Results
Bacterial Transformation
To transform NEB-5alpha competent bacteria with our reporter constructs and create our GLOW.coli, we agreed to perform a Gibson Assembly. To do this, we started by thinking about the best way to order our inserts - consisting of an inducible promoter on the one hand and a reporter gene (GFP, BFP, TaqRFP) on the other. To do this, we used the Benchling platform, which enabled us to download all the sequences we needed and carry out a Gibson assembly virtually. We agreed to linearise our plasmid pBluescript KS (+) with the enzyme HindIII, as it only occurs once in the plasmid and also has an optimal GC ratio after inclusion of the inserts, which should alleviate PCR. We took advantage of the fact that Gibson assembly requires overhangs that are complementary to the vector by inserting HindIII restriction sites once forward and once reverse into the extensions of the sequences. These do not interfere with the coding sequences but allow the insert to be accurately excised from the plasmid, provided it has been correctly inserted into the plasmid. Accordingly, we would expect to see bands as shown in Figure 1. You can see a virtual digest with HindIII, for Acetate, Butyrates and Propionates for all three reporter genes. The plasmid is 3007bp long, so a band at the appropriate level is expected. The acetate promoter glnAP2 (150 bp) together with the reporter gene GFP has a length of 976 bp, with BFP a total length of 979 bp and with TaqRFP a length of bp. Correspondingly, the bands for the butyrate promoter (pchA: 82bp) appear in the virtual digest. Since the propionate construct contains an additional regulatory sequence (prpR: 1587 bp) as an additional insert, which creates an additional HindIII restriction site in the insert, two additional bands should appear after a successful digest, in addition to the plasmid (promotor prpB: 62bp).
Gibson assembly
The first step of our cloning process involved the linearization of our plasmid with HindIII. To confirm successful digestion, we performed an agarose gel electrophoresis. In addition to the undigested vector as a control, we ran two samples of digested plasmid (c1 and c2). These two samples only differed in the volume of the enzyme used (c1 1µl vs. c2 1.5µl), with no noticeable difference observed. In both cases, we were able to detect linearized plasmid in the gel. Unlike the uncut vector, the linearized plasmid migrates more slowly through the gel due to its open linear structure, which can bind to the gel matrix and therefore move more slowly compared to the undigested, circular, and compact ring-shaped structure. As can be seen from Figure 2, the linearization was successful, so we could use the vector for the Gibson assembly
After performing the Gibson assembly according to the protocol (refer to Lab Book under Gibson Assembly), transforming the bacteria, and streaking them on LB agar plates, we first checked the negative control plate the following day. As expected, the negative control plate showed no colonies, in contrast to the ones with the transformed bacteria. This prompted us to decide to pick colonies to inoculate fresh LB medium.
After eight hours of growth, we used 3 mL oh the suspensions for DNA purification performin a Mini-Prep and 0.5µl of bacterial suspension for a Colony-PCR. The purpose of both methods was to verify the presence of our insert. As mentioned above, we incorporated HindIII sites designed to facilitate the removal of inserts at specific locations. Consequently, after DNA isolation, we performed a digestion with HindIII and subsequently loaded the DNA onto a 2% agarose gel. Upon examining Figure 3, one can observe bands that are smaller than the uncut vector and bands that are larger than the uncut vector (labeled as Plasmid). The higher band served as an indication to us that the enzyme worked. As the 1kb also could not be resolved even after an extended period, we mistakenly assumed the smallest band to be the insert. We attributed the DNA getting stuck in the well and slowed DNA migration through the gel to an excessively high gel percentage.
Figure 4A shows the Colony-PCR results for the Acetate constructs. 'Pro' stands for the primers that only amplify the promoter, while 'Rep' represents the primers that amplify the respective reporter genes. Using the same principle, Figures 4B and 4C were generated for Butyrate and Propionate. Figure 4D shows an attempt to amplify the regulatory sequence prpR. It can be stated that in none of the figures, the Colony-PCR was successful. Bands are visible, but these correspond to non-specific amplifications. Accordingly, we concluded that the Gibson Assembly did not work. Subsequently, by performing a Primer-Blast, primer binding sites in the genome of the used E. coli strain were identified. Since the generation of the primers was based on plasmid DNA and not genomic DNA, this binding was unforeseen. One additional reason the Gibson assembly didn't work is that in the process of vector linearization, we used alkaline phosphatase to remove the phosphate ends to prevent re-ligation. However, we did not consider that this process would interfere with the Gibson assembly.
By using the forward and reverse primers, which, unlike the other primers, do not have primer binding sites within the inserts, we were able to detect the plasmid at the correct size of about 3007 bp for all constructs (Figure 5). This suggests the functionality of the primers.
