Experimental Setup
Our experimental data were gathered by transforming E. coli with the constructed plasmids and measuring the optical density (OD) and fluorescence of the corresponding cultures. This series of experiments allowed us to characterise vsfGFP-0 with respect to sfGFP, the strength of the Lac promoter, as well as to characterise the Growth Slowing (GS) genes. A typical experiment was conducted in a 96 well plate and its contents were measured using a Hidex Sense Microplate Reader Type 425-311 (g/n 320-0250). There were 150µL of culture in each well in total. 130µL of these were made up of 2.5% Millers LB media with kanamycin (50ng/µL) and chloramphenicol (25ng/µL). The remaining 20µL were made of the relevant culture diluted to an OD of 0.05.
These experiments involved monitoring the optical density (OD-600nm) (and fluorescence for fluorescent cultures) of LEMO BL21 transformed E. coli every 15 minutes during a 24-hour period. For the experiments involving pLac, several IPTG concentrations were tested, ranging from 0 to 1 mM. For the Anderson promoter experiments, a different promoter from the catalogue was used in each construct. Our computational analysis was based on visual comparisons of the bacterial growth and fluorescence behaviour using the R software to produce the required graphs.
vsfGFP-0 Characterisation (Construct 1): First Attempt
Our first experiment intended to measure OD and fluorescence values of the construct containing vsfGFP-0 under the control of pLac, for an IPTG range from 0µM to 1000µM. The findings shown in Figure 1 indicate that the induction of the promoter resulted in a sustained increase in fluorescence when the vsfGFP-0 was present, compared to the empty pET28a backbone. A different sfGFP-containing plasmid (pJUMP28-1a) was included as a positive fluorescence control. In later measurements, at high IPTG concentrations, all the samples with this construct overpower the sensor after 2,400,000 relative fluorescence units (RFU). In addition, it was also seen that samples with high IPTG concentrations exhibited shorter lag phase periods. Although our results confirmed the fluorescence increment after IPTG-inducible promoter activation, some data recordings were incomplete as the optical sensor became saturated before we could observe a proper plateau in them. The 96-well plate maps (found here: supplementary data) include a detailed description of all the assessed well plates.
Fluorescence increased proportionally to IPTG concentration only when the vsfGFP-0 insert was present in the plasmid (rows A and B). The empty pET28a plasmid showed no fluorescence increase (rows C and D).
OD was increased after IPTG administration to a greater extent when the vsfGFP insert was present in the pET28A plasmid (rows A, B, E, F). The optical sensor was overwhelmed in these conditions but not in the rest of the samples. These included negative controls of PET-28 without any fluorescent genes or the inducible promoter, respectively (rows C and G). Also a different plasmid pJUMP28-1A was included as reference (row D). Lastly, a pET28a plasmid with fluorescent genes but lacking the promoter was also included as negative control (row H).
vsfGFP-0 Characterisation (Construct 1 V2): Second Attempt
In our previous experiment, the optical sensor was saturated by the fluorescence of most of the induced cultures. As a result, the working IPTG concentration range was reduced to 0 to 5 µM. vsfGFP-0 expressed by the constitutive Anderson promoter J23110 was also added as a positive fluorescence control. Additionally, this experiment intends to compare the fluorescence of vsfGFP-0 and its parent molecule, sfGFP-0. Therefore, the fluorescence of cultures containing plasmids with these genes (in a pET28a vector) was measured. In a similar way as before, fluorescence and OD were measured every 15 minutes during a 24 hour period and the results are shown in Figure 3 and Figure 4.
All inducible constructs showed an increase in fluorescence as the concentration of IPTG increased, although those from rows A and B had a much steeper increase than those in rows C and D, as shown in Figure 2. This suggests an intrinsic superiority in the brightness of the protein expressed by the bacteria. The growth control, as expected, showed near-zero levels of fluorescence and the fluorescence control saturated the sensor. Well D1, which was meant to be a blank, was contaminated when loading the plate, which matches the unexpected growth shown by the measurements.
When formatting the plots depicting fluorescence with respect to time, it is very easy to see that for every sfGFP construct at a particular IPTG concentration there is a vsfGFP-0 analogous that exhibits a higher fluorescence corresponding to the same inducer concentration every time. See Figure 3 where the fluorescence values have been corrected against the fluorescence of the blank.
