Wet lab results
The cloning process of our parts faced us with important challenges that needed to be solved at the last minute, since most of our parts were delivered in bad conditions, being held in the Mexican customs at room temperature and opened. We noticed that we could amplify the parts when the DNA template was from the resuspended stock, but we could not use the PCR product as a new template since no amplification was obtained from this, even when changing the primers, polymerase and reaction conditions (annealing temperature). This last problem was mainly attributed to the bad shipping conditions that caused the 96 well plate containing our parts to arrive open and to be held at room temperature for almost two months. As a consequence our parts presented considerable degradation that we could notice since the first amplification was made.
Figure 1. Initial PCR of the Twist parts recently opened
However, by using increasing amounts of digested DNA for the ligation process we were able to obtain a successful ligation. After overnight growth we performed a colony PCR that showed a band in the expected size, as well as showing the presence of RFP reporter gene after propagation of those bacteria in LB medium provided with the selection marker. Unfortunately we were not able to characterize the effectiveness of the sRNAs since time was a limiting factor.
Figure 2. Agarose gel 1% used to analyse the colony PCR products. From left to right: Control, Colony 1 PSB1A3+FinalCamRNA, Colony 2 PSB1A3+FinalCamRNA, Colony 1 PSB1A3+ManualCamRNA , Colony 2 PSB1A3+ManualCamRNA
One the other hand, we had the chance to characterize the Axtl primers, these primers were recovered from the cytochrome c oxidase subunit I (COX1) gene of Ambystoma mexicanum (axolotl).
| Primer |
Sequence(3' to 5') |
Lenght (bp) |
| Axtl Fw
| AGCTGGAACGGGATGAACTG |
20 |
| Axtl Rev
| CGTCGTGGTATTCCGGCTAA |
20 |
We amplified a sample with five different annealing temperatures between a range of 55°C to 70°C to identify the optimum for the Axtl primers. From this experiment, 61°C showed to be the optimum temperature. However in our first attempt we observed similar amplification levels in all temperatures, we assume this result was a product of the high sensitivity of our polymerase Primer Star Max; to resolve this we performed a second experiment using Thermo Fisher Taq Polymerase and obtained similar results, confirming this was a product of a good primer design that showed good annealing stability.
Figure 1. Agarose gel 1% used to analyse optimal annealing temperature of PCR products with Axtl primers. From left to right: 1kb Ladder, 55 °c, 58 °C, 61 °C, 64 °C, 67 °C
We plan on performing further work to expand on our solution. Despite the limitations encountered in the lab, we were able to generate other results that go beyond the lab and use mathematical and systematic approaches to expand on our project.
Software results and conclusions
As part of the mission of the project, our team designed a dataset of small RNAs using our python pipeline for targeting cloramphenicol-acetyl transferase and penicillin beta-lactamase genes in bacteria. The profile of the score vs the binding site of the created sRNAs is on figure 2.
Figure 2. Results of the sRNAs analized by sRNA Designer
As noted on figure 2, higher scores for sRNA hybridization efficiency can be achieved if the target region is located downstream of the start codon. Since this is the first attempt to merge both scores 1 and 2, our team decided to select three sRNAs with improved characteristics per gene: one with the higher score 1 value, a second one with a higher S2 value and the last one with higher final score (S1+S2). It is important to note that we have also considered the sRNAs made by basepairing to the first 24 nts as a control, since the actual protocols state that a sRNA must pair to these site.
Despite the fact that scores are normalized, the higher reported scores are not equal to 1 (S1 and 2) or 2 (final score) since the selection criteria used in the present project is only to choose those sRNAs inside a 100-nt downstream gene start region, as explained on our design.
However, after a suscessfull neural-network based model incorporation into our software (see model and software) we were able to calculate the best sRNA for downregulate the cloramphenicol-acetyl transferase gene in E. coli regardless the basepairing site previosuly mentioned limitations. It is important to note that since our neural network model was trained with reported sRNA:mRNA pairs retrieved from literature (in which the basepairng position was present) we consider that we can incorporate any sRNA that give suscessful results according to our modelling regardless the basepairing site
Table 2. sRNAs created by our software in order to downregulate cloramphenicol-acetyl transferase gene in bacteria. This ones were created considering the incorporation of our neural network model. For more details about the titles see model and software pages
Microfluidic results
The proof of concept related to the microfluidic encapsulation of the bacteriophages relies on the principle of the microfluidic coaxial device that is proposed to produce the liposome-phage. It is based on the injection of an organic phase into an aqueous phase using a constant and controlled flow rate to induce a laminar flow. In literature, this type of device has already been used for this purpose. The results showed the production of unilamellar liposomes containing each one or two phages. It has also been observed that the proportion of charged liposomes is superior when compared to the proportion of empty liposomes, proving a high rate of encapsulation [2].
