. (). (view online)

Engineering Success | TU-Eindhoven - iGEM 2024

Engineering

Trust me, I'm an engineer

~ TheVilkaz

Creating something new is rarely a linear process. Regardless of planning or research, many unexpected challenges arise. In our experiments we planned for the unexpected by using the iGEM engineering cycles to build from a minimal viable product to a final product. We also used this approach in the modeling where we iteratively improved our model.

Lab Engineering Cycle

BMV Isolation engineering cycle

For the isolation of bacterial membrane vesicles (BMVs) from both E. Coli and M. Smegmatis, multiple design cycles have been performed.

Cycle 1
Click to see the next step
Design
Literature research resulted in a first draft version of a potential protocol. This protocol can be found here. The protocol was altered where necessary to fit our project.
Build
The protocol was performed in the lab.
Test
With the help of DLS the sample that resulted from the protocol was analysed. The same sample was measured on day one and day five, both for the charachteristics and the stability of the sample.
Learn
The number and intensity mean of the sample were lower than the expected value for BMVs.
The PdI and intensity mean increased significantly over five days, confirming that the sample is not stable. The parts within the sample most likely form aggregates. To confirm this conclusion, CryoTEM measurement was performed.
CryoTEM results showed BMVs of ranging sizes in the sample. Next to that, the sample contained some protein aggregates. It was therefore concluded to improve the protocol to decrease heterogeneity.
Design
The insights of the results of the previous protocol in combination with new literature research resulted in an improved protocol. Within this new protocol, 0.45 and 0.2 µm filters were used to filter out whole bacteria and bacterial debris. 30 kDa amicon filters were used to further concentrate the sample. To overcome the issue of small membrane vesicle in the sample, a new growth medium was introduced: Sauton's minimal medium. This Sauton's medium should provide an environment that favors BMVs production at the expense of bacterial growth.
Build
The improved protocol was performed in the lab.
Test
The sample was again analysed with DLS. This time, the same sample was measured multiple days from day one till day ten.
Learn
Both the number mean and intensity mean were within the expected size range for BMVs. The PdI was also lower than in Cycle 1, suggesting that the sample was more homogenous. The sample was stable for seven days, which is expected for BMVs. As the number mean and intensity differed, it was decided to do a CryoTEM measurement to visualize the sample.
Cycle 2
Cycle 3
Design
Now that the DLS measurements suggested that the protocol was successfull in isolating BMVs, a fresh sample was prepared via performing the same protocol as in Cycle 2. The sample was analysed with a CryoTEM measurement.
Build
The protocol from Cycle 2 was performed.
Test
The sample was analysed with CryoTEM. The diameter of the BMVs was measured by hand to get an estimation of the average size.
Learn
The CryoTEM showed that the sample contained BMVs. The total amount of BMVs was lower than in the protocol of Cycle 1. The size of the BMVs, measured by hand, corresponded most to that of the number mean. Besides BMVs, the sample still contained protein aggregates. Next to that, ruptured BMVs were visible, most likely caused by the centrifugal force. Further improvement of the protocol is advised.

Cloning engineering cycle

For the cloning of the genes encoding for the fusion proteins that were to be expressed in M. smegmatis, several engineering cycles were performed.

Cycle 1
Click to see the next step
Design
In order to clone inserts 3a-3g into the plasmid pCHERRY3, by Gibson assembly, primers were designed to perform a PCR linearization of the pCHERRY3 plasmid. For all inserts gBlocks were designed with Gibson overlaps at the ends. Since all fusion proteins consist of two main domains, namely the membrane-anchor into two Gblocks that share the same Gibson overlap in the middle. The PCR program for the linearization of pCHERRY3 was designed as shown in .
Build
The PCR linearization of pCHERRY3 was performed.
Test
The PCR product was run on an agarose gel.
Learn
No clear fragment was visible. We hypothesized that the annealing temperature was too high for the primers to bind.
PCR program for the linearization of pCHERRY3.
Stage Temperature (°C) Time
Initial denaturation 98 60 s
Denaturation 98 10 s
Annealing 67 30 s
Extension 72 165 s 32X
Final Extension 72 5 min
End 4 Hold
Design
The linearization PCR was redesigned with an annealing temperature of 65°C instead of 67°C.
Build
The adjusted linearization PCR was performed.
Test
The PCR product was run on an agarose gel.
Learn
No clear fragment is visible. We hypothesized that the annealing temperature is still too high for the primers to bind.
Cycle 2
Cycle 3
Design
The linearization PCR was redesigned with varying annealing temperatures: 55°C, 60°C, 65°C, 66°C and 67°C.
Build
The adjusted linearization PCR was performed.
Test
The PCR product was run on an agarose gel.
Learn
At temperatures higher than 60°C, the agarose gel shows no clear band. However, for 55 and 60°C, the agarose gel shows a fragment at ~1.2 kb fragments instead of the expected ~5.6 kb fragment. Therefore, we formulated and tested the three hypotheses seen in .
  • There is an alternative primer binding site for which annealing occurs at annealing temperatures of 60°C and below. We tested this hypothesis by analyzing the sequence of the plasmid for alternative binding sites that would result in a fragment of approximately 1.2 kb. We did not find such a primer binding site, only one that would result in a ~1.6 kb fragment. Because the difference between these fragments is quite small, we do not exclude the possibility that this hypothesis is true.
  • The extension time of 165 seconds was insufficient for completing the replication of the full 5.6 KB fragment. In order to test this hypothesis we performed the linearization PCR with an extension time of 7 minutes. However, the same ~1.2 kb fragments were seen, proving this hypothesis incorrect.
  • The sample of pCHERRY3 plasmid does not actually contain the plasmid. We tested this hypothesis by choosing a set of restriction enzymes that should cut the plasmids into a characteristic pattern of bands on the agarose gel. The result was merely a smear of DNA fragments that were smaller than expected. Therefore, we concluded that the linearization PCR of the plasmid was not working because the plasmid was incorrect. We remade the sample of the plasmid and confirmed its sequence by whole-plasmid-sequencing.
Hypotheses resulting from Cycle 3
Learn
No bands were seen on the gel. Therefore, we formulated the following two hypotheses:
Linearizing the plasmid using a restriction enzyme prior to the PCR, will enhance the efficacy of the replication.
The primers are malfunctioning

