Team:TUDelft/Design

Team

Team Members Collaborations Team Bonding

Project

Project Description Contribution Design Engineering Success Partnership Sustainability

Lab

Experimental Details Notebook Results

Models

Overview Phage infection and toxin production 2D biofilm simulation

Parts

Overview Basic parts Composite parts Part collection

Human Practices

Human Practices Integrated Human Practices Education & Engagement Inclusion Proposed Implementation Safety

Entrepreneurship

Overview Customer Product Company Business Plan

Acknowledgements

Attributions Crowdfunding Sponsors

Achievements

Judging Form Achievements Awards and Nominations

Design

• PHOCUS
• Our target: the desert locusts
• Engineering bacteriophages instead of bacteria
• The target of the bacteriophage
• Mode of action: killing the gregarious locusts
• Mathematical models of phage infection and toxin production
• Future perspectives

PHOCUS

Safe and fast, two simple words, but vital for an effective biopesticide. The Food and Agriculture Organisation (FAO) and the national government of Kenya, a country infested by locusts, highlighted the pressing need for a pesticide with these attributes to solve the actual locust crisis. Chemical pesticides are effective but not specific to desert locust [1], and a safer alternative used, the fungus Metarhizium anisopliae, is too slow [2]. Throughout the process of designing our biopesticide, we conducted interviews with experts from different disciplines. They helped us understand the complexities associated with designing a safe and fast solution to the locust crisis. This led us to design PHOCUS: a bacteriophage based biopesticide that is fast and safe.

Figure 1. PHOCUS: a bacteriophage-based pesticide that is fast and safe.

PHOCUS is a biopesticide based on engineered bacteriophages that infect the gut bacteria of the locust. Upon infection, the bacteria produce molecules that are specifically toxic to locusts: the crystal protein, Cry7Ca1, and an RNA interference (RNAi) precursor. The Cry7Ca1 toxin, from Bacillus thuringiensis strain BTH-13, has been described to have specific toxicity against locust [3, 4]. Cry7Ca1 punctures the gut lining facilitating the RNAi precursor to reach the hemolymph, where it silences the expression of locust essential genes.

To gain more insight into the feasibility of PHOCUS as a safe and fast biopesticide, we developed two mathematical models. With them we aimed to (1) predict the speed at which the toxins would be produced, and (2) acquire insights on how the spatial distribution of the locusts gut affect the host-phage population dynamics. From the former, we learned that phage infection and toxin production is a fast process. We also learned which strategies were the best to improve the production of the toxin. For the latter model, we could determine that phage mobility is a crucial property that influences toxin production.

Our target: the desert locusts

We target the desert locust (Schistocerca gregaria) as this is the locust species that causes the most economic and environmental damage. This desert locust threatens the livelihood of one-tenth of the global population [5] and is considered to be the most important species of locust to tackle due to its ability to migrate over large distances and rapidly multiply [6].

Throughout the past months, desert locusts have been found in Northern Africa, the Arabian peninsula and the South of Asia (Figure 2). During a plague, locust swarms can spread out to over 60 countries and cover up to one-fifth of the Earth’s land [5]. After selecting the desert locust as our target, we worked towards designing a biopesticide that is specific and therefore safe for other insects and animals. Our biopesticide also needs to work fast enough to eliminate swarms.

Figure 2. Areas affected by desert locust plagues [5].

Design considerations

PHOCUS Mode of action: Why killing the locust?

Locusts exhibit phase polyphenism, which is key to the swarming behavior observed during plagues (Figure 3). While brainstorming we came up with two possible approaches to exterminate locust swarms: (1) reversing swarming behavior or (2) directly killing the locust and eliminating the swarms. Before committing to either approach, we consulted experts to verify their feasibility. Multiple experts explicitly advised against de-gregarising. According to Harold van der Valk, environmental toxicologist at Falconsult, and Em. Prof. Arnold van Huis, tropical entomologist from Wageningen University & Research (WUR), de-gregarising back to solitary individuals does not actually solve the problem, but could potentially even worsen it. Indeed, there would be the same quantity of locusts, only more dispersed, making control operations less effective or even impossible. After critically reflecting the feedback from experts, we decided to focus on killing the locusts.

Figure 3. Our target is desert locusts in the flying adult phase. We will target the gregarious desert locust. Adapted from [5]

Can we kill too many locusts?

A concern might be that we exterminate all locusts and lead the species to extinction. Something that could have devastating effects on the local ecosystems. To investigate the probability of this we spoke to Cyril Piou from the French Agricultural Research Centre for International Development (CIRAD). He is an expert on behavioral locust swarm models. He stressed that solitary locusts always remain throughout. Thus, extermination of locusts would not be possible.

Engineering bacteriophages instead of bacteria

In order to deliver the toxins to the desert locust, lytic bacteriophages will be used as a vector. Bacteriophages are a type of virus that infect and replicate within bacteria [7]. Lytic bacteriophages infect and reproduce fast as the cellular machinery of the host is taken over after infection and all energy is put towards expressing the viral genetic information.

