Engineering Success

To give Antarctic phosphatase treatment to the digested backbone and perform cloning.

An Antarctic Phosphatase (AP) reaction was setup as follows:
Digested plasmid: 20uL
AP buffer: 5uL
AP enzyme: 0.5uL
NFW: 24.5uL
The reaction was setup in a thermocycle with a 37°C step for 30 mins, followed by 80°C for 2 minutes to inactivate the enzyme. Ligation was performed as before.

The ligation mixture was transformed into DH5α and plated on tetracycline media. A single colony appeared after overnight incubation. Since the efficiency of traditional cloning decreases after AP treatment, the entire procedure was repeated and three more colonies were obtained after incubation. Colony PCR was performed with the four available colonies and three colonies appeared positive for the insert in the correct orientation.

In order to test the orientation of a non directional cloned insert we chose forward primers from the 5' end of the promoter and reverse primers from the 3’ end of our gene. The results showed that the clone was successful

To test the activity of the ACC deaminase gene in A. brasilense sp7 using the spectrophotometric assay designed by Honma and Shimomura.

The acdS gene was cloned downstream of the exaA promoter in a broad host range pCZ plasmid. This engineered plasmid was then mobilized into A. brasilense sp7 using biparental conjugation. The expression of this gene was supposed to be verified by culturing Azospirillum on minimal media with various alcohols and glycerol as a carbon source (mimicking fermentative hypoxia environments). A spectrophotometric assay was planned to observe the product of the enzyme catalyzed reaction.

The acdS gene was successfully cloned under the exaA promoter and the positive clones in E. coli DH5α were confirmed by colony PCR. After transforming the E. coli S17.1 donor strains with this engineered plasmid, conjugation was performed and transconjugated A. brasilense sp7 were selected by plating them on appropriate antibiotic media (tetracycline and ampicillin). Minimal media was prepaped, however it was seen that the growth of the transconjugated Azospirillum in minimal media was extremely slow as compared to rich LB media and so we could not proceed with further steps

It is possible that the metabolic load of the plasmid in minimal media is significant as even the control setup (with A. brasilense containing the pcZ backbone) also shows very slow growth. We might have to change our cloning backbone to some other broad host range mobilizable vector which might have lesser metabolic load in minimal media. It is also possible that we might need to optimize the composition of the minimal media to suit the growth of the trans-conjugated Azospirillum.

To obtain clear bands of ~1 kB, corresponding to the acdS gene with a clear ladder, and no bands on the negative control lane.

Gel run with a higher concentration of ladder, and run for a longer time at a lower voltage.

Clear bands were observed corresponding to the acdS gene at 1 kB, and were correlated with a clear gel ladder, and no bands in the negative control lane.

The quality and concentration of the gel ladder, and the voltage and time taken to run an agarose gel, all impact the clarity of a confirmatory agarose gel electrophoresis.

To clone exaA promotor upstream of fluorescent reporter mCherry in pet15b vector via RF Cloning.

exaA promoter in pcZ plasmid was used as a template to make megaprimer for RF Cloning. Megaprimer and the recipient plasmid pet15b were used in the secondary PCR. The annealing temperature was set to 50°C. PCR products were digested by Dpn 1 enzyme. Digested products were introduced in NEB electrocompetent E. coli by electroporation.

Colonies were not seen in the plates, implying that cloning had failed.

The annealing temperature might have been too low. The secondary PCR should be performed with a gradient of temperatures to identify the optimum annealing temperature.

To setup a gradient secondary PCR to find the optimal annealing temperature.

While performing the secondary PCR, megaprimer made during the first cyle were used. A gradient from 51°C-60°C was set up for the secondary PCR. Electroporation with digested PCR product was done.

Maximum colonies were seen when the annealing temperature was 58°C and 60°C. After inoculation, the culture did not grow well.

For troubleshooting we referred to literature describing the promotor and we found that exaA promoter forms a secondary structure which might interfere with RF Cloning process

To perform secondary PCR with DMSO to inhibit secondary structures in exaA promoter sequence

Different concentrations of DMSO were added to the PCR mix. The annealing temperature was set to 58°C; Secondary PCR was performed.