Processing the data further gives some more interesting insight. If the mean values of fluorescence at a given time are divided by the optical density at that time, the quotient will be a magnitude representing "fluorescence per cell". This is the most important metric. Since GFPs (and in particular sfGFPs) are so stable, it is necessary to obtain a normalised value for fluorescence (as indicated in Equation 9 (from biological modelling)), since in time the GFPs will build up and the values of fluorescence will become an indicator of how long the bacteria have been expressing the protein rather than how fluorescent the protein actually is. In Figure 4 it can, once more, be clearly seen that the F/OD increases as the inducer concentration increases and the vsfGFP-0 always leads in magnitude with respect to its sfGFP analogous when both are subjected to the same IPTG concentrations. Moreover, it is important to note the trough in F/OD at approximately four hours. This might indicate an offset between the times when the logarithmic phase begins for growth rate and when it begins for GFP production, or it could also denote the required maturation time for GFPs to begin fluorescing. It seems as though cells begin dividing rapidly roughly five hours before they begin producing protein, and are subsequently surpassed by a rapid increase in GFP production. This would explain the decrease in the F/OD quotient from when measurements started to be recorded until around five hours later.
Finally, utilising all the data points measuring fluorescence per IPTG concentration we can plot the trend followed by the cultures, shown in Figure 5. This indicates that the fluorescence of the cultures with each of the reporters tested increases with the concentration of inducer. However, vsfGFP-0 increases significantly faster than sfGFP establishing, indicating considerably higher brightness. At the highest IPTG concentration, overall, the measured fluorescence of the culture containing the vsfGFP-0 gene was 2.5 times more fluorescent than that of sfGFP. It should be highlighted that these trendlines do not cross the origin. Additional to potential systematic errors in measuring, this could depict leaky expression of the T7-pLac which manages the expression of the GFP gene, as has been previously described in literature[2]. The error bars were measured based on the standard error of mean of each point. It should be noted that mean was done between the corresponding plasmids (A and B or B and C). Moreover, as seen at 3.5μΜ, the error bars overlap. This happens because the two values have a range higher than 500,000 RFU OD-1. In addition, these relationships could be used further to predict relative maximum F/OD depending on the concentration of inducer. This shall be explored further.
In conclusion, it seems reasonably clear the improvement vsfGFP-0 presents over its parent molecule sfGFP as a reporter, as equivalent gene expression results in significantly higher fluorescence in the former.
For additional information see our registry part page: http://parts.igem.org/Part:BBa_K4939000
Promoter Activity Calculation
As shown in Figure 1 from the Biological Modelling section, the promoter activity is a magnitude that quantifies transcription and translation of vsfGFP-0 stimulated by the pLac promoter. This measure is extremely important as it can be used as a means of calibrating the pLac inducible promoters against constitutive promoters such as the Anderson promoters. Based on the modelling Equation (11), it is expected that the promoter activity is affected by the growth rate (\(\mu\)), the fluorescence per cell at steady-state (\(f_{ss}\)) and the maturation constant (\(m\)).
The parameters are determined from the graphs illustrated above. The growth is determined from the relationship of the log2OD over time, where the slope is calculated. The \(f_{ss}\) is determined from the F/OD graph illustrated in Figure 4 above, where the plateau indicates the maximum fluorescence per cell, while the maturation constant is determined from the Equation (1) below after finding the maturation time. The maturation time is the period taken when the non-fluorescent vsfGFP-0 protein begins to fluoresce. Assuming that the maturation time is similar to the parent strain, sfGFP, is 13.6 min, the maturation constant is calculated as follows:$$\begin{equation}\frac{\ln(2)}{\text{GFP Maturation time (h)}} = m\end{equation}$$$$\begin{equation}\frac{\ln(2)}{13.6/60} = 3.0059\end{equation}$$
As a result, The table of the measurements and calculations can be seen below in table 2 Promoter activity Calculation:
The trend shown in the relationship between the two variables is an illustration of a sigmoidal curve as modelled by Hill’s kinetics..It represents the non-linear behaviour of the promoter activity rate as it slowly decreases. Therefore, it is estimated that the maximum promoter activity is at \(1.45\times 10^6\text{RFU}\;\text{OD}^{-1}\text{h}^{-1}\) with the equilibrium constant being \(0.225\times 10^6\text{RFU}\;\text{OD}^{-1}\text{h}^{-1}\)
Using the Anderson promoter activity values,the promoter activity of pLac at varying [IPTG] can be reduced by taking the same particular F/OD value for Anderson and pLac promoters and assuming that the level of fluorescence is indicative of the same level of gene expression and thus promoter activity. This can be validated by calculating the promoter activity of two of the Anderson promoters, and using their strength relative to one another, to confirm that the change in promoter activity between the two is still correlated to this difference in strength. This methodology constitutes a way to determine absolute promoter activity metrics for virtually any inducible promoter, with the only parameter needing to be swapped out within the experimentation being the promoter and associated inducer.
