Introduction

The MRI Contrast Agent bacterioferritin was integrated into our system to produce a visible change in an MRI scan once the AND-gate of the Tumor Sensor has been activated (Figure 1). By changing the signal, bacterioferritin alerts the physician that:

• the bacteria have colonized the correct location,
• there are enough bacteria to produce a full dose of the Anti-Cancer Toxin, so the risk of sub-dosing and eventually producing a drug-resistant tumor is minimized and
• the Anti-Cancer Toxin, located on the same plasmid and under the control of the same promoter as the MRI Contrast Agent, has started to accumulate.

For more details about bacterioferritin and its role in our system, go to our description of the MRI Contrast Agent.

Overview of the Experiments

Thanks to professor Markus Rudin and his team from the Animal Imaging Center at the Institute for Biomedical Engineering of ETH Zurich, we were able to perform experiments in a small-animal MRI scanner.

After the design phase, we first determined the concentration of AHL needed for full induction of the system. To do this, we measured a dose-response curve using AHL as inducer and measuring sfGFP fluorescence as an output. Second, we performed an SDS-PAGE analysis to confirm that bacterioferritin was indeed expressed along with GFP after induction with the appropriate concentration of AHL. Finally, we grew the bacteria in an iron-supplemented medium to observe the effects of bacterioferritin overexpression on the MRI signal.

Our most relevant results are presented below. For a detailed protocol describing each experiment, visit Protocols.

Phase I: Initial System Design

To test wether overexpression of bacterioferritin leads to a detectable signal change in an MRI scanner, we transformed E. coli Nissle 1917 with a plasmid containing an AHL-inducible promoter (P_Lux) that controls the expression of both a green fluorescent protein (GFP) and bacterioferritin. Since LuxR is necessary for AHL-mediated (P_Lux) activation, a plasmid consisting of constitutively expressed LuxR was also transformed into the bacteria (Figure 2).

For the plasmid expressing bacterioferritin, P_Lux promoter was based on BBa_R0062. Ribosome binding site was designed with the Salis Lab RBS Calculator to reach maximum expression. Superfolder GFP sequence was based on BBa_K515105, while the sequence for our bacterioferritin (BBa_K2500000) was based on BBa_K1438001. Silent mutations were introduced to codon-optimize for expression in E. coli Nissle 1917.

For the plasmid expressing LuxR, an Anderson promoter with a relative strength of 1 was used (BBa_J23100. Ribosome binding site was designed with the Salis Lab RBS Calculator to reach maximum expression. LuxR was based on BBa_C0062.

Phase II: Tests and Optimization

Fluorescence Measurement to Obtain the AHL Dose-Response Curve

OBJECTIVE
Determine the dose-response curve and the concentration of AHL needed for full induction of the system by measuring the fluorescence after induction with different concentrations of AHL. This is critical to begin the characterization of our system and fit the first parameters of our model.

PROCEDURE
Biological triplicates of E. coli Nissle 1917 transformed with a plasmid coding for an AHL-inducible promoter that controls the expression of bacterioferritin and GFP and a plasmid coding for LuxR (Figure 2) were transferred in a 96-well plate and induced with twelve different concentrations of AHL (from 0 to 10^-2 M). Fluorescence and absorbance were measured in a plate reader over a period of 4 hours. A detailed protocol is available in Protocols.

RESULTS
Based on measurements of fluorescence over time, a time point t = 200 minutes was chosen as representative of the plateau region (Figure 3). The relationship of fluorescence at that time point and the concentration of AHL used for induction was plotted to obtain the AHL dose-response curve (Figure 4).

CONCLUSION
Based on the data, we decided to use 1E-4 M of AHL for full induction of the system in future experiments. The measurements should be made at t = 200 minutes after induction. We could also fit with statistical significiance the expression level of luxR in our system thanks to our model: it is equal to 41 ± 7 nM.min-1

SDS-PAGE to Confirm AHL-Induced Expression of Bacterioferritin

OBJECTIVE
The concentration of AHL needed for full induction of the system was calculated based on fluorescence measurements and the assumption that bacterioferritin is co-expressed alongside GFP upon activation of P_Lux (Figure 3). To confirm that bacterioferritin is indeed co-expressed, an SDS-PAGE analysis is performed.

