Antibiotics represent the most effective treatment against bacterial infections. Since the discovery of penicillin by Alexander Fleming in 1928, many new antibiotics have been constantly developed and were successfully applied to treat life-threatening diseases. This significant advancement in medicine saved millions of lives and still does today. However, fighting microorganisms has never been a completed task, but rather an ongoing race between drug discovery and pathogens developing resistances. Thus, multi-drug resistant bacteria still constitute a major threat for humanity, as infectious diseases represent the second leading cause of death worldwide. [1][2]

One major reason, for the steady increase of antimicrobial resistances is the “inappropriate use of antimicrobials”. Due to excessive prescription and application in livestock farming, little amounts of antibiotics are nearly found everywhere, even in drinking water. These low-dose and non-lethal concentrations containing habitats, allow bacteria to adjust and develop resistances.[2][3]

As β-lactams make up a large percentage of all antibiotics used, the project preferentially focused on this class of broad-spectrum antibiotics. Carbapenems, penicillin derivatives, cephalosporins and monobactams represent the four main classes of the beta-lactams that sum up to over 100 different active substances. All compounds of this particular group can be easily identified by their common chemical structure: the β-lactam ring (see Figure 1).[4]

To address the increasing development of multi-drug resistant bacteria our iGEM Team aims at developing a novel beta-lactam biosensor in Bacillus subtilis based on the genetics of the bla-operon found in Staphylococcus aureus (for a detailed description consider our Design section below). The genetically engineered biosensor will help to (I) reliably detect even minimal antibiotic concentrations of compounds from the beta-lactam family in waste and drinking water and (II) unravel producer strains of yet unknown β-lactam related antibiotics. After extensive characterization of the detection range and sensitivity, the greater goal is to combine the functional β-lactam biosensor with our Peptidosomes. Thereby, we would prove the applicability of EncaBcillus as a completely new cultivation platform. Encapsulation of this whole-cell biosensor, will allow an easier and safer handling of the bacteria and thus making them more appealing for field applications, like for example in sewage treatment plants. As a proof of principle we used six β-lactams and two controls (water and bacitracin) to evaluate our biosensor (Table 1).

Finally, we could demonstrate a fully functional biosensor encapsulated in Peptidosomes. For this we used the strain showing the strongest response, broadest detection range and the highest viability when tested with different β-lactams. Peptidosomes containing the biosensor, were exposed to different β-lactams and development of the luminescence signal was obtained every hour over a time period of 5 hours.

In pre-tests, we investigated the growth of B. subtilis wild type and the PenP mutant, upon exposure to different concentrations of each tested β-lactam (data not shown). After that, we narrowed these down to two concentrations per antibiotic (see Table 3). Since we did not want to kill our biosensors, we focused on β-lactam concentrations which result in a slight inhibition of growth. These concentrations were used for all following experiments to further characterise our biosensors and the effect of the B. subtilis native β-lactamase (PenP) (see Figure 4). Plate reader experiments were performed in Mueller Hinton (MH) media, induction with the antibiotics was carried out after one hour of incubation at 37˚C in the plate reader. Growth was monitored every five minutes for at least 18h.

As expected, the data from the pretest performed in triplicates in Figure 4 show, that the presence of the β-lactamase PenP plays a key role in facilitating survival at higher antibiotic concentrations. The wild type strain B. subtilis W168 is therefore able to grow at higher antibiotic concentrations as the mutant W168 penP::kan^R when treated with ampicillin, carbenicillin, cefoperazone, cefoxitin and penicillin G (see Figure 4). However, this was not the case for cefalexin and bacitracin. Here, we could not observe any difference in growth inhibition between the wild type and the penP mutant in regard to the tested concentrations (see Figure 4). Cefalexin showed a very strong inhibitory effect on the growth of both wildtype and the mutant. For this reason, we chose a relatively weak final concentration (see Table 4 below). Furthermore, we noticed a growth inhibition of the mutant W168 penP::kan^R during the stationary phase, especially when treated with carbenicillin, cefalexin and penicillin G (data not shown). For this reason, we selected weaker antibiotic concentrations for the upcoming experiments with this mutant.