After the transformation of digested PCR products into NEB electrocompetent cells, the culture after inoculation showed growth. Confirmatory PCR revealed that our clones were positive.

Two of our samples have been given for sequencing for final confirmation, but we haven’t received the sequencing results yet. After the final confirmation, we plan on characterising the activity of the promotor by preparing minimal media and adding different alcohols in varying concentrations and observing the fluorescence readout.

We began working on the A.B510 GSM by first making sure that the reaction and metabolite names have the standard nomenclature. Since it looked in order, we shifted our entire design to be based on the A.B510 GSM.

We set constraints of the Exchange reactions as required- input reactions would have a lower bound of -1000 and upper bound of 0 and output reactions would have lower bound of 0 and upper bound of 1000.

To check for internal inconsistencies, we then ran FBA with 0 carbon source as input as a sanity check. The expectation was to get 0 growth rates. We also ran FBA with some positive values for the carbon sources, where the expectation was to get some non-zero value as the growth rate.

We got 0 growth rates in all the situations, implying that there was some issues with the model. The first hypothesis was that there are gaps in this model too.

To find the gaps, we ran the GSM findGaps.m function and worked through the FastGapFill tutorial provided by the Cobra Toolbox developers on their Github.

We found 277 root gaps, which lead to 1569 reactions being blocked downstream. Since this number was too big to ignore, we tried running the fastGapFill function.

After a lot of troubleshooting with the code, we found that the Cobratoolbox developers had an error in their code on Github, which was presumably giving us errors at our end. We emailed them, and this issue was resolved at their end in a few days.

We then tried gap-filling again, but there were 0 added reactions to our model, which meant that the function was filling gaps without having to add any new reactions.

After some more troubleshooting, we learned that the function was changing the reaction bounds of exchange reactions to allow both-way flow- into the bacteria as well as out of it. This did not make biological sense for excretory metabolites. We then had no option but to manually look into the dead-ends to see why this issue was occurring. After doing about 15 dead ends, it was evident that the manual curation of the model would again be infeasible. We also learned that the GSM had been made using automation software by the authors too, and we had come full circle back to square one!

We decided to look for a fully curated GSM of the closest related bacteria to Azospirillum. Since Pseuodomonas is closely related to Azospirillum, and it has well established GSMs, we decided to use this GSM to make order-of-magnitude approximations for our system. We could also test whether our logic and workflow is valid and then have the code and its analysis ready for when an Azospirillum GSM is available to us.

We made the GSM complete by adding the reactions specific to our system and the metabolites needed for the same.

From there, we tested the flux through biomass with 0 input carbon sources and free carbon sources, to see if biomass is being produced without having any input, and checking that we have positive flux with free carbon input.

We then ran simulations to extract the maximum rate at which our system can degrade ACC, and the dependencies of ACC and oxygen to the growth rates of the bacteria.

To find the water covered regions from Sentinel-2 optical data using image segmentation.

K-means clustering algorithm was coded in python. Sentinel-2 optical data for a district in Punjab was downloaded.

Clusters were found using k-means clustering and then the water pixels were labelled. These water pixels were different from the real-life data and hence the model was ineffective.

We realised that image segmentation using k-means is not a good algorithm to work with in this case.

To find the water covered regions from Sentinel-2 optical data using image segmentation using R-CNN.

R-CNN was coded in python. Sentinel-2 optical data for a district in Punjab was downloaded.

Clusters were found using R-CNN and then the water pixels were labelled. These water pixels were different from the real-life data and hence the model was ineffective.

We realised that image segmentation using R-CNN is not a good algorithm to work with in this case. The training set was probably ineffective.

To find the waterlogged regions using QGIS and SNAP.

NDWI,NDVI,NDMI and MNDWI ratios were calculated. Sentinel-2 bands data for a district in Punjab was downloaded. Cutoffs for all these ratios were found for waterlogging areas.

All the ratio maps were calculated and the maps were combined together to find waterlogging hotspots.

We found waterlogged areas for the district. We can extend this algorithm to the entire country.