Engineering bacteriophages instead of bacteria

We decided to engineer bacteriophages and not the bacteria. There are several reasons why using engineered bacteriophages is a superior delivery strategy than engineered bacteria:

• After phage infection, bacteriophages hijack the bacterial transcription/translation system to redirect all resources to the production of new phage particles. If we would use engineered bacteria, upon delivery not all of the cell's energy and resources would be dedicated to producing our toxic molecules.
• By using phages we take advantage of the fact that the bacteria are already present in the gut. In contrast to bacteria, which first needs to settle inside the gut and out-compete the bacteria that are already present, our factories are already available.
• In the locust gut, the phages are expected to produce higher levels of toxin and faster than bacteria. We do not need to deliver many phages to reach high toxin levels as they propagate and spread upon infection. If we would use bacteria they would need to take over the gut microbiome or we would have to deliver very high numbers of bacteria inside the gut to produce enough toxin.

Design considerations

Using lytic over lysogenic phages

Bacteriophages can infect bacteria following two different life cycles: the lytic and the lysogenic cycles [8]. In the lytic cycle, bacteriophages inject their DNA in the host bacteria upon infection. This DNA is then transcribed and translated by the host's bacterial metabolic machinery after which the host produces new phages. Eventually the host generates endolysin, causing the cells to burst and release the replicated bacteriophages (Figure 4) [8]. In contrast, upon infection a lysogenic phage integrates its genome into the host DNA. The phages genetic material replicates together with that of the host [9].

Figure 4. The replication cycle of a lytic phage, adapted figure [10].

We will use lytic phages for different reasons. First of all, we choose to use lytic phages as the produced molecules are released from the bacteria during lysis. This is important because we need to release the toxin from the host bacteria in order to be able to work. Secondly, we need to produce high amounts of toxic molecules as fast as possible. Because lytic phages are virulent, we increase the number of infected bacteria that will produce these molecules. Thirdly, lytic phages represent a safer option than using lysogenic ones, as horizontal gene transfer by transduction is not present or less likely [11]. (For more information about the safety considerations read Lytic phage Safety).

The target of the bacteriophage

The bacteriophages will target the gut bacteria of the locust (Figure 5). The bacterial community composition of the desert locust gut has been studied thoroughly to investigate the interactions between the locust and locust gut bacteria [12, 13, 14, 15]. Most of these studies looked at the role and abundance of Pantoea agglomerans, as there were reasons to believe that this bacterium was involved in the production of phenolic components of a locust cohesion pheromone. However, a recent study shows that neither P. agglomerans, nor any other specific bacterial species, is consistently present in the gut of the desert locust [16]. Nevertheless, this study shows that gregarious, field-collected, desert locusts maintain a core population of the bacterial Enterobacter genus [Dr. O. Lavy, personal interview]. To account for any variety in the core Enterobacter population in the locust gut, our biopesticide consists of a cocktail of phages targeting a variety of bacteria from the Enterobacter genus.

Figure 5. The bacteriophages encounter their target bacteria inside the locust gut.

Proof of concept: Model bacterium and phage

For our proof of concept we chose Escherichia coli (E. coli) BL21 (DE3) as the bacterial host model, since it belongs to the Enterobacteriaceae family, the genus we will target with our bacteriophage cocktail. In order to choose a model phage, Dr Franklin Nobrega, PI of the Microbial interactions lab, University of Southampton, advised us about different E.coli phages. We chose the T7 phage over our second candidate, the lambda phage, for two main reasons [17]:

• T7 has a short latent time, defined as the period needed for the phage to reproduce inside an infected host cell.
• T7 has a big burst size, defined as the number of newly synthesized phages produced by an infected cell.

Phage engineering

Several techniques exist to engineer the phage genome, we decided to use the Bacteriophage Recombineering of Electroporated DNA (BRED) method (Figure 6) [18]. This approach is frequently used for engineering phage genomes and can be used to delete, insert and replace genes. We chose to engineer phages with the BRED method as it leads to higher frequencies of modified phages than other methods [19].

Figure 6. BRED. dsDNA containing homologous regions to the bacteriophage are electroporated into cells containing bacteriophages. Recombination results in recombinant bacteriophages, containing the dsDNA fragment [20].

As the space is limited in the T7 genome for inserting genes of interest [Dr. Franklin L. Nobrega, personal interview], we replaced inert genes of the T7 genome to increase our chances of a successful insertion. Another reason for replacing non-essential genes is that it increases the likelihood that the phage works after it has been engineered, as the phage still contains its essential genes.

Gene expression from the phage DNA occurs in three stages: the early stage, middle stage and late stage [21]. To maximize expression we hypothesised that by replacing early genes toxin expression would be higher. Early genes are expressed in an earlier stage, as the DNA is expressed earlier we thought this would correspond to a higher level of protein. Three non-essential T7 genes were considered: 0.6A (early gene), 1.1 (early gene) and 4.3 (middle gene) [7].

In order to determine which of these genes should be substituted with our recombinant gene, we performed proof of concept experiments using enhanced GFP del6 (229) (eGFP) as a reporter, substituting each of the target non-essential genes separately. In order to perform BRED, the eGFP nucleotide sequence was amplified and modified to contain homologous regions to the phage genome (Figure 7) (see Results). Then, to verify its correct insertion, we did PCR with primers that hybridize with GFP and that flanked the insertion site.