From these first experiments, we selected the final antibiotic concentrations for the upcoming plate reader experiments with the biosensor strains (see Table 4)

In our first experiment, we performed plate reader assays in a 96 well plate format and measured growth (OD₆₀₀) and luminescence output for 18 hours every 5 minutes. Induction with the β-lactam antibiotics occurred after 1 hour. All strains have been tested in triplicates under the same conditions. Strains with the genotype penP::kan^R have been induced with lower concentrations compared to the wild type strain W168 (see Table 4 above).

After induction, we anticipate a strong increase in luminescence signal for strains containing the full set of constructs (P_blaZ_lux or P_blaR1I_lux, P_veg_blaR1 or P_xyl_blaR1, P_lepA_blaI), thus representing functional biosensors. Besides the biosensor constructs, we also tested all physiological controls missing one essential composite of the biosensor`s heterologous one-component system (data not shown). The control strain W168 (wild type) and control 1, will presumably not show any luminescence output, while the positive control 2 is expected to show a steady luminescence signal regardless of the presence of any antibiotic compound.

As the β-lactamase PenP confers resistance to β-lactam antibiotics in B. subtilis, a rising luminescence signal is estimated for the P_penP_lux constructs post induction. Generally, the P_xyl promoter is activated by adding xylose as an inducer to the medium. In the plate reader experiments, 0.2% xylose was added to the cells (beginning with the day culture) to activate expression of blaR1.

Further we propose biosensor strains carrying the genotype remark penP::kan^R to give a stronger signal in presence of β-lactam compounds, as they cannot be degraded by the PenP enzyme.

As shown in Figure 5, the wildtype W168 (black with white dots) shows no increase in RLU/OD₆₀₀ values when induced with the different β-lactam antibiotics and controls. Control 1 (black tight stripes) behaves similarly to the wild type strain. The slight decrease of control 2 (light grey) in the bar chart where induction with ampicillin and carbenicillin happened, is mostly explained by the high growth inhibition caused by the chosen concentrations for W168 (with functional PenP). Most of the times, the constitutive expression of the lux operon resulted in an RLU/OD₆₀₀ of over 1.3 million for control 2 (see Figure 5).

Biosensor 1 gives an overall good signal for all β-lactam antibiotics tested, but also shows a higher basal activity in absence of the β-lactam compounds of 40.000- 90.000 RLU/OD₆₀₀ (see Figure 5, bar chart with bacitracin and dH₂O). Further, we could observe a difference in signal intensity dependent on the β-lactam antibiotic tested. Therefore, biosensor 1 gives the highest signal in presence of penicillin G, cefoxitin and cefoperazone with up to 2.7 million RLU/OD₆₀₀. Ampicillin and penicillin G again show a weaker increase in signal produced by biosensor 1, which could be due to the same reason as for control 2 (see Figure 5).

For biosensor 2, the detection range and sensitivity is comparable to biosensor 1, This strain strongly senses cefoxitin, ampicillin and cefoperazone reaching up to 2.4 million RLU/OD₆₀₀. Even the basal activity of the P_blaZ promoter in biosensor 2, as shown in the bar charts with bacitracin and dH₂O, conforms with the one from biosensor 1.

The activity of biosensor 3 can be induced by adding xylose (0.2% final concentration) to the sample. In absence of xylose, the signal is very weak as can be seen in the bar with the dark blue stripes (see Figure 5). This is due to the fact, that without xylose, nearly no receptor molecules are localized in the bacteria`s inner cell membrane and the signal cannot be detected and amplified. For induction with cefoperazone though, the weak receptor density on the surface because of the leaky expression of blaR1, seems to be enough to cause a quite powerful signal already. From this can be concluded, that the receptor seems to have a high affinity to cefoperazone. In the bar charts of bacitracin and dH₂O (see Figure 5), biosensor 3 shows the lowest basal promoter activity compared to the other two biosensors tested. Here, we could also observe a high sensitivity for penicillin G, cefoxitin, cefoperazone and ampicillin. In general, all biosensors show good performance in liquid MH-Medium. They able to detect the six β-lactams reliably by giving a signal way stronger than the basal luminescence.

Furthermore, the analysis of the induction of P_penP by different β-lactam antibiotics unfolded that this promoter seems to be constitutively active during exponential phase (see Figure 6). As the exact promoter length and potential regulatory regions upstream are still unidentified, two versions (short and long) of the promoter have been examined.

