Results

Modelling

Clorobiocin blocks the ATP-binding site of bacterial gyrase subunit B (GyrB). The high sequence similarity between the ATP-binding site of GyrB and human topoisomerase II leads to blocking of the ATP binding site of human topoisomerase II rendering clorobiocin toxic to humans. Our aim was to improve the binding characteristics of our aminocoumarin in a way that it selectively binds GyrB but not the human topoisomerase II. A second advancement is that we use a β-lactam ring to make it specific to organisms that have a β-lactamase, like MRSA. The docking poses of our new, activated aminocoumarin (with cleaved lactam ring) with GyrB (PDB:1KZN) scored better than novobiocin and clorobiocin indicating a higher affinity to our target protein.

Figure 1: Interaction diagram between the activated aminocoumarin and the binding pocket of Gyrase B (1KZN)

Next, we compared the binding affinity of a novobiocin-resistant variant of GyrB to our newly synthesized aminocoumarin. The activated aminocoumarin is differently positioned in the crystal structure (PDB:1AJ6) than novobiocin and showed better binding indicating that it is even effective against the resistant variant.

Figure 2: Interaction diagram between novobiocin and the binding pocket of Gyrase B (1AJ6)

Figure 3: Interaction diagram between the activated aminocoumarin and the binding pocket of Gyrase B (1AJ6)

The inactivated form of our antibiotic (lactam ring closed) was also docked in a reverse orientation in the ATP binding site of the off-target human topoisomerase II (PDB:1ZXM). The unforeseeable consequences for binding potentially render it ineffective in the human cell.

Figure 4: Positioning of CBN in the binding pocket of human topoisomerase type II (1ZXM)

Figure 5: Positioning of LC in the binding pocket of the human topoisomerase type II (1ZXM)

Chemistry

Our new antibiotic should have an alternative Ring A with a β-lactam ring as a warhead. We synthesized 3-(((1-carboxypropan-2-yl)amino)methyl)-4-hydroxybenzoic acid using the duff reaction and reductive amination (LINK TO ChemIntro). The synthesis worked fine, but clean-up attempts by crystallization and extraction failed. Purification by HPLC worked fine. Extended storage in acidified methanol led to methylation of the compound.

Figure 6: Structures of the desired products and the methylated derivatives.

Figure 7: MS spectra of the LC-MS analysis. Elution time from LC from top to bottom: minute 5.0; 5.4; 6.0 and 7.6 corresponding to the non-methylated, methylated, dimethylated and trimethylated compound.

Biochemistry

In order to produce our new antibiotic, the synthesized Ring A derivative has to be accepted by one of the L-enzymes (CloL, CouL, NovL or SimL) which catalyze the formation of the amide bond between Ring A and B. The conjugation of CouL and CloL into the CloQ-defective strain Clo-SA02 was successful. SimL and NovL could not be cloned. LC-MS analysis showed that the feeding experiment with the synthesized Ring A derivative was successful.

Figure 8: On the left: Desired product of feeding, on the right: MS chromatogram and MS spectrum with desired m/z.

The next important step is to close the warhead, the β-lactam ring. First, we had to clone and express β-lactam synthetase. Cloning worked fine, but expression was not optimized due to time constraints.

Test System

aminocoumarin resistant GyrB from E.coli (GyrR_EC). Both are under control of inducible promoters (pRha and arac, respectively). We established the system using carbenicillin and clorobiocin. A suitable clorobiocin amount was found by performing a clorobiocin dilution series on different E.coli strains (XL1-blue and a TolC-knockout strain, following named TolC) by us and on Corynebacterium glutamicum by iGEM team Franconia. As expected, the XL1-blue strain is almost resistant while TolC is more susceptible to clorobiocin. C. glutamicum as a gram-positive bacterium showed by far the most significant zones of inhibition compared to the E.coli strains. The SHV-1 construct provides resistance to carbenicillin in a dose-dependent manner in XL1-blue, while the construct does not seem to work as good in TolC.

Figure 9: Different L-rhamnose concentrations (0.1, 0.2, 0.4 and 0.8%(w/v)) were used to induce SHV-1 expression. 1000 µg carbenicillin were used. pSB1C3-SHV-1 is used as a control because it does not express the protein.

The GyrBR_EC construct shows a reduced zone of inhibition in TolC. The small zone of inhibition in XL1-blue makes it hard to determine a difference after induction.

Figure 10: Figure 10: Zone of inhibition in induced TolC samples is reduced compared to the control.

Team:Tuebingen/Demonstrate

Results

Modelling

Chemistry

Biochemistry

Test System