Figure 7. dsRNA targeting a non-essential gene. Due to the homologous overhangs (marked in red) recombination occurs.

Selecting engineered bacteriophages

In the case that our engineering efficiency is lower than the expected 10% [19], we need a method to select the engineered phages. For doing so, we will use a CRISPR Cas9 system (Figure 8). We designed a plasmid to contain the coding sequences for single-guide RNA (sgRNA). This sgRNA targets one of our three non-essential T7 genes, 0.6A, 1.1 or 4.3. This plasmid will be expressed in E.coli together with a plasmid carrying Cas9 (pCas9-CR4). When one of the non-essential genes are present, they will be recognized by the Cas9 guided by the sgRNA and Cas9 will cut DNA at this position. As all engineered phages do not contain the non-essential genes anymore they will not be cut, and they will be the only phages forming plaques.

Figure 8. Method to select for the engineered bacteriophages. A sgRNA is designed to recognize and cut the wild type genes. The non-engineered phages will be cut and only engineered phages survive.

Maximizing gene expression

We worked together with UT Austin, this partnership was drylab based. UT Austin focused their project on modelling gene expression of engineered T7 genomes in E. coli using a simulation framework called Pinetree. To gain more knowledge they ran Pinetree model simulations wherein they compared our three different GFP replacement locations. With this data we gained information on the expression of our engineered T7 genomes. We learned that replacement of the T7 gene 4.3 was best suited for optimal protein expression. In addition, we learned which sequence characteristics are relevant for optimizing a phage genome for overexpression of the locust specific toxin. For more information, see Partnership.

Mode of action: killing the gregarious locusts

Our goal is to kill the gregarious locusts present in the swarms by using a specific and unique complementary approach that combines the insecticidal activity of the crystal toxin Cry7Ca1 and the gene silencing effect of RNA interference (RNAi). The bacteriophage will contain the DNA coding for the Cry7Ca1 toxin from Bacillus thuringiensis (Bt), as it has been shown to be effective against locusts [3, 4], as well as a short-hairpin RNA (shRNA) that will specifically target unique sequences of locust essential genes.

Although both these options have been shown to work against locusts, they are not yet a good fit to be implemented in the field. For instance, the Cry7Ca1 toxin, specific to locust, would have to be provided to them. However, when exposed to the expected harsh environmental conditions, the protein could quickly lose its activity. In order to circumvent this issue, we propose to produce them together with the bacteriophages. Furthermore, the dsRNA would have to be injected into the locust, rendering the control of swarms unfeasible. To overcome this problem, we designed a method to deliver dsRNA to the necessary location through the locust gut. By combining existing toxins, and a novel delivery method, we can develop an effective biopesticide.

After further research we realised that co-expressing Cry7Ca1 and shRNA together would enhance their effectiveness. The Cry7Ca1 toxin will poke holes into the locust gut [3, 4]. These holes will favour the shRNA produced to reach the hemolymph, the location where shRNA has shown to work best. This combined approach will also decrease the change of resistance formation (see Safety).

Bacillus thuringiensis crystal protein Cry7Ca1

Insecticidal crystal proteins (Cry) from Bacillus thuringiensis (Bt) are used worldwide to control pests since they are toxic and highly specific [22]. Cry proteins are natively present as protoxins in the insect gut. The toxin needs to be activated by proteases, once activated it oligomerizes to form a pore in the locust gut membrane, which ultimately causes locust death (Figure 9) [23]. We use the Bt toxin Cry7Ca1, as it has been described to be toxic against the locust of the species Locusta migratoria manilensis, from the order Orthoptera [4]. Moreover, Cry7Ca1 has been reported to have little or no toxicity towards Lepidoptera, Coleoptera and Diptera, and is therefore a good candidate to make our bio-pesticide safe [24].

Figure 9. Cry protein pore formation mode of action. Cry proteins are naturally produced by B. thuringiensis as a relatively inert protoxin that is ingested by insect larvae. The protoxin is then solubilized in the insect gut and processed by gut proteases, leading to the formation of a pore structure in the cell membrane. This leads to cell lysis, gut damage and eventually larvae death.

Design of the Cry7Ca1 toxin BioBrick

The Cry protein, Cry7Ca1 (GenBank accession no. EF486523) from the Bt strain BTH-13 is a three-domain Cry protein with a molecular weight of 129 kDa, whereas the activated toxin has a molecular weight of 68 kDa (Figure 10). For our project, we chose the Cry7Ca1 toxin as it has been reported to be more efficient against L. migratoria than the protoxin [4].

Figure 10. Schematic overview of the Cry7Ca1 protein domains. The full length protein is referred as Cry7Ca1 protoxin, whereas the activated version is referred as Cry7Ca1. The numbers and arrows indicate the position of the correspondent amino acids.

An expression cassette was designed for the Cry7Ca1 activated toxin (BBa_K3407001), using the viral promoter and terminator from T7 (Figure 10), the model phage used in phage engineering experiments. The Cry7Ca1 gene was codon-optimized for expression in E. coli using the GenSmartTM Codon Optimization Tool [25]. The BBa_K3407001 includes a N-terminal His-tag and a thrombin cutting site to facilitate purification and characterization [3], as the His-tag can therefore be removed after protein purification.