Since all the results indicated that the Gibson Assembly did not work and we suspected dephosphorylation as the cause, we proceeded to conduct another Gibson Assembly, this time without dephosphorylation. We followed the Gibson protocol as before and used the same plasmid. Prior to streaking the transformed bacteria on the plates, we spread X-Gal (5-Bromo-4-chloro-3-indolyl-β-D-galactoside) on the LB agar plates. This allowed us to perform a blue-white selection. For blue-white selection, a plasmid can be used that contains the gene for β-galactosidase (lacZ gene) at the position where the transgene is inserted into the plasmid (Multiple Cloning Site). Inserting an insert into the Multiple Cloning Site inactivates the galactosidase. As a result, after transformation, the transgenic organisms, in contrast to the non-transgenic organisms, do not contain a functional galactosidase, causing their colonies to remain white, allowing to be isolated based on their lack of color (Fig. 5).
As seen in Figure 6, both white and blue colonies grew. So we used a pipette tip to pick the white colonies and subsequently conducted another Colony-PCR. As Figure 7A demonstrates, bands in the Propionate constructs are visible and correspond to the expected band lenghts. The labelling corresponds to Figure 4.
The insert containing the Propionate promoter includes an additional spacer to extend the short promoter with 62 base pairs to 252 base pairs. In the GFP and BFP constructs, a band at approximately 250bp can be detected, which supports the amplification of the correct band. Additionally, in the wells containing the primers for the reporter genes, bands with 750 base pairs can be observed. Since the reporter genes, GFP, BFP, and TaqRFP, all range between 740 and 750bp, this band also indicates a successful amplification of the desired insert.
Figure 7B displays the Colony-PCR using primers for the regulatory sequence prpR, which is 1616bp long. A prominent band is visible in the RFP construct, which does not exist in the other constructs (exceeding 5000bp). As the entire construct, including the promoter and regulatory sequence, is 5702bp in length, the amplification of the entire construct cannot be ruled out. Overall, it can be concluded that the Colony-PCR for Propionate suggests that the Gibson Assembly worked. Due to a lack of time, planned experiments with the successful Gibson were not conducted in a timely manner but will be continued in the future.
GLOW.coli
Despite the fact that our Gibson Assembly did not work, we proceeded to use our constructs and incubated them with Short-Chain Fatty Acids (SCFAs) as outlined in the lab notebook. Keeping our failed Gibson Assembly in mind, our aim was nevertheless to investigate whether any difference in fluorescence could be observed when exposing the same number of bacteria to different concentrations of SCFAs. To accomplish this, we utilized all nine constructs (Acetate, Butyrate, and Propionate, each with GFP, BFP, and TaqRFP). The concentrations of SCFAs used were 0.01 FE, 0.1 FE, 1 FE, 10 FE, 100 FE, and 1000 FE, with 'FE' representing 'Fecal Equivalent.' To see if our constructs can detect SCFA in concentrations that are biologically relevant we measured the SCFA concentration in Fecal Equivalents. Where we calculated the mass of the specific SCFA in 0.01g, 0.1g, 1g, 10g, 100g and 1000g of mouse feces. So, the total mass of a specific short chain fatty acid, that is present in 1g of mouse feces corresponds to 1 Fecal Equivalent. Which then can be scaled up or down. These calculations were done based on the measurements in Shaidullov, Sorokina and Sitdikov3 et al..
As a control, we used the autofluorescence of the bacteria without stimulation to confirm the effect of SCFAs.
It should be noted here that the concentrations greater than 10-100 FE exceed the concentrations of SCFAs that are realistic to achieve in an experimental setup. However, it was interesting to observe the impact of SCFAs on bacteria when exceeding physiological concentrations.
In Figure 8-10, the graphs depicting the fluorescence over a seven-hour time frame are displayed. At first glance, it is noticeable that the fluorescence in the Acetate (Fig. 8) and Butyrate (Fig. 9) constructs increases significantly with higher concentrations compared to lower concentrations. The same trend is observed for Propionate (Fig. 10), but only in the GFP construct. In the case of all GFP constructs, it is worth noting that, when considering the scale, the fluorescence is significantly higher than in the other constructs, while the range remains consistently comparable for both Butyrate and Propionate. However, since the Gibson Assembly did not work, one possible reason for this could be found in the dilutions. We weighed the powdered SCFAs and subsequently created dilution series. If errors were made in the initial weighing, these errors would have propagated through all experiments since we used the same dilutions for all three constructs.
