Results from our project are separated into three distinct phases: cloning our plasmid constructs, evaluating those constructs via a growth assay, and purifying our protein expression products.

Cloning

Four plasmid constructs were designed to investigate encapsulin facilitated expression of antimicrobial peptides (AMPs): an empty pETDuet vector (vehicle control); pET expressing a single T4GALA encapsulin monomer (negative control); pETDuet expressing the AMP HBCM2 fused to a Tobacco Etch Virus (TEV) protease site, flexible linker, and T4GALA targeting peptide (positive control); and pETDuet expressing HBCM2, TEV protease site, linker, targeting peptide, and a T4GALA encapsulin monomer (experiment) (Figure 1A).

Figure 1. A) Diagram of our three protein-expressing constructs. B) Agarose gel showing PCR amplification of the T4GALA encapsulin from its original plasmid. Lane 1: ladder, lane 3: pET vector with T4GALA, lane 5: PCR-amplified T4GALA. C) Agarose gel for extraction of double restriction digest of pET plasmid for linearization. From left: ladder, digested pET.

The T4GALA encapsulin construct and empty pETDuet vector containing regulatory elements and ampicillin resistance was generously donated to our team by the Geissen Lab. Our HBCM2, TEV site, linker, and targeting peptide insert with overhangs for Gibson assembly were ordered from Integrated DNA Technologies (IDT). PCR was used to extract the T4GALA monomer from our donated plasmid (Figure 1B). A double restriction digest and gel extraction were used to linearize our pET vector (Figure 1C). Gibson assembly reactions were performed to create our pET + AMP and pET + AMP + encapsulin constructs. Successful assembly was confirmed by Sanger sequencing (Figure 2), and all four plasmids were transformed into the same strain of BL21 E. coli.

Interestingly, we noted several mismatches between our AMP insert and its Sanger sequencing (Figure 2A). Sequencing proved challenging due to nonspecific primer binding, and after trying multiple sequencing primers early in the season, we attributed these mismatches to technical issues and proceeded with experimentation. In the context of our later results, however, we acknowledge the possibility that there was a potentially problematic mismatch with the sequence of our insert that could have affected the potency of the AMP.

Figure 2. A) BLAST comparison of sequencing data (Query) to expected result (Sbjct). B) A representative chromatogram segment from our sequencing.

Growth Assay

We reasoned that if the encapsulin was mitigating the toxicity of the AMP during expression, the growth rate of bacteria expressing the AMP + encapsulin would be higher than the growth rate of bacteria only expressing the AMP alone. To test this, we designed an assay in which we inoculated wells of Luria-Bertani (LB) broth containing 100 μM ampicillin with each of our four bacterial strains. Each strain featured three biological and three technical replicates. Protein expression was induced in half of the samples with 100 μM IPTG at time=0. OD600 measurements were taken by a plate reader every 5 minutes for 12 hours. Logistic regressions were performed on the growth phase of each well, and the growth rate was recorded for each. To compare these growth rates, t-tests were performed with considerations made for the variance due to our biological and technical replicates [1].

Figure 3: Growth assay average growth rates for 4 experimental groups: empty pET vector, encapsulin only, AMP only, and encapsulin + AMP. Error bars represent 95% confidence intervals. A) Growth rates for uninduced gene expression B) Growth rates for induced gene expression via IPTG.

As seen in Figure 3B, bacteria expressing our AMP and encapsulin had a higher average growth rate than bacteria expressing the AMP without encapsulin, suggesting the mitigating effects of the encapsulin on AMP toxicity. Although this trend was not statistically significant in our small-scale experiment, it was observed in spite of the added metabolic burden of encapsulin expression. This suggests that even if the encapsulin failed to increase the growth rate of the AMP + encapsulin bacteria relative to those expressing just AMP, the encapsulin may still have mitigated toxicity and potentially improved protein yield. Further experiments are warranted to investigate this effect, such as increasing the number of replicates or the amount of protein expressed by each cell. Increased expression could be achieved by growing the bacteria in a richer media like terrific broth (TB) media or waiting to induce with IPTG at a higher OD600 following initial bacterial growth.

