Overview

We achieved it!

We used a wide variety of analytical techniques to check for the results of our experiments. At first, Thin Layer Chromatography(TLC) was used to detect the presence of C16 and C18 FFA in our samples and compare their intensities. When we couldn’t get a concrete inference from TLC, we proceeded with Liquid Chromatography - High Resolution Mass Spectrometry(LC-HRMS). Here, we could clearly see an overall increase in the free fatty acids(FFA) content with a specific increase in C18:1 FFA. The FAA1 deleted strain showed a 3-fold increase in fatty acids as compared to our control strain.

For validating the design of our engineered efflux pump, GROMACS Molecular Dynamics(MD) simulations were run, and the calculated trajectories were analyzed and interpreted. An overall trend supporting our hypothesis was observed, showing greater binding affinity of the engineered pump towards alkanes as compared to fatty acids. This result was compared and contrasted against the trends observed in the wild-type ATP-Binding Cassette(ABC2) transporter, which displayed similar degrees of binding affinity towards both alkanes and fatty acids

Thin Layer chromatography

Experiment

To check for the increase in the free fatty acid composition caused due to the expression of JcFatA and JcFatB, we first decided to perform a TLC. We performed TLC with 3 samples - Po1g transformed with empty pYLEX (control), Po1g transformed with pYLEX-JC(G1), Po1g transformed with pYLEX(G3). We normalized the OD₆₀₀ values and made cell pellets of the samples and extracted the FFA using the chloroform-methanol extraction (Folch extraction).

The samples after drying were re-suspended in 100 µL of chloroform and then spotted on the silica TLC plate. We also spotted the C16:0 (palmitic) and C18:0 (stearic) fatty acid standards along with these samples to get to know if we have the desired fatty acids or not.

The mobile phase used was 60% ethyl acetate in hexane. The TLC plate was then stained with PMA(phosphomolybdic acid) and slightly heated over a flame to visualize the spots.

C₁₆ - palmitic acid (C₁₆) standard

C₁₈ - stearic acid (C₁₈) standard

P - Po1g with empty pYLEX (control)

M - mix spot of P and G3

G3 - Po1g with pYLEX-JC insert (colony 3)

G1 - Po1g with pYLEX-JC insert (colony 1)

Inference

We cannot distinguish between the C₁₆ and C₁₈ fatty acid composition of the samples by TLC. The intensities of all the spots are quite similar except for the fact that the FFA content in G1 is slightly higher as compared to P.

TLC is unable to show the difference between C18:0, C18:1, and C18:2 FFA composition in the sample, so we had to proceed with LC-HRMS.

From the TLC we can infer that we have C₁₆ and C₁₈ FFA in our samples.

Liquid Chromatography-High Resolution Mass Spectrometry (LC-HRMS)

We sent 5 samples for LC-HRMS:

• Po1g transformed with empty pYLEX (A), Po1g transformed with pYLEX-JC(G1) (A), Po1g transformed with pYLEX-JC(G3) (A) - all normalized to OD₆₀₀ = 12.
• Po1g transformed with empty pYLEX(B), FAA1 (B) - normalized to OD₆₀₀ = 5.

The FFAs were extracted from the cells using the chloroform-methanol extraction method and the sample was completely dried using a centrivap. The vials were then parafilmed and sent to the CSIR-National Chemical Laboratory’s (NCL) LC-HRMS facility for analysis. There they were resuspended in a 2:1 ratio in 200µL acetonitrile:methanol and 10µL of the sample was injected into the LC-HRMS equipment. A standard of a known concentration of 15:0 fatty acid was used for quantification.

Upon receiving the result, we isolated the peaks corresponding to our fatty acids using their molecular weights and formulae. We got the expected peaks for 16:0, 18:1 and 18:2 fatty acids, but did not get any peak for 18:0.

Upon receiving the raw data, we analysed it by extracting the areas under our relevant peaks, subtracting the area (if any) under blank, and then finding the total area (the given area values were only for 10 uL injection whereas the total resuspension in solvent mix was 200 uL) and then normalising the area with respect to the area under the standard. This gave us the amount of each of our components present in nanomoles. We normalised it further to find it for unit OD and per mL of culture (Note: Here OD value has been used as a proxy for cell number). This was then plotted for each fatty acid peak for each of our samples and we were in for a pleasant surprise.

Here we see that upon adding our thioesterases to the system, the composition of the fatty acids had an overall increase in the organism. We hypothesise that this may be simply due to the addition of more thioesterases to the system. The native FAS complex of Y. lipolytica already makes fatty acids, but if there is excess of the Acyl-ACP/Acyl-CoA substrate leftover then these can be utilised by the Jatropha curcas thioesterases to produce more fatty acids, thus increasing the overall content.