The RLU/OD₆₀₀ values shown in Figure 6 indicate a moderate promoter activity during exponential growth. There is no noticeable difference between the strains with a functional PenP enzyme (a) and the penP mutant (b) (see Figure 6). We could not observe a particular activation by β-lactam antibiotics, which suggests that this enzyme is produced might have other functionalities, too.

We expected the control substances (water and bacitracin) to not cause any luminescence signal at the edge of the inhibition zones. The β-lactam antibiotics should lead to a glowing halo when tested with the three different biosensor versions. The wildtype strain and control 1 should not show any signal, since both strains are lacking the lux operon. In the case of control 2 a luminescence signal should be spread over the whole plate, due to the constitutive expression (P_veg) of luciferase. Figure 7 sums up the results of the disk diffusion assay for all strains tested. After 24 hours of incubation at 37°C, plates were photographed under day light conditions and under a chemiluminescence dock (with two minutes exposure time).

As expected, the wildtype and control 1, show no luminescence signal, while control 2 leads to a strong luminescence signal spread across the entire plate (Figure 7, top). Neither bacitracin, nor dH₂O lead to an detectable output, accounting for all strains tested. While in liquid medium biosensor 1 behaves similar compared to biosensor 2, there is a tremendous difference in the detection capability. Biosensor 2 showed detection for all β-lactams tested. In contrast, biosensor 1 only showed a luminescence signal for cefoperazone, cefoxitin and cefalexin. Further, the luminescence halo around the cefoxitin disk is quite broad compared to the others, indicating an increased diffusion of the compound into the lawn. Although, biosensor 1 was activated by penicillin G in liquid medium, we could not observe an induction on plate.

Biosensor 2 was activated by all of the β-lactam compounds tested. Ampicillin, cefoxitin, cefalexin and cefoperazone strongly activate the system, while penicillin G and carbenicillin just show a weak induction of the signal on plate. These findings go along with the results obtained in liquid medium in the previous experiments. On the plate with the lawn of Biosensor 3, all β-lactams could be detected efficiently when 0.2% xylose was added. In contrast to Biosensor 1 and 2, there is a very weak luminescence halo around the cefalexin disk. Also, this halo seems not to de directly at the edge where the cells are in contact with the antibiotic, but rather a bit further off the inhibition zone. Without induction of Biosensor 3 with 0.2% xylose, we could not detect any luminescence signal, demonstrating that the receptor (BlaR1) is crucial for detection and signal transduction, standing in line with results obtained in liquid medium (data not shown).

After evaluating our biosensor versions, we were also curious to follow up on the native B. subtilis β-lactamase PenP. Thus, we performed disk diffusion assays with the P_PenP (short and long version) reporter strains and checked if any of the β-lactams would lead to a luminescence signal (Figure 8). Unfortunately, none of the tested substance lead to a notable luminescence. We again could only observe a weak basal promoter activity (as in liquid) with both reporter strains. Taking these results together with the observations in liquid (Figure 6), we can state that the native β-lactamase in B. subtilis dose not respond to any of our tested β-lactams. Yet, we could observe increased sensibility of our biosensors, when PenP, is knock-out, accounting for ampicillin in liquid conditions (Figure 5) and for all antibiotics tested on solid agar plates (Figure 7). We could also clearly demonstrate increased sensitivity in terms of resistance against β-lactams, when B. subtilis is lacking PenP (Figure 4).

As a final characterisation of our biosensor, we carried out Dose Response assays in order to investigate the detection sensitivity. In this experiment, eleven different concentrations of the β-lactam antibiotics and bacitracin (control) have been tested together with the biosensor 2. This biosensor was chosen due to its reliable performance in liquid and on solid medium and because all necessary components are constitutively expressed without the addition of any further substances (as with biosensor 3).

The obtained results indicate that biosensor 2 shows the highest dynamic range for cefoperazone and is capable to even sense very low concentrations of this compound. Biosensor 2 shows the weakest dose response towards cefalexin, which is in agreement with all previous experiments. The antibiotics cefalexin, ampicillin and penicillin G lead to a decrease in luminescence signal for concentrations higher than 10^-1, due to the growth inhibition at these concentrations. Over all, a high dynamic range upon antibiotic exposure can be detected for all compounds starting at concentrations above 10^-3.