PROCEDURE
Protein lysates were obtained from bacteria treated with different concentrations of AHL. To determine the concentrations of proteins, needed to prepare the samples for SDS-PAGE, a Bradford protein assay was preformed prior to the SDS-PAGE analysis. A detailed protocol is available in Protocols.

RESULTS

After determination of the total protein concentrations by Bradford protein assay, samples were prepared accordingly and subjected to SDS-PAGE analysis. Bands sized approximately 18.4 kDa (which corresponds to bacterioferritin) were visible in the samples induced with AHL. There were no bands in uninduced samples or the negative control (Figure 5).

CONCLUSION
Bands corresponding to bacterioferritin were visible in the samples induced with AHL. With this, to co-expression of bacterioferritin and GFP in the test strain for MRI experiments is confirmed.

Phase III: Demonstration of the Function

Magnetic Resonance Imaging of Bacterioferritin-Expressing E. coli Nissle 1917

OBJECTIVE
To test whether overexpression of bacterioferritin leads to a detectable signal change in an MRI scanner, bacteria are grown in iron supplemented medium, induced with AHL and imaged in an MRI scanner.

PROCEDURE
Biological triplicates of E. coli Nissle 1917 transformed with bacterioferritin (Figure 2) were grown in four different experimental conditions (with and without induction and with and without iron supplementation) and imaged in a 4.7 T small animal MRI scanner. T2 maps and T2-weighted images were generated since iron bound to bacterioferritin converts from its ferrous to ferric state and becomes paramagnetic, which results in shortening of the transverse (T2) relaxation time in MRI. [1] E. coli MG1655 (DE3) (T7lacO-bfr), transformed with an IPTG-inducible promoter that controls bacterioferritin expression, was used as a positive control. Additionally, a bfr-knockout E. coli K-12 from the Keio collection was included into the experiment to see how much the endogenous bacterioferritin contributes to the signal when the bacteria are grown in the presence of iron. A detailed protocol is available in Protocols.

RESULTS
Bound to bacteriferritin, iron is in its paramagnetic ferric state which shortens the T2 relaxation time in the area, and result in a visible signal drop in the T2-weighted image. [1] Both of these effects, shortening of the T2 relaxation time (Table 1, Figure 6) and a visible change in the T2-weighted image (Figure 7) were experienced by the bacteria in this experiment that were grown in the presence of iron and induced to overexpress bacterioferritin.

Table 1. T2 relaxation times for different samples. Iron (+) indicates growth in an iron-supplemented medium, while Induction (+) indicates induction of bacterioferritin expression by AHL or IPTG. pLux-bfr M52H represents our E. coli Nissle transformed with bacterioferritin under the control of an AHL-responsive promoter, while T7lacO-bfr has bacterioferritin expression controlled by an IPTG-inducible promoter. T2 relaxation times are significantly lower for samples where bacterioferritin expression is induced and iron is supplemented to the growth medium.

Iron/Induction	pLux-bfr M52H	T7lacO-bfr	bfr-knockout
–/–	303 ± 11 ms	304 ms	343 ms
–/+	303 ± 22 ms	328 ± 3 ms	N/A
+/–	225 ± 1 ms	221 ± 6 ms	223 ± 6 ms
+/+	166 ± 10 ms	193 ± 16 ms	N/A

All the bacteria were resuspended and imaged in PBS, after washing of the culture medium. To test if any unwashed iron could mask the signal, the T2 relaxation time of pure PBS was compared to PBS supplemented with iron. The results showed that the free iron in the medium only slightly changes the signal (T2 = 332 ms in PBS and T2 = 316 ms in iron-supplemented PBS) and should not interfere with the measurements (Figure 8).

As observed with other strains (Table 1), the bfr-knockout strain experienced a change in the signal when grown in an iron-supplemented medium, suggesting that even in the absence of bacterioferritin, other iron-storage systems present in the bacteria contribute to iron binding and background noise (Table 1, Figure 8).

CONCLUSION
We believe the results demonstrate the usability of bacterioferritin as an MRI contrast agent in vitro and confirms the potential to use it as an in vivo reporter of tumor sensing. All samples where bacterioferrition overexpression was induced and iron supplementation was provided showed a significant change in the signal, as expected. There was, however, a more modest level of signal change experienced by all bacteria grown in an iron-supplemented medium, which can probably be contributed to presence of various inherent bacterial iron-storage proteins, as the knockout experiments suggest.