We started all our initial transformations in DH5a strains of E. coli. Cloning in DH5a was never consistent, except for the positive controls. This suggested that something was wrong with our plasmid inside of DH5a.

Of the few transformations that were successful, such as (Fig. 1), we still had issues. When performing Colony PCR on these colonies, hardly any had actually taken up the plasmid (Fig. 2). While 5 out of 7 colonies tested for XMT-MXMT-DXMT plate and 1 out of 7 colonies tested for TDH-KBL plate, all the other genes were negative. Additionally, we had a hard time growing these colonies, potentially suggesting that they had somehow lost the plasmid entirely.

We eventually found out that the reason for the low plasmid stability over generations was due to our selection of a RecA- strain of E. coli such as DH5a (Duarte et al. 2021). The lack of the RecA protein can especially make the chloramphenicol resistance gene susceptible to deletion (Volff et al. 1997). Therefore, we switched to a RecA+ strain of E. coli such as MC1061.

We found much more successful cloning in MC1061 compared to DH5a. Following a 3-day incubation at 30 C, MC1061 E. coli colonies with XMT-Mxmt-Dxmt, TDH-KBL, and FBA-TPI genes grew (Fig 3).

When performing colony PCR on these colonies, 15 out of 15 of all tested colonies of FBA-TPI showed successful insertion, 7 out of 7 of all tested colonies of TDH-KBL showed successful insertion, and 2 out of 7 of all tested colonies of XMT-Mxmt-Dxmt showed successful insertion (5 negative bands likely attributed to gel loading error). Transformations in MC1061 E. coli showed a much higher rate of positive colonies than transformations in DH5a E. coli (Fig. 4).

This higher insertion success rate in MC1061 cells than DH5a cells is likely due to the presence of a presence of the recA protein in MC1061 cells. The recA gene aids in the recombination of homologous DNA, but is engineered to be dysfunctional in DH5a cells in order to prevent unwanted recombination of the plasmid. However, the PNZ8148 plasmid requires the function of the recA gene in E. coli for proper amplification, so MC1061 E. coli being recA+ is likely the cause for the higher rate of positive colonies in MC1061 than DH5a (Azizpour et al. 2017).

Additionally, colonies grew much better for MC1061 cells when incubated at 30 degrees celsius compared to 37 degrees celsius. Initial cloning of assembled PNZ8148 in both DH5a and MC1061 cells with incubation at 37 degrees celsius only produced growth on the DH5a plates. We know transformation worked due to the fact that we were able to verify some colonies with colony PCR. Therefore, it must have been a lack of cloning efficiency that caused low growth in MC1061 cells. When repeated with MC1061 cells at 30 degree incubation, colonies grew (Fig. 3).

In all transformation attempts in MC1061 and DH5a, no colony PCR attempts on SIPDC, PtPDC, or PgPDC showed successful amplification. These being the only constructs that code for phenylalanine decarboxylase, the enzyme which produces phenylethylamine, we speculate that due to lack of full nisin promoter repression in E. coli, toxic levels of phenylethylamine or a deprivation of phenylalanine could arise in E. coli, killing the cell (NICE Expression Handbook).

Two transformation attempts in MC1061 E. coli, with assembled plasmids extracted from positive MC1061 colonies and purified yielded no colony growth. After discussing cloning difficulties with the company which provided our team the PNZ8148 plasmid, we were informed of an accidental mutation in the repC region of the plasmid. We hypothesize that this mutation may be the cause of our amplification difficulties and expect more success with the fixed plasmid.

When we purified the plasmid from our E. coli liquid cultures, we then electroporated that DNA into Lactococcus. However, nothing grew on chloramphenicol plates with both M17 and LB media, even after 5 days at 30 degrees Celsius.

Both LB media and M17 media, a media designed specifically for the Lactococcus species, were used to grow L. lactis. L. lactis grown overnight in LB media had a mean optical density (measured at a wavelength of 600 nm in 1 cm light path) of 1.152, while L. lactis grown overnight in M17 media with the same conditions had a mean OD600 of 1.033. We speculate that there is little significant difference in growth efficiency between L. lactis bacteria grown in M17 and LB media.