We decided to test the first hypothesis. We choose the PstI restriction enzyme, which cuts within the sequence of the insert that is not being copied by this linearization, see . However, This strategy once again resulted in no visible bands on the agarose gel, making it seem more likely that the primers are the cause of the problem.

Test
The PCR product was run on an agarose gel.
Build
The adjusted linearization PCR was performed.
Design
With the new sample of pCHERRY3 plasmid with the correct sequence, we redesigned the linearization PCR, with varying annealing temperatures of 55°C, 60°C, 65°C, 66°C and 67°C and with the extension time of 7 minutes. To increase chance of success, we performed the reactions with the Phusion Master Mix as well as Q5-polymerase.
Cycle 4
pcher3 vector
Figure 1. llustration of the restriction of the pCHERRY3 vector prior to the linearization PCR.
Cycle 5
Design
We designed and ordered new primers to reattempt the linearization PCR with varying annealing temperatures of 55°C, 60°C, 65°C, 66°C and 67°C and with the extension time of 7 minutes. We did not only reperform the linearization PCR with the new primers but also with the combinations of a reverse primer from the old set and a forward primer from the new set, and vice versa. Additionally, since the plasmid is high in GC-content, we decided to also perform a reaction with a GC-enhancer. We performed the reactions with the Phusion Master Mix as well as Q5-polymerase.
Build
The adjusted linearization PCR was performed.
Test
The PCR product was run on an agarose gel.
Learn
No bands were seen on the gel. We decided to change the cloning strategy from Gibson assembly to restriction ligation cloning.
Design
With the new sample of pCHERRY3 plasmid with the correct sequence, we redesigned the linearization PCR, with varying annealing temperatures of 55°C, 60°C, 65°C, 66°C and 67°C and with the extension time of 7 minutes. To increase chance of success, we performed the reactions with the Phusion Master Mix as well as Q5-polymerase.
Build
Restriction ligation cloning was performed for fusion protein (3c) PorinN50-GFP.
Test
The resulting plasmids were visualized on an agarose gel. Next, the plasmids were transformed into TOP10 E. coli cells, plated on selective media, cultured, miniprepped, and sent to Azenta for sequence confirmation.
Learn
On the agarose gel of the plasmids with insert (3c) PorinN50-GFP, bands are seen at the expected height (~6.5 kb), but other bands are also seen at higher molecular weight. The gel can be viewed here.

The Azenta results confirmed the sequence of the insert. This still has to come true. In the next steps, we plan to clone the plasmids with the inserts 3a, 3b and 3d-3g into pCHERRY3 by restriction ligation cloning as well. For these plasmids, we will also amplify the gBlocks prior to restriction ligation, to increase the yield.

Cycle 6

Modeling Engineering Cycle

In parallel to the lab, the model also went through multiple cycles. We worked in weekly sprints, where we would also look at the long term planning and make adjustments to see which milestones were still feasible. The modeling results inform the future steps of the lab in future engineering cycles to be performed after the iGEM competition. More details about the modeling can be found on the modeling page.

Cycle 1
Click to see the next step
Design
We want to start with a preprocessed data file available publicly. In this way we have results as fast as possible. We took some inspiration from the Tongji China 2018 team for the intial idea but quickly went in another direction after literature research.
Build
The main tool that we use, pVACtools, requires the data file to have some annotations, so we implemented a step to add these. Another input is required, the HLA type, which is available at the data source.
Test
The main tool that we use, pVACtools, requires the data file to have some annotations, so we implemented a step to add these. Another input is required, the HLA type, which is available at the data source.
Learn
We learn that other data sources may not provide an already processed file and the HLA type.
Design
The goal is to not use the already processed data.
Build
We create an end-to-end pipeline that processes raw DNA data and determines the HLA type. This involves learning about a lot of bioinformatics tools, renting cloud computing, following tutorials and bugfixing.
Test
We confirm that we get approximately the same results and that the HLA type was correctly determined.
Learn
Neither the processed file or our pipeline is taking advantage of all information, mainly RNA is not integrated
Cycle 2
Cycle 3
Design
The goal is to also integrate RNA information in the pipeline.
Build
We need to learn new tools but the concepts are often similar which makes things faster. We upgrade our cloud computing because it is taking a long time to run.
Test
We find that RNA information is now present in the output and that the results changed.
Learn
We were already aware that our data had a limitation. It contained only tumor DNA/RNA data and not a healthy control.
Design
After a long time we got access to a restricted dataset with a healthy control. We want to analyse this data to see the effect
Build
We extend our pipeline to consider the healthy control. We also run the original pipeline without the control.
Test
We check if the results of the tumor-only pipeline differs from the pipeline with the control.
Learn
We find no differences for this particular patients but cannot draw a general conclusion from this. We identify a way to reduce the need for a healthy control, proteome similarity search, but will not have time for another engineering cycle.
Cycle 4