Results

In the following subsections, we present our fluorescence measurements over time, a cytometry study showing a more microscopy view of how our cells behaved and our calculations of promoter strength using Relative Promoter Units (RPU). Notice that all of our data is publicly available here, and all software developed to calculate the results presented below are available here.

Fluorescence results - plate reader

As described above, we have recorded fluorescence and optical density (OD) over time of a 96-well plate organized as mentioned before. We show below fluorescence measurements after 11h of experiments, when all colonies presented an OD of about 0.5 u.a., except of course LB and LB+Cam which do not have any bacteria. Notice that we have for three biological replicates of three technical replicates, as suggested for extra credit. If you are interested in our complete data, we have made it public here.

Table 1 (click for larger view): Fluorescence readings after 11h of experiment, not normalized by OD yet (see data below).

LB and LB+Cam are our baseline controls: they set the zero-level for the fluorescence. Although their levels are very close indeed, a Mann-Whitney U-statistic test showed that the antibiotic Cam is inducing a fluorescence that should not be disregarded (p-value < 0.5%). Thus, we will normalize all of our measures with respect to LB+Cam activity.

To better visualize this data, we show in the figure below The activity of each device averaged for each colony (considering the standard deviation). This fluorescence measure is normalized by LB+Cam. Notice how Device 3 fluorescence is close 1.0, i.e., not much is induced more than our zero-level activity. This is somehow expected, as this promoter is indeed very weak (see Relative Promoter Units below).

Figure 2 - Fluorescence per cell for each of the three colonies and for each device. While across colonies there is some difference, within the same colony the standard deviation is rather small. For all of our cases, Device 1 is the stronger promoter, followed by Device 2. Device 3 was always the weaker promoter.

The activity difference is definitely greater than the standard deviation of our measurements, guaranteeing our the hierarchy Device 1 > Device 2 > Device 3. We can see the time evolution of the fluorescence relative to the baseline control for each device.

Figure 3 - Time evolution of the relative fluorescence per cell, normalized by the LB+Cam activity. All colonies are in their log-phase, suitable for evaluating their promoter strengths (see below). Notice that for all times, Device 1 is stronger than Device 2, while Device 3 is by far the weaker.

The figure above clearly shows that Device 1 induced significantly more activity. Still, Device 2 seems to perform better than Device 3. However, to quantitatively ensure that these activity differences are statistically relevant, we have performed a Mann-Whitney U-statistic test (p-value under 0.01%). Therefore, Device 1 is definitely stronger than Device 2, which in turn is also stronger than Device 3.

Cytometry

Here, we performed flow cytometry with biological triplicate. Because populations are well distinguished, quadrant gating method can be used to discern GFP expressing cells from those not expressing the GFP selection marker. FL1-H vs. FL2-H plots were divided in four quadrants - Q4 (lower left), Q3 (lower right), Q2 (upper left) and Q1 (upper right) - defined by analyzing negative control with no plasmid and fitting all events in the LL quadrant. Therefore, as FL1 detector (530/30 filter) collects green emission we obtained fluorescence intensity histograms for each sample, FL1-H vs. Counts.

Negative control (no plasmid)

Autofluorescence from DH5-alpha E. coli cells and histogram markers FL1-H- and FL1-H+ were set, therefore, any event in the Q3 quadrant represents GFP fluorescence.

Figure 4 - E. coli DH5-alpha cells. The cells were excited at 488 nm and detected using FL1-H (FITC channel). At the left, dot plot obtained by flow cytometry correspondent to autofluorescence whereby the lower left quadrant was defined to contain all cells as control. Histogram on the right shows autofluorescence levels. Both dot plot and histogram were used to set baseline and threshold.

Negative control (with plasmid)

As a second negative control, DH5-alpha E. coli cells containing a chloramphenicol resistant plasmid (pSB1C3) were used and no significant difference in the cells autofluorescence was detected when compared with wild DH5-alpha E. coli cells.

Figure 5 - E. coli DH5-alpha cells with chloramphenicol resistant plasmid pSB1C3 without insert DNA. The cells were excited at 488 nm and detected using FL1-H (FITC channel). At the left, dot plots obtained by flow cytometry correspondent to cells fluorescence and, at the right, histograms analysis of fluorescence levels.