RNAi approach

RNA interference (RNAi) is a biological response against double-stranded RNA (dsRNA) (Figure 11). It can be used to regulate expression of protein-coding genes [26]. Presence of dsRNA in the hemolymph has shown to lead to robust RNAi-mediated gene silencing in most tissues from the desert locust reaching maximal effect at very low doses (30 ng) [27, 28]. As injecting dsRNA into swarming locusts is not feasible we will circumvent this by producing the shRNA with bacteriophages inside the locust gut. We also thought of co-expressing the shRNA with other proteins to avoid degradation (see Future perspectives for RNAi).

Figure 11. Schematic of the RNAi approach. When shRNA enters the cell, it is processed by Dicer into short interference RNA (siRNA). RNA-induced silencing complex (RISC) if formed, and it hybridizes with the target messenger RNA (mRNA). This leads to cleavage of the mRNA and eventual gene silencing.

Short-hairpin RNA (shRNA) production

Instead of producing dsRNA we will produce shRNA to silence desert locusts genes. This choice was motivated by a few design considerations: First, the expected cleavage products of shRNA are more predictable than those of longer dsRNA sequences, hence providing more control to minimize possible off-target effects. Decreasing off-targets is determinant to the safety of this approach. Second, shRNAs are simple to transcribe and self-assemble, and they are specific and stable [30].

shRNAs consist of two palindromic sequences containing the mRNA target sequence, which self-assemble into dsRNA upon transcription (Figure 12). The shRNA will only target desert locust essential genes that will lead to locust death. To create our biopesticide we will have to use an RNA target finder to ensure specificity. These programs find off-targets in other organisms by performing BLAST searches against databases as the 1000 Insect Transcriptome Evolution (1KITE) dataset or off-target finder [31, 32].

We propose to produce multiple shRNA in tandem (tshRNA) (see Future perspectives for RNAi). To cleave each individual shRNA from the transcript we suggest expressing Mini-3 (BBa_K3407002). Mini-3 is a sequence specific endoribonuclease from Bacillus subtilis [33, 35], it will cut a specific sequence located at the bottom of each hairpin. Mini-3 was chosen as it shows a high sequence-dependent cleavage [33]. After cleavage, Mini-3 produces a GG 3’ overhang, which will be used by Dicer to cut the shRNA into small interfering RNAs (siRNA). These siRNA target and subsequently cleave the complementary mRNAs, causing gene silencing [35] (Figure 11).

Transport of shRNA to the hemolymph

The shRNA has to reach the hemolymph effectively. To transport the dsRNA to the hemolymph, we thought of a delivery protein (see Future perspectives) that includes an RNA binding domain (RBD) from human Fox-1 (feminizing locus on X) (BBa_K3407004). We designed the shRNAs (BBa_K3407006) to have a sequence in their loop that is recognised by RBD of Fox-1 (BBa_K3407004). In this way, we expect a high affinity and specific binding of the shRNAs to Fox-1 [36]), such that it could be transported to the hemolymph.

We designed an shRNA targeting eGFP del6 (229) from pUC57-OriLR-deGFP plasmid (BBa_K3407006). The dsRNA region of the hairpin corresponds to 27nt from the eGFP gene (nt 78 to 105), and is the one recognised and processed by Dicer to form siRNA. The shRNA has a single-stranded RNA (ssRNA) loop region designed to be recognised by Fox-1 RBD (BBa_K3407004). The dsRNA region contains a GG overhang that would represent the overhang left when Mini-3 (BBa_K3407002) cleaves a tshRNA, and that represents a good substrate to process the shRNA by Dicer into a siRNA.

Figure 12. (A) Schematic representation of the shRNA in vitro production using T7 RiboMAX kit from Promega. Primer pairs are annealed in the T4 ligase buffer by heating at 95ºC for 5 minutes and left cooling at room temperature for 30 minutes. Once template is annealed, T7 RNA polymerase is added to transcribe the DNA template at 42ºC for 2 hours. shRNAs self assemble as soon as they are produced. (B) DNA template design for in vitro shRNA production. T7 promoter followed by eGFP target sequences with inverted repeats are linked by the loop region, and finally ends with a GG overhang.

shRNA are produced in vitro using the T7 RiboMAX RNAi express kit (Promega) (Figure 12), which makes use of the viral T7 RNA polymerase. The DNA transcription template is synthesized as a set of complementary primers that anneal to form a dsDNA sequence containing a T7 promoter, the target sequence as two inverted repeats linked by the loop sequence, and finally ended by two GG corresponding to the overhang desired to test.

We have designed and produced shRNA with a GG 3’ overhang scar in vitro under the T7 promoter (BBa_K3407022) using the phage T7 polymerase. We have shown that our shRNA (BBa_K3407006) specifically binds the Fox-1 RBD domain (BBa_K3407004), the key protein of the proposed delivery protein. In addition, we have demonstrated that the in vitro produced shRNA can be cleaved by Dicer. We also observed that shRNA with protruding 3’ GG overhangs can be processed to produce siRNA, constituting the first step in RNAi mediated gene silencing. For more information, see Results.

