Strains and plasmids used in this work are listed in Table 1. Bacterial cultures were grown in LD broth (10 g/L of tryptone, 5 g/L of yeast extract, and 5 g/L of NaCl) or M9tx minimal medium (0.1% NH₄Cl, 1.6% Na₂HPO₄⋅12H₂O, 0.3% KH₂PO₄, 0.5% NaCl, 0.013% MgSO₄, 0.001% CaCl₂, trace elements, and 0.05% Triton X-100). Solid media were prepared with LD10 (LD broth supplemented with 1% agar). When needed, media were supplemented with 0.4% (w/v) glucose, 0.4% (w/v) gluconate, ca. 0.2% (w/v) 2-KG, and 20 mM of succinate. 2-KG was prepared from calcium 2-KG as described (Swanson et al., 2000).

The Prestwick Chemical Library (Prestwick Chemical Inc., Illkirch, France) was a kind gift of P. Seneci and P. Landini. The 1,120 library compounds are at 2 mg/ml concentration in 100% dimethyl sulfoxide (DMSO). For the first screening of the Prestwick Chemical Library and the secondary assays, PAMO107 (i.e., PAO1 ΔgntP) and PAO1 cultures in M9tx supplemented with either 0.4% glucose or gluconate were diluted to OD₆₀₀ 0.01 in the same medium. Culture aliquots (145 μl) were distributed in the wells of 96-well plates, and 5 μl of Prestwick Chemical compounds was added (final concentration, 66.7 μg/ml). As controls, some microcultures were supplemented with 5 μl of DMSO or 33.0 μg/ml of ciprofloxacin. The plates were incubated at 37°C with slow agitation for 24 h, and the growth was monitored by measuring the OD₆₀₀ at the EnSight microplate reader (PerkinElmer, Waltham, MA, United States). R was calculated as (A–B)/A, where A and B are the OD₆₀₀ reached after incubation of the culture without (with DMSO only) and with sanguinarine, respectively.

Sanguinarine chloride (MW 367.78 g/mol; hereafter sanguinarine) stock solution was prepared at 2 mg/ml (i.e., 5.4 mM) in 100% DMSO and twofold serially diluted in DMSO to 0.0625 mg/ml. Overnight cultures of PAMO107 in LD were washed twice in M9tx with 0.4% gluconate and resuspended in 1 ml of the same medium at OD₆₀₀ 0.01 in glass test tubes. Cultures were incubated 24 h at 37°C in the presence of 33 μl of either 2 mg/ml or diluted sanguinarine samples (or the same volume of DMSO for control cultures). The OD₆₀₀ was measured at the EnSight reader (PerkinElmer). The 50 and 90% inhibitory concentration (IC₅₀ and IC₉₀) were estimated as the sanguinarine concentrations that reduced the OD₆₀₀ by 50 and 90%, respectively, compared with that of the control culture with DMSO. Three biological replicates were analyzed for each strain/condition.

PAO1 cultures grown in M9tx supplemented with 0.4% gluconate and either 66.6 μg/ml of sanguinarine chloride in DMSO or the same volume of DMSO only were fixed with formaldehyde (0.37% final concentration) for 30 min at 37°C with shaking. Cells were pelleted, washed with phosphate-buffered saline (PBS) buffer, and resuspended in 1/10 volume of PBS. Cell suspension (10 μl) was layered on microscope slides precoated with a thin layer of 1.5% agarose and observed with a Leica DMRA2 widefield microscope (Leica Microsystems, Wetzlar, Germany) using the 100 × immersion objective. Fluorescence of sanguinarine was detected in the UV channel (excitation–emission filters, 360/40–470/40).