In addition to the previously noted sequencing data, a possible explanation for the smaller-than-expected effect of AMP toxicity on bacterial growth is the fusion of the targeting peptide and linker to the AMP-only positive control. These domains (targeting peptide + linker) were added so they better matched the experimental construct that also had these added to facilitate encapsulation. Additionally, previous experiments suggested adding these domains did not affect the AMP’s toxicity [2].

Protein Purification

In parallel with the growth assay experiments, we explored the use of our encapsulin system to express and purify HBCM2. BL21 E. coli were transformed with either a pETDuet + encapsulin plasmid or a pETDuet + encapsulin + AMP plasmid. These transformants were then grown at 37 degrees Celsius and shook at 250 rpm in LB broth with 100 uM ampicillin. At an OD600 of 0.6, the bacteria were induced with 200 μM IPTG to promote protein expression. The proteins were expressed for 4 hours at 37 degrees C while shaking at 250 rpm. Harvested E. coli were then lysed and centrifuged to produce a clear lysate. The thermostability of our encapsulin allowed us to then heat the clear lysate to 85 degrees C, inducing precipitation of native E. coli proteins. The efficacy of these techniques was confirmed with a protein gel (Figure 4).

Figure 4: SDS-PAGE gel showing heat protein purification. Note lanes 4, 6, and 7 showing the increasing purity of samples heated to 25, 65, and 85 degrees, respectively. From left: ladder, clear lysate using dilute lysozyme, clear lysate using moderate lysozyme, clear lysate using high lysozyme, clear lysate with higher concentration of cells per unit lysis buffer, clear lysate heated to 65º C, clear lysate heated to 85º C.

While this process produced a relatively pure protein sample, we investigated the use of ammonium sulfate precipitation for further purification. Samples of heat-purified protein were incubated at 4 degrees C for 1 hour with ammonium sulfate at concentrations ranging from 10% to 100% saturation. After resuspension of collected pellets of precipitated protein and subsequent dialysis, a protein gel showed the ability of this process to increase the purity of our protein samples (Figure 5). We found that low concentrations of ammonium sulfate were able to selectively precipitate our encapsulin-AMP complex, as seen by the distinct T4GALA Monomer and AMP fusion bands at a 10% ammonium sulfate concentration.

Figure 5. SDS-PAGE gel demonstrating ammonium sulfate purification. From left: lanes show protein pellets collected after incubating with ammonium sulfate at concentrations ranging from 10% to 100% in 10% intervals. Note the high purity of the lower concentration bands.

Next, our team began working to open the purified encapsulins, separate them from our AMP (HBCM2), and use our TEV site to cleave the AMP from the targeting peptide. Efforts in this area are ongoing, but we are optimistic that this can be accomplished with accessible techniques and high efficiency by first introducing the encapsulins to a mildly acidic buffer and then exploiting the drastic differences in size and charge between the encapsulin and AMP for definitive separation.

Finally, we designed plasmids for experiments that explore the addition of multiple AMPs to each targeting peptide for a maximized yield of AMP from each encapsulin. Each encapsulin produced represents a fixed energetic cost and a fixed number of encapsulated targeting peptides, regardless of the number of AMPs attached to each targeting peptide. We hypothesize that, to a threshold, there will not be a significant increase in toxicity from adding additional AMPs to each targeting peptide. Our computational modeling team estimated the maximum number of AMPs that could be chained together per targeting peptide, and this will inform the length of the chains we plan to test. More information on the completed modeling that continues to guide our design can be seen on the model page.

References

1. Blainey, P., Krzywinski, M. & Altman, N. Replication. Nat Methods 11, 879–880 (2014). https://doi.org/10.1038/nmeth.3091
2. Lee, T. H., Carpenter, T. S., D'haeseleer, P., Savage, D. F., & Yung, M. C. (2020). Encapsulin carrier proteins for enhanced expression of antimicrobial peptides. Biotechnology and bioengineering, 117(3), 603–613. https://doi.org/10.1002/bit.27222