The diffusion of molecular species at the interface of the organic phase with the aqueous phase induces the formation of liposomes by auto assembly due to the high polar environment created when the organic phase encounters the aqueous phase [2,3]. In COMSOL, the channels were constructed by extruding work planes containing their transversal geometry. The measures of the device were the same as the ones reported by Vladisavljević, G. T et al. [1] in order to compare the results of the simulation. The inner channel has a round shape (0.58 and 1 mm for the inner and outer diameter respectively) and the outer channel has a squared shape (1mm and 1.2 mm for the inner and outer side length). At the end of the inner channel a cone-shape structure with an inner diameter of 209 m was constructed. From the materials database of COMSOL, glass borosilicate was assigned to the domains and frontiers of the channels and water was used for both inner and outer channel’s fluids.
In order to obtain the velocity field and the flow line of fluids in the channels, the physic module of laminar flow was used. Two entries corresponding to the inner and outer channels were defined; a flow rate of 1.610-10 was used for the inner channel during all simulations and the flow rate of the outer channel was 0.6, 3 and 4.8 ml/h for each of the three simulations. For the boundaries, no slip condition was assigned to the channels of the walls since they are solid. For the computation, a tetrahedral mesh was used because we constructed a 3D model.
First, a stationary study was modeled in order to obtain the velocity field and the flow lines. In the model, the maximum velocity is reached at the center of the outlet of the inner channel as reported in the simulation made by Vladisavljević, G. T et al. [1]. The velocity at the walls is zero as shown in the velocity field of our three simulations.
The velocity of fluids in the channels depends on their cross-sectional area; they are inversely proportional. At the end of the inner channel, the velocity is higher than at the entrance and it decreases when the fluid enters the outer channel as a result of the augmentation of the cross-sectional area and the drag force from the aqueous phase (see Figure 1, 2 and 3). The results of our simulation indicate that as the flow rate of the aqueous phase increases; the velocity at the center of the outer channel also increases.
Figure 1. Velocity field computed for flow rates of 0.6 ml/h for both the organic and aqueous phases.
Figure 2. Velocity field computed for flow rates of 0.6 ml/h and 3 ml/h for the organic and aqueous phases respectively.
Figure 3. Velocity field computed for flow rates of 0.6 ml/h and 4.8 ml/h for the organic and aqueous phases respectively.
When comparing the values of the maximum velocity in the inner channel obtained by our simulation using a static flow rate for the organic phase and a variable flow rate for the aqueous phase, the values are comparable in magnitude; 910-3m/s and 810-3m/s for the value reported in literature and the one obtained with our three simulations respectively. However, when the flow rates of aqueous phase are increased, there is a difference in the maximum velocity reached at the outer channel, for example for flow rates of 0.6 ml/h and 4.8 ml/h for the organic and aqueous phase respectively, the maximum velocity at the outer channel is three times bigger in our simulation (410-3m/s and 1210-3 for the literature and the model respectively). This mismatch could be explained by the difference in values of the properties of fluids used for each simulation; neither the type of fluid nor the density used for the simulations are specified by Vladisavljević, G. T et al. [1]. In the future, the density of the organic mix and the phage solution need to be determined to achieve more accurate results with our simulation.
Finally, the particle tracing module was used in order to observe the trajectory of the bacteriophages. Solid particles simulating the phages contained in the aqueous phase are released from aleatory positions at the inlet of the outer channel in the model in COMSOL. One particle was released each 0.1 s to observe its behavior. As shown in the animation, the velocity reached by the phages is the same as the one reported in the velocity field. The flow lines respect the boundaries of the model indicating its correct functioning.
In the future, we propose to fabricate the device in order to use microscopy to m onitor in real time the behavior of the fluids in the system. If it works correctly, a jet must be observed at the end of the inner channel; when increasing the flow rate of the aqueous phase, the jet must be shorter and thinner. If the size of the jet is closer to the diameter of the inner channel, a better mixing will be achieved. A better mixing leads to produce smaller liposomes according to Vladisavljević, G. T et al. [1]. It has been stated that the mixing grade depends on the flow rates and properties of the fluids and on the geometry of the channels, specifically on the diameter of the exit of the inner channel [1,2]. We expect our model to produce liposome-phage complex correctly because the behavior of our simulations is comparable to the one reported in literature.
Nebulizer prototype
In order to make the delivery of the liposome-phage complex correctly we created a model prototype using electronics and 3D printing. The prototype made was based on the design of a membrane nebulizer "Omron NE-022" [3], designing a model that simulated its operation using an ultrasonic nebulizer, in light of a vibrating membrane that would be implemented in the real device. The base or box of the device was made by 3D printing, while SLA/DLP printing for the dispenser, as well as FDM printing for the rest of the system, On the other hand, for the electronics, we adapt the design of “Nano mist” [2] modifying it to extend the button to the base of the device and changing the Lipo battery for a Li-ion MS18650 battery.
Figure 4. Video of the nebulizer prototype.