Samples (5 ml) of M9tx medium supplemented with 0.4% gluconate and 66.6 μg/ml of sanguinarine in glass tubes were mock-inoculated or inoculated with PAO1, PAMO107, or PAMO108 at OD₆₀₀ = 0.01 and incubated overnight at 37°C with gentle agitation. The cells in the supernatants were removed by centrifugation, and the supernatants were filtered through 0.45-μm pore size filters. Supernatants (4 ml) were freeze dried. The solid residue was extracted with ca. 1 ml of MeOH [high-performance liquid chromatography (HPLC) grade] by stirring in a sealed flask for 1 h. MeOH was added by micropipette; but for reliability and precision of sanguinarine quantification, the MeOH amount was measured by weight. The suspension was then filtered on a 0.45-μm syringe filter, diluted up to 10 ml, and analyzed by HPLC in triplicate. The analyses were performed using a Kinetex^® -5μm C18 100 Å column (150 mm × 4.6 mm, i.d.) on a JASCO PU4180 equipped with a MD4015 PDA detector. A fixed 1 ml/min flow rate of the mobile phase was used. The analyses were performed with gradient elution using a mobile phase consisting of a solvent A (0.1% formic acid in deionized water) and a solvent B (0.1% formic acid in acetonitrile). The following program was employed: 20% B at 0–5 min, 21–90% B at 5–19 min, maintained 90% B for 10 min, and then 90–20% B in 3 min. The injection volume was 20 μl. Sanguinarine signal (retention time 9.55 min) was integrated at the wavelength of 254 nm. For the calibration of sanguinarine, standard solution in MeOH (HPLC grade) was initially prepared; however, the resulting calibration curve was not reliable due to a marked matrix effect of the residue of the medium extracted during sample preparation. Thus, a calibration was performed using the cell culture medium (Supplementary Figure 1). One milliliter of a stock solution of sanguinarine in DMSO (6.7 mM) was added to 18.5 ml of the medium and stirred thoroughly. The volatiles were then evaporated under vacuum, and the solid residue extracted three times with 2 ml of MeOH. The combined extractions were collected in a 10-ml measuring flask and made up to the mark. Four standard calibration solutions of sanguinarine ranging from 98.2 to 12.3 μg/ml were prepared by dilution of the mother solution and injected in triplicate.

2-KG was measured in cell-free supernatants as described (Lanning and Cohen, 1951). In brief, overnight cultures of PAO1 and PAMO108 were centrifuged, and the bacterial cells washed twice with M9tx and diluted to OD₆₀₀ = 0.01 in M9tx supplemented with 0.2% 2-KG and 0.4% gluconate. Cultures were grown to OD₆₀₀ = 0.2–0.3 at 37°C, and 1.0 OD₆₀₀ (corresponding to ca. 2 × 10⁸ cfu) samples were withdrawn and centrifuged. Bacterial pellets were washed twice in M9tx, resuspended in 1 ml of M9tx containing 15 μg/ml of 2-KG, and supplemented with either 33 μl of 2 mg/ml of sanguinarine in DMSO (final concentration 66.6 μg/ml) or an equal volume of DMSO and incubated 30 min at 37°C with shaking. The samples were centrifuged 5 min at 12,000 × g, and the cell-free supernatants were mixed with 0.5 ml of a freshly prepared 15 mg/ml solution of o-phenylenediamine in 0.25 N of HCl. The reaction mixtures were heated 30 min at 100°C in a boiling water bath and cooled to room temperature. Absorbance at 330 nm (A₃₃₀) was read in a spectrophotometer and used to determine the 2-KG concentration by comparison with standard curves generated by testing samples containing known 2-KG concentrations in M9tx and DMSO as described above. Since sanguinarine absorbs light at 330 nm (Janovská et al., 2009), the A₃₃₀ of samples without 2-KG and with sanguinarine was considered as background absorbance and subtracted from the A₃₃₀ of all sanguinarine-containing samples (see Supplementary Figure 2 for the assay description and 2-KG standard curves obtained in the presence or absence of sanguinarine).

P. aeruginosa proteins essential for growth with 2-KG as the sole carbon source, but dispensable for growth on gluconate or glucose, are the KguT transporter and the KguK and KguD cytoplasmic enzymes. In addition, a ΔgntP mutant strain growing with gluconate as the sole carbon source would require also Gad, which oxidizes gluconate to 2-KG in the periplasm, besides the three above proteins (Figure 1). Thus, the growth of a ΔgntP mutant on minimal medium with gluconate as carbon source would be prevented by molecules inhibiting one out of these four proteins. On the contrary, such molecules should not interfere with the ΔgntP mutant growth on glucose, which enters the cell due to Glt and is converted into 6-phosphogluconate by Glk and Zwf (Figure 1). Thus, through a simple two-step screening based on the differential growth of a ΔgntP strain in media containing either glucose or gluconate as carbon source, it would be possible to identify specific inhibitors of the 2-KG branch of glucose utilization.