One confounding result was that we did not get any peaks for 18:0 (stearic) fatty acid in our LC-MS graph. We hypothesise that this happened because of the native desaturases of the yeast which convert 18:0 fatty acid to 18:1 and 18:2 by dehydrogenation. So even if the Jatropha curcas thioesterases had produced 18:0 fatty acid, it may have been converted to the unsaturated 18:1 and 18:2. This could also help partly explain why there is a significantly higher yield of 18:1 and 18:2 fatty acids as compared to 16:0, which is not significant.

To check how the composition (percentages) of our fatty acids changed, we plotted each fraction with respect to the total fatty acid composition analysed. So for example, to compare 16:0, we divided 16:0 normalised area by (16:0 + 18:1 + 18:2) normalised area to get the percentage of our analyte mixture that was 16:0 (palmitic) fatty acid. We did this for each sample and for each fatty acid. We also calculated the same values for the Jatropha oil fatty acid composition from literature (since the values varied among sources, we took acceptable ranges) and made the comparisons to obtain the following graph.

From this graph we can see that although we did not replicate the exact composition of the fatty acids in jatropha oil-based SAFs, we got intermediate values for the percentages of 18:1(oleic acid) and 18:2(linoleic acid) between our empty control and Jatropha oil-based SAFs. This is expected as we will not get the exact compositions of the Jatropha curcas plant’s fatty acids upon expressing it heterologously in a yeast. So we have got compositions which seem to be a combination between the two. We can’t see the same changes in 16:0 and we think this could be due to the already low abundance, so the differences may not be significant with respect to experimental error. Thus it is best not to make any comments about that data.

We had also attempted to perform deletions of genes in the native organism involved in fatty acid degrading processes like FAA1 responsible for beta-oxidation and Alk1 for alkane metabolism to fatty alcohols. Upon talking to scientists we had learned that it usually takes years to perform such deletions, so we weren’t very hopeful about the success of this part of the project. However we succeeded in the selection for the deletion strains one day before our LC-HRMS and thus were able to include this along with a separate control normalised at OD₆₀₀ = 5 for comparing to see if our overall fatty acid production had increased in this strain. We were thrilled to see such strong results for this.

Conclusion: We were successful in the transformation of our chassis Y. lipolytica with the thioesterases JcFATA and JcFATB from the plant Jatropha curcas as is clearly seen by the significantly altered fatty acid profiles of the transformed organism. We also succeeded in performing the deletion of FAA1 in our chassis to make it produce an overall larger amount of fatty acids and thus fuel.

Scope for Improvement

Due to lack of time and space, we were not able to perform duplicates and triplicates for our LC-HRMS results as we would have preferred to. This can be done to ensure further robustness of results.

Untransformed Po1g strain would have been the ideal control for comparing our FAA1 deletion to, as teh deletion was performed in previously untransformed organisms. However the untransformed Po1g had a very slow growth rate as compared to the FAA1 deletion strain and due to this we weren’t able to grow it upto a good OD₆₀₀ value in the time crunch we faced towards the end.

Data for native fatty acid compositions of Y. lipolytica are very variable across strains and also vary across different growth conditions. Due to this, there was no reliable literature available for the exact fatty acid composition of Po1g strain grown in YPD medium. Hence we had to take our control composition to be the native composition. This assumption based on a single measurement is not ideal and more experiments should be undertaken to establish the native fatty acid composition of this strain.

Hydrocarbon Efflux Pump Engineering

Our aim was to engineer the already existing ABC2 transporter from Y.lipolyitca with the objective of biasing the specificity of the pump towards alkanes compared to fatty acids. This could enable the extraction of the produced hydrocarbons directly from the medium feasible and also ensures that the yeast cells do not die of the toxic effects of the increased hydrocarbon accumulation.

Using cycles of docking, mutations and structural predictions, we arrived at a preliminary design for the engineered efflux pump from the wild type ABC2 transporter. This design was validated by running 100ns GROMACS MD simulations for the following systems:

• WT efflux pump with heptadecane as the ligand
• WT efflux pump with palmitic acid as the ligand
• Variant efflux pump with heptadecane as the ligand
• Variant efflux pump with palmitic acid as the ligand

For each system , the simulations were ran thrice and the generated 4 x 3 =12 trajectories analyzed. A trend of increased binding affinity towards alkanes and a decreased binding affinity towards fatty acids was observed in the case of the variant pump in comparison with the wild type ABC2 transporter, hence validating our engineering efforts.

A detailed description of the 12 MD trajectories and the associated visualizations can be found in the Efflux Pump Engineering page linked here.