However, this is only theoretical as the Andersons over-saturated the sensor and therefore, the steady state was never measured. Furthermore, more data would be needed in order to predict the promoter activity with more accuracy and for more ranges of IPTG.
Anderson Promoters + vsfGFP-0 (Construct 2)
A relevant part of our project consists of characterising the strength of the selected Anderson promoters in our genetic system. Anderson promoter collection represents a versatile option to test transcription activity. It comprises a wide-covering catalogue of constitutive promoters with constant activities with known values (promoter strength), defined by their particular nucleotide sequence. In our case, this series of experiments would have helped to determine the fluorescence values that we should expect during a lesser or greater activity of our PARSE genetic system.
We found that fluorescence increased for all the Anderson constructs whereas it did not for the positive and negative growth controls (uninduced pLac-vsfGFP-0 and LB respectively). The fluorescence control with sfGFP-pJUMP28-a showed an increase in fluorescence as well. The rate of fluorescence increase of the Anderson constructs was different depending on the promoter used and followed the qualitative behaviour expected based on their relative strengths described in the Anderson catalogue page.
The fluorescence increase rate and the maximum fluorescence values were taken into account for the calculation of the overall activity of our system. These calculations were essential for the subsequent comparison and estimation of the PARSE system activity using an inducible promoter.
Fluorescence intensity increased at a high rate for the plasmids cloned both with J23106 and J23110 and a lower rate for J23114 (see figure 10), which was expected given that their strengths are meant to be comparable and high with respect to J23114 based on the Anderson catalogue page. However, J23106 showed a slightly lower increase rate than J23110 which contradicts the relative strengths recorded in the Anderson catalogue. J23100 was also cloned and tested but showed a very slow fluorescence increase rate, comparable to the uninduced pLac, which was very unexpected since this was, theoretically, the strongest promoter. We later confirmed that the genetic sequence cloned was not right through sequencing.Fluorescence intensity increased at a high rate for the plasmids cloned both with J23106 and J23110 and a lower rate for J23114 (see figure 10), which was expected given that their strengths are meant to be comparable and high with respect to J23114 based on the Anderson catalogue page. However, J23106 showed a slightly lower increase rate than J23110 which contradicts the relative strengths recorded in the Anderson catalogue. J23100 was also cloned and tested but showed a very slow fluorescence increase rate, comparable to the uninduced pLac, which was very unexpected since this was, theoretically, the strongest promoter. We later confirmed that the genetic sequence cloned was not right through sequencing.
It is worth noting that, initially, the team intended to form the link between gene expression and fluorescence (or growth rate in the case of the GS) with respect to the less-arbitrary (i.e. constitutive) Anderson promoters, rather than the inducible pLac. Nevertheless, since the data for the Andersons could not be measured entirely as the plate reader’s sensor was overwhelmed, we resorted to characterising vsfGFP-0 and the GS employing our inducible system.
In parallel, the effect of the Anderson promoters on the bacterial growth in E. coli was studied. The OD600 values were simultaneously measured in the same samples and collected during the full duration of the experiment. We found that bacteria transformed with a plasmid containing vsfGFP-0 under the control of an Anderson Promoter divided slower when compared to vsfGFP-0 under the control of pLac, as well as the empty pET28a vector and a GFP-containing plasmid (pJUMP28-1a) as independent constitutive GFP expression control(Figure 11).
One possible explanation for the decrease in this natural cell function can be attributed to global protein synthesis restraints caused by the metabolic burden of augmented GFP production in these genetic profiles [1]. The metabolic burden can be defined as the draining of nutrients, raw materials and energy from the cell metabolism as a result of an abnormal increase in protein production. In our case, the directly affected function is the bacterial growth. Further experiments would be required to confirm this hypothesis.