Positive control

For comparaison purposes, a positive control expressing GFP (BBa_I20270 - a GFP expression device in the pSB1C3 backbone (chloramphenicol resistant) was also measured.

Figure 6 - E. coli cells GFP expression controlled by BBa_J23151 promoter. GFP was excited at 488 nm and detected using FL1-H (FITC channel). At the left dot plots obtained by flow cytometry correspondent to BBa_I20270 and histograms on the right show analysis of GFP expression in biological triplicate

Device 1

Figure 7 - E. coli cells GFP expression controlled by BBa_J23101 promoter. GFP was excited at 488 nm and detected using FL1-H (FITC channel). At the left dot plots obtained by flow cytometry correspondent to Device 1 (J23101 + I13504 ) and histograms on the right show analysis of GFP expression in biological triplicate

Device 2

Figure 8 - E. coli cells GFP expression controlled by BBa_J23101 promoter. GFP was excited at 488 nm and detected using FL1-H (FITC channel). At the left dot plots obtained by flow cytometry correspondent to Device 2 (J23106 + I13504 ) and histograms on the right show analysis of GFP expression in biological triplicate

Device 3

Figure 9 - E. coli cells GFP expression controlled by BBa_J23101 promoter. GFP was excited at 488 nm and detected using FL1-H (FITC channel). At the left dot plots obtained by flow cytometry correspondent to Device 3 (J23117 + I13504 ) and histograms on the right show analysis of GFP expression in biological triplicate

Comparison of expression rates of J23101, J23106 and J23117

Using flow cytometry we also conclude that promoter strength decreases from J23101>J23106>J23117 (Device 1>Device2>Device3).

Figure 10 - Comparison of GFP expression controlled by BBa_J23101, BBa_J23106 and BBa_J23117 promoters. GFP was excited at 488 nm and detected using FL1-H (FITC channel).

Relative Promoter Units

According to the Anderson Promoter Collection [1], Device 1 should be stronger among the tested promoters in this experiment. Moreover, we can numerically measure a relative promoter strength and compare with the values available in the Anderson Collection. To measure then, we have to evaluate the following equation while our colonies are in the log-phase of their growth: \[ RPU_{\phi} = \frac{ \left\langle \frac{dF_{\phi}}{dt}\frac{1}{OD_{\phi}} \right\rangle }{ \left\langle \frac{dF_{J101}}{dt}\frac{1}{OD_{J101}} \right\rangle } \]
Notice that this measure is evaluated with respect to Device 1, thus we will be able to measure the promoter strength of devices 2 (J23106) and 3 (J23117). In compliance with the Anderson Collection, we set to 0.7 the promoter strength of the device 1 (J23101). All our colonies were in the log-phase (colocar figura!). See below our results, the expected values and the relative error.

This shows that our Device 2 was on point with very little deviation, and Device 3 is within the error margin. We believe our measurement could be improved by ######

We have also measured the promoter strengths of all controls to compare with the devices being tested, as shown below. We have normalized everything to Relative Promoter Units (RPU) using Device 1 as reference. Notice that it is not a surprise that Device 1 would induce activity visibly higher than Device 3, being almost 15 (fifteen) times stronger.

Figure 11 - Promoter strengths measured in Relative Promoter Units (RPU) for all controls and devices. Device 1 strength was used as reference, set to 0.7 RPU. As expected, our negative controls (-control 1 and 2 in the figure) have present almost no strength at all. Devices 2 and 3 show strengths pretty close to what were expected from publicly RPU measurements from Anderson Collection. Finally, our positive control (+Control in the figure) has intermediary strength, which is reasonable since it is a mutation from another promter from Anderson Collection which should be stronger than Device 3, but weaker than Device 2.

The positive control shows similar activity as the device 2. The positive control I20270 uses the J23151 promoter, a mutant from one of the promoters in the Anderson Collection, namely, J23114. It would be natural to expect, then, to behave similarly, although nothing is really certain. Using the Anderson Collection catalogue available on Registry [1], the relative strength of J23114 part is 0.10, which is stronger than Device 3, but weaker than Device 2. We show in the above figure that this is also the case for J23151. Moreover, we have measured J23151 promoter strength as 0.2 ± 0.1 RPU, which is almost twice the strength of J23114.

Measuring Promoter Strength using a Camera + Gimp

aaaaa

Team:Brasil-USP/interlabstudy/results