In order to demonstrate that our engineered system- the burden monitor- worked, we tested it under multiple realistic conditions. We decided that we would approach this in three main ways: (1) By using the burden monitor to measure the burden of constructs with known measured promoter strengths, (2), By using the burden monitor to quantify the burden of 330 BioBricks from the iGEM Kits, and (3) By identifying "interesting parts" from the data generated from method 2, and verifying that these parts have functions that are likely to impose burden on a host cell.
Using these 3 methods, we were able to confirm both the validity and the reproducibility of the burden monitor.
Demonstration 1: Anderson Series
We chose the Anderson Series of Constitutive Promoters↗ to demonstrate the validity of our burden monitor. The Anderson Series of Constitutive Promoters consists of parts that each contain a promoter of varying strength, an RBS of fixed strength, and an RFP whose degree of expression is dictated by promoter strength. The 2006 Berkeley iGEM team previously measured the strengths of each of these parts. Having this promoter strength information allowed us to hypothesize that if the burden monitor was indeed accurate, parts with strong promoters would express higher levels of burden and parts with weak promoters would express lower levels of burden on their host cell.
We first confirmed that the parts from the series expressed RFP differentially according to their strengths by growing overnight cultures of each and looking for the color spectrum from dark red to light red to colorless.
The primary indicator of burden is a reduction in growth rate. To identify burden among the Anderson Series parts, we took 5uL of each O/N culture from Figure 1 and placed them in triplicates in a 96-well plate with 195uL of LB media. OD600 measurements were then taken over a course of 6 hours during the exponential growth phase. We expected strains with strong Anderson promoters to have lower growth rates (due to high levels of burden) and strains with Anderson weak promoters to have higher growth rates (due to low levels of burden). The results shown below in Figure 2 confirm these expectations.
Takeaways from Figure 2:
- Anderson Series parts each impose different levels of burden on their host cell, causing them to grow at different rates.
- Cells containing strong Anderson Series promoters grow at a slower rate than those with weak promoters.
After observing growth rates, we introduced a second variable: GFP expression rate. This is where we used the genomic GFP feature of the burden monitor. We carried out the same protocol (loading each culture onto the 96-well plates in triplicates), but this time we also measured GFP expression rate. We expected strains containing strong Anderson promoters to express GFP at a lower rate than strains with weaker promoters. The results below in Figure 3 confirm these expectations.
Takeaways from Figure 3:
- Strains containing strong Anderson Series promoters reallocate more of their ribosomes away from their genome and towards RFP expression (translational burden).
- Strains containing strong Anderson promoters express GFP at lower rates than strains with weaker promoters.
We collated the burden values (percent reduction in growth rate) for the Anderson Series Promoters obtained from the burden monitor measurements into Table 1 below:
Takeaways from Table 1:
- Stronger promoters in the Anderson series generally exhibit higher burden than weaker ones.
Demonstration 2: Large Scale Application
We applied this burden measurement system to 330 BioBricks from the iGEM distribution kits, including all of the frequently used iGEM parts.
As mentioned earlier, we define burden as the percent reduction in growth rate. Figure 4 shows the normalized mean growth rate for all constructs that were measured, and the distribution displays the growth rate reductions associated with those constructs. Burdensome constructs harbor lower growth rates, and this is shown by the progressively darker blue coloring as growth rates get lower. For more information, visit the results page↗.
Takeaways from Figure 4:
- Cells containing different BioBricks grow at different rates.
- Majority of these BioBricks are non-burdensome. This is because most parts do not contain promoters and RBS sequences, which are required to sequester cellular machinery and impart burden.
The application of this measuring system also allowed for us to differentiate between the burden types of constructs. The burden assays provided two main forms of data: GFP expression rates from the burden monitor and growth rates, for each BioBrick. After running all the data through an R pipeline, we were able to study the trends between GFP expression and growth rates to identify if a part: (1) was burdensome, (2) imposed only translational burden, or (3) imposed both translational and "other" burden. Figure 5 helps visualize this and provides evidence that a construct can impose more than one type of burden.
Takeaways from Figure 5:
- Our measurement system provides valuable data to assess the burden of a BioBrick: the growth rate of cells containing each BioBrick, and the types of burden associated with each BioBrick.
- Information extracted from these graphs are used to characterize the burden and evolutionary stability of each part.
- This is a reliable and scalable measurement method, as we were able to successfully measure the burden of 330 different BioBricks.
The data extracted from both of these graphs were later used for the model, which allowed us to characterize the metabolic burden of the BioBricks by calculating their burden values. For more information, visit the measurement page↗ and the modeling page↗.
Demonstration 3: Interesting Burdensome Parts
Figure 6 displays a sample of the interesting parts that we had identified after running burden assays on many parts. The growth rates of these parts are lower than the average "no-burden" parts, and what makes them even more interesting is that many of them deviate from the regression line, which is a sign that they may possess other burden. This would mean that they experience a burden from sources other than ribosomal reallocation. After isolating these interesting samples from a large script output, we were able to identify these BioBricks and learn about their compositions (see table below). Furthermore, it's crucial to discover the distinction between cells with solely translational burden and cells with other burden on top of translational burden. For example, if a cell is experiencing only translational burden, there are a variety of ways in which one could reduce translation until you need expression, thereby minimizing the loss of the construct. However, if the burden is both translational and something else, any attempts to reduce burden may result in the cell still producing something toxic or forgoing a metabolic pathway that's vital for the cell; therefore, killing the cells. With these very real possibilities in mind, understanding these parts and their sources of burden is important in allowing researchers to potentially design more efficient experiments.
Represses the expression of YFP via a cis-repressive element called cr5 which sequesters the ribosome binding site of YFP
RFP Coding Device
Chloramphenicol resistance gene including its native promoter and ribosome binding site
Apple fragrance generator
IAA biosynthetic genes under control of the Pveg2 promoter
Plac + LacZ + bglX. The E. coli periplasmic β-glucosidase gene bglX under the control of the lac promoter. The native ribosome binding site is present.
Constitutive production of LasR trans-activator
LasR/PAI1 Inducible Promoter + RBS (B0034) + GFP
PFixK2-lambda C1-B0034-amilCP blue light sensor output
Strong promoter strong RBS combination for high expression levels of proteins
Takeaways from Figure 6:
- Burden assay data can be used to identify burdensome parts.
- Working backwards in this way (identifying parts that assume burdensome positions on the growth rate vs GFP expression rate graph and then finding out more about the function of these parts) can help us identify a broad range of burden sources.
In summary, we proved the validity and functionality of the burden monitor in three ways: (1) By using the burden monitor to measure the burden of strains containing the Anderson Series of Promoters and confirming that stronger promoters impose a higher level of burden on host strains than weaker promoters, (2) By using the burden monitor to measure 330 BioBricks from iGEM distribution kits and using this data to classify burden into either the 'translational' or 'other' category, and (3) By using the burden monitor-generated data to identify burdensome parts and learn more about what makes a BioBrick burdensome (apart from ribosomal reallocation).