFU in a one week time span. This PDE model also aligns with our experimental findings that the RFP system was effective, however it did not reach as significantly to later segments of the colon. This result is also seen in the soil microcosm models with infected populations reaching 1e7 CFU. 

Summary

Figure. Difference between infection of replicative and non-replicative systems. Left: E. coli RFP replicative system infected indicator strain in the colon. Right: E. coli RFP non-replicative system infected indicator strain in the colon.

The model of the colon shows significant efficacy for the satellite phage systems. The replicative model for the RFP system in the colon shows higher infectivity than the non-replicative system. 

Summary

Figure. Model of E. coli Replicative RFP System in the Colon. In order the subplots are the indicator strain (CFU), infected indicator strain (CFU), satellite (PFU), replicative system (CFU), infected replicative system (CFU), colon microbial community (CFU), and nutrient (%). 

Summary

Figure. Model of E. coli Non-Replicative RFP System in the Colon. In order the subplots are the indicator strain (CFU), infected indicator strain (CFU), satellite (PFU), colon microbial community (CFU), and nutrient (%). 

Interestingly, in contrast to the results of the previous models, the E. coli replicative kanamycin targeting CRISPR system has a higher population of indicator strain (CFU) than the non-replicative system.

Summary

Figure. Model of E. coli Replicative Kanamycin Targeting CRISPR System in the Colon. In order the subplots are the indicator strain (CFU), satellite (PFU), replicative system (CFU), infected replicative system (CFU), colon microbial community (CFU), and nutrient (%). 

Summary

Figure. Model of E. coli Non-Replicative Kanamycin Targeting CRISPR System in the Colon. In order the subplots are the indicator strain (CFU), satellite (PFU), colon microbial community (CFU), and nutrient (%). 

  • Both replicative and non-replicative of each system significant impact with infected populations ranging from 2e5 to 3.5e5
  • The replicative system has more of an effect than the non-replicative system. The replicative and non-replicative RFP models show clear results that the non-replicative system has less effect than the replicative system showing peak populations of approximately 2e5 and 2.5e5 CFU of infected indicator strain respectively. 

We developed a novel partial differential equation (PDE) model for our satellite phage systems in the soil. This model uses advection-diffusion fluid dynamics, Monod growth kinematics, and population dynamics to describe the interactions and efficacy of our satellite phage systems. Our results demonstrate that both our replicative and non-replicative systems have a measurable effect and our replicative systems have more of an effect than the non-replicative satellite part systems.

Similar results can be seen in the models of the soil microcosms as in the results for the colon models. The replicative and non-replicative RFP models show clear results that the non-replicative system has less effect than the replicative system showing peak populations of approximately 2e5 and 2.5e5 CFU of infected indicator strain respectively. 

Summary

Figure. Difference between infection of replicative and non-replicative systems. Left: Indicator population of E. coli Replicative Kanamycin Targeting CRISPR System. Right: Indicator population of E. coli Non-Replicative Kanamycin Targeting CRISPR System in Soil. 

The soil microcosm models of the E. coli replicative and non-replicative kanamycin targeting CRISPR system show similar higher killing efficiency for the replicative system than the non-replicative system. 

Summary

Figure. Model of E. coli Replicative RFP System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), infected indicator strain (CFU), attached infected indicator strain (CFU), satellite (PFU), replicative system (CFU), attached replicative system (CFU), soil microbial community (CFU), and nutrient (%). 

Summary

Figure. Model of E. coli Non-Replicative RFP System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), infected indicator strain (CFU), attached infected indicator strain (CFU), satellite (PFU), soil microbial community (CFU), and nutrient (%). 

The above two figures show the results for the replicative and non-replicative E. coli RFP system in the soil microcosms. The non-replicative system has a peak population of infected indicator strain between 2e5 and 2.5e5. The replicative system has a peak population of infected indicator strain between 2.5e5 and 3e5. 

Summary

Figure. Model of E. coli Replicative Kanamycin Targeting CRISPR System/ M. aichiense and Phagelet System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), satellite (PFU), replicative system (CFU), attached replicative system (CFU), soil microbial community (CFU), and nutrient (%). 

Summary

Figure. Model of E. coli Non-Replicative Kanamycin Targeting CRISPR System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), satellite (PFU), soil microbial community (CFU), and nutrient (%). 

The two above figures show the replicative and non-replicative kanamycin targeting CRISPR system. The replicative system has a greater killing effect than the non-replicative system with a minimum population of indicator strain around 3e5 and 3.5e5 respectively. 

Refined parameters from comparison with Microcosm Data

After preliminary results from the soil microcosms we were able to adjust the model parameters to better represent the dynamics. 

Summary

The diffusion rate of bacteria and transducing units was determined by spread from the ‘center’ to the ‘edge’ measurements using the following equation to estimate the diffusion coefficient

Sample Image

The burst size was re-estimated from the effect size. The diffusion rate of nutrients was re-adjusted accordingly to the new diffusion rate of bacteria and transducing units.

Sample Image Sample Image Sample Image

Figure. Model of E. coli Replicative RFP System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), infected indicator strain (CFU), attached infected indicator strain (CFU), satellite (PFU), replicative system (CFU), attached replicative system (CFU), soil microbial community (CFU), and nutrient (%).

Sample Image Sample Image

Figure. Model of E. coli Non-Replicative RFP System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), infected indicator strain (CFU), attached infected indicator strain (CFU), satellite (PFU), soil microbial community (CFU), and nutrient (%).

Sample Image Sample Image

Figure. Model of E. coli Replicative Kanamycin Targeting CRISPR System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), satellite (PFU), replicative system (CFU), attached replicative system (CFU), soil microbial community (CFU), and nutrient (%).

Sample Image Sample Image

Figure. Model of E. coli Non-Replicative Kanamycin Targeting CRISPR System in Soil. Model of E. coli Non-Replicative Kanamycin Targeting CRISPR System in Soil. In order the subplots are the indicator strain (CFU), attached indicator strain (CFU), satellite (PFU), soil microbial community (CFU), and nutrient (%).

These new parameters from the soil microcosm experiments show a further spread which matches our experimental results.

References