Mathematical models of phage infection and toxin production

Sufficient amounts of toxin need to be produced inside the locust gut for our biopesticide to be effective. In the previously mentioned Cry7Ca1 study, they indicated that approximately 0.87 µg/ml of activated Cry7Ca1 toxin is required for a 50% lethality in Locusta migratoria manilensis [4]. Similarly, the injection of 30 ng of dsRNA in the hemolymph of the desert locust is required for a maximum knockdown effect [27]. Before testing our biopesticide on actual locusts, we decided to use mathematical models to gain insights on how our biopesticide would function and on the speed at which the toxins are produced. From this, we obtained knowledge about which phage properties are relevant to consider for an effective functioning of our biopesticide.

Host-phage interactions

First, we adapted a model of phage infection and toxin production based on the work of Beretta and Kuang (1998) [37]. This model describes the temporal evolution of susceptible bacteria, infected bacteria, phages and toxin concentration (Figure 13A). The aim of this model is to investigate how host-phage population dynamics influence the concentration profile of the produced toxin and which parameters are most important to consider for optimal toxin production. Furthermore, this model is extended (Figure 13B) to investigate if the rate at which bacteria turn resistant against phages could potentially be a threat to the effectiveness of our biopesticide.

A

B

Figure 13. A. Schematic representation of our phage infection and toxin production model. B. Schematic representation of our phage infection and toxin production model including phage resistance

Influence of spatial effects on host-phage interactions

Our first model describes host-phage interaction dynamics as an isolated system (i.e. the spatial organization of the bacteria is not taken into account). Most likely, the bacteria in the locust gut reside in so-called microcolonies, which are small groups of cells packed close to each other [38, 13]. The spatial organization of cells in these microcolonies can influence phage propagation and toxin production in the locust gut. We hypothesized that spatial effects, such as phage mobility and substrate diffusivity, may have an important influence on the performance of our biopesticide. To investigate these effects on phage infection and toxin production, we modelled the temporal evolution of the number of bacteria, phages and produced toxin with a more detailed 2D biofilm simulation framework. Furthermore, as our biopesticide does not target all bacteria of the locust gut, we investigated the influence of biofilm heterogeneity on the resulting toxin production.

Figure 14. Visual representation of the output created with our 2D biofilm simulation model

Future perspectives

Future perspectives for RNAi

In this project we have validated some parts of our RNAi approach, laying the ground for future investigations. We suggest further research would focus on:

• Expression of extra proteins: We suggest to co-express the shRNA together with:
  • SGPI-2, a peptidic serine protease inhibitor from the desert locust. The high levels of serine proteases in the gut of the locust represent a threat for phage and recombinant protein integrity. Expression of SGPI-2 can inhibit serine proteases inside desert locusts [40] and thus protect the shRNA from degradation.
  
  
• A delivery protein for shRNA transport from gut to hemolymph. We suggest to use a protein made from fused domains consisting of:
  • Fox-1 RBD. As mentioned before, it is expected to bind strongly with the loop region of the shRNA.
  • Protein kinase R (PKR) RNA-binding motif, it contains two dsRNA binding domains (DRBD). It has a high affinity to bind dsRNA unspecifically and can protect the dsRNA from degradation by dsRNAses [41].
  • Trypsin modulating oostatic factor of the flesh fly Neobellieria (Sarcophaga) bullata (Neb-TMOF). TMOF is resistant to protease degradation in the desert locust gut and is able to cross the locust gut barrier [42].
  
  

• Efficient shRNA production approaches. There are many in vivo approaches to efficiently produce RNA structures with strong RNAi response and a specific silencing output. As we need to package the DNA sequence for the shRNA and the Cry7Ca1 into the phage genome, we have limited space. To limit the insert size, RNA will be produced in vivo through a single expression cassette to limit the use of promoters. To our knowledge, no approach has successfully met the criteria needed for our case: target specificity and in vivo production efficiency in a single expression cassette. We suggest developing a novel tshRNA production approach.

  As predicted by RNA folding software programs [43], it is possible to produce multiple stable shRNAs under the same promoter. Here, A T7 promoter controls the expression of the tshRNA transcript with all shRNAs transcribed one after the other. An “ACCU'' sequence at the base of each hairpin is recognized by Mini-3, Mini-3 cleaves the shRNA out of the transcript. Next, Dicer will process the shRNA into siRNA and these will then bind to the mRNA.

Co-expression of shRNA with Mini-3 and YmdB

shRNA is susceptible to be degraded by the bacteria RNAses when produced in vivo. In order to circumvent this problem, the RNAse activity can be reduced by expression RNAse inhibitors. An example is YmdB (BBa_K3407003), an RNAseIII inhibitor from E. coli [44]. Therefore, it would be ideal to co-express this inhibitor together with any dsRNA molecule. As our final idea is to produce tshRNA, co-expression of Mini-3 (BBa_K3407002) would also be needed to cleave the tshRNA and obtain the individual shRNAs.

Therefore, we designed, cloned and over-expressed Mini-3 (BBa_K3407018) and YmdB (BBa_K3407019) in compatible plasmids that will allow the co-expression together with an overexpressed tshRNA. The three selected plasmid backbones from the BglBricks collection [45], present different antibiotic resistant genes as well as different inductors (Table 1).