Often, it is observed that the fluorescence peak decreases over time, especially at higher concentrations. Occasionally, fluctuations are noticeable, alternating between a peak and nadir. In the future, we intend to conduct the same experiment with the constructs where the Gibson Assembly was successful.
Subsequently, we conducted a literature review and found that certain substances can alter the morphology of bacteria, for example, by damaging the cells and thereby increasing autofluorescence.
Discussion
A problem we encountered while performing our experiments was the presence of autofluorescence in our sample which distorted our results. Therefore, it is crucial to find ways to minimize or prevent the presence of autofluorescence to ensure the success of future experiments. Solutions for these problems can be found in literature1.
According to Surre, J., Saint-Ruf, C., Collin, V. et al.1, prokaryotic cells such as E. coli can exhibit an intrinsic fluorescence due to the presence of fluorescent cellular structural components and metabolites. Examples for these compounds are flavines and NAD, which are responsible for the majority of cytoplasmic autofluorescence in E. coli, since they play an important role in cell metabolism. Both, the fluorescence of FAD and NAD are redox dependent; while the oxidized form of FAD exhibits a strong fluorescence, the reduced form is less fluorescent. For NAD it is the other way around.
Because many fluorescent compounds are present in the cell, cellular autofluorescence encompasses a large spectral range since different endogenous fluorophores emit fluorescence at different wavelengths. Therefore, the autofluorescence spectra oftentimes overlap with the spectra of fluorophores that are introduced for research purposes, hindering fluorescence microscopy and flow cytometry measurements. Generally, the autofluorescence depends on the metabolic state of the cells, as well as on different stressors (e.g.: the use of antibiotics), that affect the cells 1 In their studies, Surre, J., Saint-Ruf, C., Collin, V. et al. found that autofluorescence in E.coli increases with bactericidal treatment. For this, β-lactam antibiotic ampicillin was used, leading to an increase in green autofluorescence. Interestingly, this phenomenon could not be observed when a bacteriostatic, protein-synthesis inhibiting antibiotics such as tetracycline were used.Since Carbenicillin, the antibiotic that we used for our experiments, also is a β-lactam antibiotic, another protein-synthesis inhibiting antibiotic needs to be chosen for future experiments to prevent an increase of autofluorescence. For this, the use of another vector that has a resistance to a protein-synthesis inhibiting antibiotic such as tetracycline, might be necessary.
Since the increase of autofluorescence can mostly be attributed to the presence of flavin molecules, and flavine biosynthesis is increased when cells are in severe stress1 (e.g: when a β-lactam antibiotic is used), we expected that the change in antibiotic will lead to a significant decrease of autofluorescence. Since the autofluorescence observed by Surre, J., Saint-Ruf, C., Collin, V. et al. is very similar to the fluorescence spectra of pure flavins, this poses another problem for our experimental set up, specifically, because the emission wavelength of flavines (525 nm)2 is near to the GFP emission wavelength of 510 nm2, and both molecules are excited at similar wavelengths, leading to a spectral overlap.
Lastly, autofluorescence can be further decreased by using PBS instead of cell culture media for the measurements. This is due to the fact that most cell growth media also exhibit fluorescence, since they contain cellular extracts. However, this can be circumvented by the measuring the cells in PBS Buffer instead of LB media or by using M9 medium, for future experiments.
Outlook
For future research, it would be interesting to investigate the metabolic activity and vitality of bacterial cells after incubation with different concentrations of SCFAs. When considering the gut, there are specific locations with a higher fluid flow than others. For instance, the flow is not as strong in a villus due to its folding. Additionally, there are areas, also known as niches, where certain bacteria are more abundant. Since not all bacteria produce SCFAs to the same extent, the concentration of SCFAs could be concentrated in specific areas of the intestine where the flow is less vigorous, leading to a local increase in concentration. This raises the question of whether the potentially resulting higher concentrations may be toxic to other bacterial species, which in turn could alter the balance in the bacterial flora, favoring an imbalance in SCFAs.
References
[1]Surre, J., Saint-Ruf, C., Collin, V. et al. Strong increase in the autofluorescence of cells signals struggle for survival. Sci Rep 8, 12088 (2018). https://doi.org/10.1038/s41598-018-30623-2
[2] Thermo Fisher GFP data sheet
[3] Shaidullov IF, Sorokina DM, Sitdikov FG, Hermann A, Abdulkhakov SR, Sitdikova GF. Short chain fatty acids and colon motility in a mouse model of irritable bowel syndrome. BMC Gastroenterol. 2021 Jan 26;21(1):37. doi: 10.1186/s12876-021-01613-y. PMID: 33499840; PMCID: PMC7836204.