Growth Slowers (Construct 3): Characterisation of the expression of the PARSE system on the bacterial growth in E.coli
The PARSE genetic system developed by iGEM Sheffield is composed of a pET28a plasmid containing one GS gene. When activated, by means of IPTG (the inducer), the GS is expressed and the expected response is a decrease in the rate of growth of the culture. This response can be modulated by varying the concentration of IPTG. To confirm that the GS genes worked in E.coli, and to calculate the growth slowing impact they have on the bacteria, a series of experiments were performed.
The first test consisted in determining if the sole presence of the pET28a vector produced any inhibitory response on bacterial growth. For this purpose, the effect of different concentrations of IPTG was compared in pET28a-transformed E. coli (see figure 14). Despite some inhibition observed in the lowest concentration tested (1µM), the higher concentration (2 -4µM) did not show any inhibitory effect on growth. With this result, we discarded the possibility of an intrinsic inhibitory effect of the vector.
Considering that a PARSE plasmid contains both a GS and a fluorescent reporter gene, the next objective was to confirm that the GS was the only significant cause of the resulting growth inhibition. From the previous experiments, the highest tested concentration of IPTG was taken (4 µM) to compare the bacterial growth when the vector contained a GS gene and when it contained vsfGFP-0. This would help determine whether the growth-slowing effect observed was due to the metabolic burden imposed by sheer protein production rather than the identity of the protein produced.
A representative example of our findings is shown in Figure 15, where bacterial growth was markedly affected when the GS was present in the plasmid. If growth-slowing depended mostly on metabolic burden only it would be likely that the constructs transformed with vsfGFP-0 would grow at a lower rate than the ones transformed with the GS, since compared to most GS, vsfGFP-0 is made of at least three times as many residues.
We decided to test a variety of growth slowers with different functions, some from E. coli phages and some from Staphylococcus phages. This way we tested both for feasibility proposing different applications as well as context-independence.
By comparing the increase in OD at different IPTG concentrations with a growth control we can determine whether the GS have indeed worked or not. Figures 14-19 show this. Out of the six cloned growth slowers, gp2, gp240 and AsiA yielded positive results. Gp240 is a DNA clamp inhibitor derived from a Staphylococcus phage. The two other successful growth slowers come from Escherichia phages. This could suggest homology between the clamp structures of both species, and may further implicate their broad applicability to several genera of bacteria. Additionally, gp240 has a different function to gp 2, which is an RNA polymerase inhibitor, and having interventions at different levels in metabolic pathways allows for a variety of applications.
On the other hand, gp79, gp104 and AsiA did not yield any significant change in growth rate with respect to the growth control. Later, it was confirmed through sequencing that the Alc construct had an incorrect DNA sequence.
The next comparison was focused on the growth inhibition effect of a wider IPTG concentration range. An example of this comparison is shown in Figure 20 (for gp240 gene) and Figure 21 below (for AsiA gene). As expected, a higher expression of the GS gene (by means of a higher IPTG concentration) was reflected in reduced bacterial growth. The highest concentration included (1000 µM) exhibited the maximum growth inhibition. A similar comparison was performed for each GS gene.
Bacterial growth (in OD units) was progressively reduced when gp240 expression was induced by inductor presence (0-1000 µM). The greatest decrease occurred in the presence of 1000 µM IPTG.
Bacterial growth (in OD units) was progressively reduced when AsiA expression was induced by inductor presence (0-1000 µM). The greatest decrease occurred in the presence of 1000 µM IPTG.
Finally, several GS were simultaneously compared to identify differences between their growth inhibition effects. Taking into account that each GS acts through a different mechanism of action, this comparison could help to select the best GS-inducer concentration - response profile according to particular experimental needs. This comparison for the three successful GS is shown in figure 22.
Bacterial growth (OD) was progressively reduced as GS expression was increased by increasing inducer concentration (0-1000 µM). In all cases, the greatest decrease occurred in the presence of 5µM IPTG.
Supporting Information for the R-code and supporting files can be found in this Gitlab repository
References
[1] Bentley, W. E., Mirjalili, N., Andersen, D. C., Davis, R. H., & Kompala, D. S. (1990). Plasmid-encoded protein: the principal factor in the "metabolic burden" associated with recombinant bacteria. Biotechnology and bioengineering, 35(7), 668–681. https://doi.org/10.1002/bit.260350704 Rahmen, N., Fulton, A., Ihling, N., Magni, M., Jaeger, K. E., & Büchs, J. (2015). Exchange of single amino acids at different positions of a recombinant protein affects metabolic burden in Escherichia coli. Microbial cell factories, 14, 10. https://doi.org/10.1186/s12934-015-0191-y