Table 1: Plasmid backbones used for expression of Mini-3 and YmdB.

Plasmid:	pBbB7a - GFP	pBbA2k - RFP	pBbE8c - RFP
Induction	IPTG	aTc (tet-ON)	Arabinose
Resistance	Ampicillin	Kanamycin	Chloramphenicol
Expression level	High	Medium-high	Medium-high
Copy	High	High	Medium-high
Origin of replication	p15A	pUC	pBBR1 Rep

• Finding specific RNAi targets. RNAi can potentially target any gene of the locust. Besides killing, other target genes can cooperate to help control outbreaks or its consequences, such as silencing Neuropeptide F to reduce food intake [46], or the genes per or tim to reduce offspring [47].

Future perspectives for computer simulations

From modelling the phage infection and the toxin production, we have identified which kind of design strategies would be most effective to improve our biopesticide. From our 2D model of phage infection and toxin production we obtained an insight in how the spatial organization of bacteria could influence the toxin production. However, there are still some aspects of this spatial distribution that could be explored with this model. Future simulations could investigate:

• How the diffusion of our toxic molecules could influence the resulting toxin concentration in space. One of the advantages of PHOCUS is that the target bacteria are already relatively close to the gut lining of the locust, which is the place where our toxic molecules should be delivered. This localization of toxin production could lead to higher toxin concentrations at the gut lining, which then depends on the diffusivity of the toxins in- and outside the biofilm. The effect of this diffusivity could be modelled by incorporating a separate PDE for the toxin.


• How a heterogenous phage cocktail targeting different bacteria affects toxin production. In the end, we designed PHOCUS as a cocktail of phages intended to target multiple bacteria. Depending on the abundance of the different bacteria and the phage properties, such as burst size and lysis time, different toxin concentration profiles could be obtained. Therefore, this model could be used to investigate how much such phage properties could differ from each other without negatively affecting the maximum amount of toxin produced at a given time.