In order to characterize the smells of each of the 4 chosen chemicals, we designed smell tests by creating solutions of specific concentrations for each chemical and surveying a group of individuals regarding the intensity of sensation they experienced for specific types of smells. In attempt to replicate the percentage makeup of each chemical as it appears in chocolate in order to calculate the solvent concentrations we would use for the experient, we determined that appropriate solvent concentrations for HDMF (Furaneol), phenylethylamine, pyrazine, and theobromine would be 1 mg/mL, 1 mg/uL, 5 ug/mL, and 10 mg/uL, respectively. We asked each participant to rate the intensity, from a scale of 1-5, of a specific smell in regards to each of the four chemicals. This included the smells sweetness, nuttiness, fishiness, earthiness, bitterness, sourness, fruitiness, caramel scent, roasted scent, saltiness, and vanilla scent. The below graphs show the average responses for each scent between the four chemicals from the survey:

The average intensity of smell in between all scents for furaneol was 2.43 from a scale of 1-5, while for phenethylamine it was 1.63, pyrazine 1.61, and theobromine 1.68. While the experiment was designed to measure the resemblance each chemical had to a specific smell, the strength of a chemical’s odor can often increase the intensity of each scent it resembles and bring out scents that otherwise would not have been noticed in a chemical of generally weaker odor. Therefore, because furaneol’s concentration in this experiment resembles its naturally occurring concentration in chocolate, it can be noted that in order to compare more specifically the resemblance of each particular scent between chemicals, further trials should be conducted decreasing the concentration of furaneol to more accurately distinguish between scent resemblance and general chemical odor strength. That furaneol’s concentration is closer to its natural presence in chocolate shows it dominates the smell profile, but adjustments to the concentration could give a clearer picture of the roles played by the other chemicals.

Additionally, we asked participants to give their own short-answer input as to what the chemical smelled like to them. For furaneol, 75% of responses commented on its sweet, caramel scent. 71% of responses regarding phenethylamine commented on its fishy scent. Pyrazine had more varied responses, 20% of which noted its fishy smell, 20% of which noted its caramel smell, 20% of which noted its nutty smell, and 20% of which noted its fruity/raspberry scent. Likewise, theobromine had varied responses, with 36% of participants commenting on its fishy smell, and a variety of comments on its garlic, marinara, vinegar, alcohol, and musky scents.

In characterizing the scents of each of the chemicals, we find that furaneol dominates the smell profile, and reducing the concentration of furaneol in further experiments that involve mixing chemicals in different concentrations will likely be necessary to ensure the right balance of chemicals needed to resemble the chocolate smell. Furthermore, in highlighting the sweet, caramelized scent of furaneol, the fishy scent of phenethylamine, the more varied yet fishy smell of theobromine, and the distinctly varied and complex smell of pyrazine, we are able to create standards to compare to after producing the chemicals via gene expression in L. lactis.

Now informed of the presence of the repC mutation in our original PNZ8148 plasmid, we plan to repeat the assembly of our constructs using the fixed plasmid. The repC mutation could explain many of the difficulties we faced cloning in both E. coli and L. lactis.

Additionally, we plan to find out what is causing the phenylethylamine pathway to not grow inside of E. coli. We plan to do this in two steps: 1) check for lack of promoter repression in E. coli by creating a green fluorescent protein pathway and checking whether or not there is fluorescent even without nisin present, and 2) directly transform ligated DNAproducts into Lactococcus, thus skipping through E. coli cloning.

We also hope to conduct additional experiments in order to better determine the concentrations of mixed chemicals that best produce a chocolate-resembling odor. We hope to perform further smell tests surveying human responses to mixtures of our chemicals with different concentration conditions. As noted in the above description of our smell test, we would also like to repeat the original smell test experiment with the same conditions, but with a lower concentration of furaneol, in order to better characterize scent resemblance regardless of chemical odor intensity.