References

1. Zhang, L., Lecoq, M., Latchininsky, A., & Hunter, D. (2019). Locust and Grasshopper Management. Annual Review of Entomology, 64(1), 15–34. https://doi.org/10.1146/annurev-ento-011118-112500
2. Roussi, A. (2020). Why gigantic locust swarms are challenging governments and researchers. Nature 2020 579:7799. https://www.nature.com/articles/d41586-020-00725-x
3. Jing, X., Yuan, Y., Wu, Y., Wu, D., Gong, P., & Gao, M. (2019). Crystal structure of Bacillus thuringiensis Cry7Ca1 toxin active against Locusta migratoria manilensis. Protein Science, 28(3), 609–619. https://doi.org/10.1002/pro.3561
4. Wu, Y., Lei, C. F., Yi, D., Liu, P. M., & Gao, M. Y. (2011). Novel Bacillus thuringiensis δ-endotoxin active against Locusta migratoria manilensis. Applied and Environmental Microbiology, 77(10), 3227–3233. https://doi.org/10.1128/AEM.02462-10
5. Locusts, facts and photos. (n.d.). Retrieved October 24, 2020, from https://www.nationalgeographic.com/animals/invertebrates/group/locusts/
6. Frequently Asked Questions (FAQs) about locusts. (n.d.). Retrieved October 24, 2020, from http://www.fao.org/ag/locusts/en/info/info/faq/index.html
7. Bacteriophages: Biology and Applications - Google Books. (n.d.). Retrieved October 24, 2020, from < a href="https://books.google.nl/books?hl=nl&lr=&id=6sTBd2EJpmYC&oi=fnd&pg=PA29&dq=phages+infect+bacteria&ots=j0fY0fm6fs&sig=tIckCDuOjW0sZwV13HHBBTZNtI8&redir_esc=y#v=onepage&q=phages%20infect%20bacteria&f=false">https://books.google.nl/books?hl=nl&lr=&id=6sTBd2EJpmYC&oi=fnd&pg=PA29&dq=phages+infect+bacteria&ots=j0fY0fm6fs&sig=tIckCDuOjW0sZwV13HHBBTZNtI8&redir_esc=y#v=onepage&q=phages%20infect%20bacteria&f=false
8. Salmond, G. P. C., & Fineran, P. C. (2015). A century of the phage: Past, present and future. In Nature Reviews Microbiology (Vol. 13, Issue 12, pp. 777–786). Nature Publishing Group. https://doi.org/10.1038/nrmicro3564
9. Paul, J. H., & Jiang, S. C. (2001). Lysogeny and transduction. Methods in Microbiology, 30, 105–125. https://doi.org/10.1016/s0580-9517(01)30042-9
10. Figure 1: The phage replication cycle. | Nature Reviews Microbiology. (n.d.). Retrieved October 24, 2020, from https://www.nature.com/articles/nrmicro2315/figures/1
11. Yutin, N. (2013). Horizontal Gene Transfer. In Brenner’s Encyclopedia of Genetics: Second Edition (pp. 530–532). Elsevier Inc. https://doi.org/10.1016/B978-0-12-374984-0.00735-X
12. Dillon, R. J., & Charnley, A. K. (1995). Chemical barriers to gut infection in the desert locust: In Vivo production of antimicrobial phenols associated with the bacterium Pantoea agglomerans. Journal of Invertebrate Pathology, 66(1), 72–75. https://doi.org/10.1006/jipa.1995.1063
13. Dillon, R., & Charnley, K. (2002). Mutualism between the desert locust Schistocerca gregaria and its gut microbiota. In Research in Microbiology (Vol. 153, Issue 8, pp. 503–509). Elsevier Masson. https://doi.org/10.1016/S0923-2508(02)01361-X
14. Dillon, R. J., Vennard, C. T., Buckling, A., & Charnley, A. K. (2005). Diversity of locust gut bacteria protects against pathogen invasion. Ecology Letters, 8(12), 1291–1298. https://doi.org/10.1111/j.1461-0248.2005.00828.x
15. Dillon, R. J., Vennard, C. T., & Charnley, A. K. (2002). A note: Gut bacteria produce components of a locust cohesion pheromone. Journal of Applied Microbiology, 92(4), 759–763. https://doi.org/10.1046/j.1365-2672.2002.01581.x
16. Lavy, O., Gophna, U., Gefen, E., & Ayali, A. (2019). The Effect of Density-Dependent Phase on the Locust Gut Bacterial Composition. Frontiers in Microbiology, 9(JAN), 3020. https://doi.org/10.3389/fmicb.2018.03020
17. Campbell, A. M. (n.d.). Bacteriophages.
18. Marinelli, L. J., Piuri, M., Swigoňová, Z., Balachandran, A., Oldfield, L. M., van Kessel, J. C., & Hatfull, G. F. (2008). BRED: A Simple and Powerful Tool for Constructing Mutant and Recombinant Bacteriophage Genomes. PLoS ONE, 3(12), e3957. https://doi.org/10.1371/journal.pone.0003957
19. Pires, D. P., Cleto, S., Sillankorva, S., Azeredo, J., & Lu, T. K. (2016). Genetically Engineered Phages: a Review of Advances over the Last Decade. Microbiology and Molecular Biology Reviews, 80(3), 523–543. https://doi.org/10.1128/mmbr.00069-15
20. Chen, Y., Batra, H., Dong, J., Chen, C., Rao, V. B., & Tao, P. (2019). Genetic engineering of bacteriophages against infectious diseases. In Frontiers in Microbiology (Vol. 10, Issue MAY, p. 954). Frontiers Media S.A. https://doi.org/10.3389/fmicb.2019.00954
21. Guzina, J., & Djordjevic, M. (2015). Inferring bacteriophage infection strategies from genome sequence: Analysis of bacteriophage 7-11 and related phages. BMC Evolutionary Biology, 15(Suppl 1), S1. https://doi.org/10.1186/1471-2148-15-S1-S1
22. Pardo-López, L., Soberón, M., & Bravo, A. (2013). Bacillus thuringiensis insecticidal three-domain Cry toxins: Mode of action, insect resistance and consequences for crop protection. In FEMS Microbiology Reviews (Vol. 37, Issue 1, pp. 3–22). FEMS Microbiol Rev. https://doi.org/10.1111/j.1574-6976.2012.00341.x
23. Mendoza-Almanza, G., Esparza-Ibarra, E. L., Ayala-Luján, J. L., Mercado-Reyes, M., Godina-González, S., Hernández-Barrales, M., & Olmos-Soto, J. (2020). The cytocidal spectrum of Bacillus thuringiensis toxins: From insects to human cancer cells. In Toxins (Vol. 12, Issue 5, p. 301). MDPI AG. https://doi.org/10.3390/toxins12050301
24. Song, L., Gao, M., Dai, S., Wu, Y., Yi, D., & Li, R. (2008). Specific activity of a Bacillus thuringiensis strain against Locusta migratoria manilensis. Journal of Invertebrate Pathology, 98(2), 169–176. https://doi.org/10.1016/j.jip.2008.02.006
25. GenSmart Codon Optimization. (n.d.). https://www.genscript.com/tools/gensmart-codon-optimization
26. Hannon, G. J. (2002). RNA interference. In Nature (Vol. 418, Issue 6894, pp. 244–251). Nature Publishing Group. https://doi.org/10.1038/418244a
27. Wynant, N., Verlinden, H., Breugelmans, B., Simonet, G., & Broeck, J. Vanden. (2012). Tissue-dependence and sensitivity of the systemic RNA interference response in the desert locust, Schistocerca gregaria. Insect Biochemistry and Molecular Biology, 42(12), 911–917. https://doi.org/10.1016/j.ibmb.2012.09.004
28. Santos, D., Broeck, J. Vanden, & Wynant, N. (2014). Systemic RNA interference in locusts: Reverse genetics and possibilities for locust pest control. In Current Opinion in Insect Science (Vol. 6, pp. 9–14). Elsevier Inc. https://doi.org/10.1016/j.cois.2014.09.013
29. Robinson, R. (2004). RNAi Therapeutics: How Likely, How Soon? PLoS Biology, 2(1), e28. https://doi.org/10.1371/journal.pbio.0020028
30. RNAfold web server. (n.d.). Retrieved October 20, 2020, from http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi?PAGE=3&ID=MLjhUGXlaY
31. 1KITE Start. (n.d.). https://www.1kite.org/
32. Good, R. T., Varghese, T., Golz, J. F., Russell, D. A., Papanicolaou, A., Edwards, O., & Robin, C. (2016). OfftargetFinder: A web tool for species-specific RNAi design. Bioinformatics, 32(8), 1232–1234. https://doi.org/10.1093/bioinformatics/btv747
33. Głów, D., Pianka, D., Sulej, A. A., Kozłowski, Ł. P., Czarnecka, J., Chojnowski, G., Skowronek, K. J., & Bujnicki, J. M. (2015). NAR Breakthrough Article: Sequence-specific cleavage of dsRNA by Mini-III RNase. Nucleic Acids Research, 43(5), 2864–2873. https://doi.org/10.1093/nar/gkv009
34. RCSB PDB - 4OUN: Crystal Structure of Mini-ribonuclease 3 from Bacillus subtilis. (n.d.). https://www.rcsb.org/structure/4OUN
35. Tijsterman, M., & Plasterk, R. H. A. (2004). Dicers at RISC: The mechanism of RNAi. In Cell (Vol. 117, Issue 1, pp. 1–3). Cell Press. https://doi.org/10.1016/S0092-8674(04)00293-4
36. Auweter, S. D., Fasan, R., Reymond, L., Underwood, J. G., Black, D. L., Pitsch, S., & Allain, F. H. T. (2006). Molecular basis of RNA recognition by the human alternative splicing factor Fox-1. EMBO Journal, 25(1), 163–173. https://doi.org/10.1038/sj.emboj.7600918
37. Beretta, E., & Kuang, Y. (1998). Modeling and analysis of a marine bacteriophage infection. Mathematical Biosciences, 149(1), 57–76. https://doi.org/10.1016/S0025-5564(97)10015-3
38. Mead, L. J., Khachatourians, G. G., & Jones, G. A. (1988). Microbial Ecology of the Gut in Laboratory Stocks of the Migratory Grasshopper, Melanoplus sanguinipes (Fab.) (Orthoptera: Acrididae). Applied and Environmental Microbiology, 54(5), 1174–1181. https://doi.org/10.1128/aem.54.5.1174-1181.1988
39. Spit, J., Zels, S., Dillen, S., Holtof, M., Wynant, N., & Vanden Broeck, J. (2014). Effects of different dietary conditions on the expression oftrypsin-and chymotrypsin-like protease genes in the digestivesystem of the migratory locust, Locusta migratoria. Insect Biochemistry and Molecular Biology, 48(1), 100–109. https://doi.org/10.1016/j.ibmb.2014.03.002
40. Spit, J., Breugelmans, B., Van Hoef, V., Simonet, G., Zels, S., & Broeck, J. Vanden. (2012). Growth-inhibition effects of pacifastin-like peptides on a pest insect: The desert locust, Schistocerca gregaria. Peptides, 34(1), 251–257. https://doi.org/10.1016/j.peptides.2011.06.019
41. Gillet, F. X., Garcia, R. A., Macedo, L. L. P., Albuquerque, E. V. S., Silva, M. C. M., & Grossi-de-Sa, M. F. (2017). Investigating engineered ribonucleoprotein particles to improve oral RNAi delivery in crop insect pests. Frontiers in Physiology, 8(APR), 256. https://doi.org/10.3389/fphys.2017.00256
42. Zhu, W., Vandingenen, A., Huybrechts, R., Vercammen, T., Baggerman, G., De Loof, A., Poulos, C. P., Velentza, A., & Breuer, M. (2001). Proteolytic breakdown of the Neb-trypsin modulating oostatic factor (Neb-TMOF) in the hemolymph of different insects and its gut epithelial transport. Journal of Insect Physiology, 47(11), 1235–1242. https://doi.org/10.1016/S0022-1910(01)00086-5
43. RNAfold web server. (n.d.). Retrieved October 24, 2020, from http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi?PAGE=3&ID=MLjhUGXlaY
44. ymdB - O-acetyl-ADP-ribose deacetylase - Escherichia coli (strain K12) - ymdB gene & protein. (n.d.). Retrieved October 5, 2020, from https://www.uniprot.org/uniprot/P0A8D6
45. Lee, T. S., Krupa, R. A., Zhang, F., Hajimorad, M., Holtz, W. J., Prasad, N., Lee, S. K., & Keasling, J. D. (2011). BglBrick vectors and datasheets: A synthetic biology platform for gene expression. Journal of Biological Engineering, 5. https://doi.org/10.1186/1754-1611-5-12
46. Wielendaele, P. Van, Dillen, S., Zels, S., Badisco, L., & Broeck, J. Vanden. (2013). Regulation of feeding by Neuropeptide F in the desert locust, Schistocerca gregaria. Insect Biochemistry and Molecular Biology, 43(1), 102–114. https://doi.org/10.1016/j.ibmb.2012.10.002
47. Tobback, J., Boerjan, B., Vandersmissen, H. P., & Huybrechts, R. (2011). The circadian clock genes affect reproductive capacity in the desert locust Schistocerca gregaria. Insect Biochemistry and Molecular Biology, 41(5), 313–321. https://doi.org/10.1016/j.ibmb.2011.01.008

Special thanks to our sponsors!