In the primary screening, the 1,120 compounds of the Prestwick Chemical Library were tested for their ability to inhibit the growth of the PAO1 ΔgntP mutant in minimal medium with gluconate as the sole carbon source. The growth was performed in 96-well plates and was evaluated by measuring the OD₆₀₀ before (OD₆₀₀^t0) and after (OD₆₀₀^t24) 24 h of incubation at 37°C in the presence of the Prestwick compounds. One hundred fourteen compounds that did interfere with growth (i.e., with OD₆₀₀^t24 - OD₆₀₀^t0 ≤ 0.08) were found in the primary screening. Among them, 25 had already known antibacterial activity (i.e., antibiotics, antibacterial, anti-infective, and antiseptic compounds) and were not further analyzed. The 89 remaining compounds (Supplementary Table 1) were subjected to further analyses.

Among the compounds selected in the primary screening, inhibitors of different metabolic routes and processes essential to sustain growth in minimal medium should be present. In order to find specific inhibitors of gluconate oxidation and/or 2-KG import, we applied a secondary screening; and in particular, we tested (i) the growth of ΔgntP strain with gluconate (i.e., in the same conditions of the primary screening) to confirm the results of the primary screening and select compounds with the highest inhibitory activity. We considered as inhibitors the compounds determining a ≥ 80% growth reduction with respect to the growth of the control culture with DMSO only; and (ii) the growth of the ΔgntP strain with glucose and of the gntP⁺ PAO1 strain with gluconate, which in both cases does not require Gad, KguT, KguD, or KguK.

Seventeen compounds were found to inhibit the growth of the ΔgntP strain with gluconate by at least the 80% (Figure 2). The compound showing the best performance (i.e., preventing the ΔgntP strain growth with gluconate and affecting to a limited extent the growth of the ΔgntP strain with glucose or of the wt PAO1 strain with gluconate) was sanguinarine (Figure 2, arrow), which was further analyzed.

We also observed that 30 compounds reduced the growth of the ΔgntP strain with both glucose and gluconate and of the gntP⁺ PAO1 strain with gluconate by at least 50% (Supplementary Table 2), showing thus growth inhibition in minimal medium irrespective of the carbon source.

The effect of sanguinarine on the growth of PAO1 and its ΔgntP derivative was tested in aerated cultures growing in minimal medium with gluconate as the carbon source. We confirmed that also in such growth conditions (i.e., in 1 ml cultures in glass tubes instead of 96-well plates), sanguinarine inhibited the growth of the ΔgntP mutant, whereas it did not prevent the growth of PAO1, which was slightly stimulated (Figure 3A). With respect to the ΔgntP mutant growing in gluconate medium, we found that sanguinarine had IC₅₀ between 8.3 and 4.2 μg/ml (i.e., between ca. 22 and 11 μM), whereas the IC₉₀ was between 66.6 and 33.0 μg/ml (Figure 3B).

Sanguinarine may prevent growth on gluconate of the ΔgntP strain by inhibiting either factors involved in 2-KG import and utilization (namely, KguT, KguD, or KguK) or the Gad protein, which converts gluconate into 2-KG (Figure 1). In the first case (i.e., inhibition of KguT, KguD, or KguK), both PAO1 and the ΔgntP strain would not grow in the presence of 2-KG as the carbon source and sanguinarine (Raneri et al., 2018). On the other hand, if Gad were the putative sanguinarine target, both strains would grow with 2-KG as the carbon source irrespective of the presence of sanguinarine. As shown in Figure 4, we found that sanguinarine inhibits the growth of both PAO1 and its ΔgntP derivative in the presence of 2-KG as the sole carbon source. Thus, sanguinarine does not inhibit gluconate oxidation by Gad, but either 2-KG transport through KguT or 2-KG conversion into 6-phosphogluconate by KguD/KguK.

To check whether sanguinarine could enter Pseudomonas cells, we measured by HPLC its concentration in the spent medium of PAO1, ΔgntP, and ΔkguT cultures grown in M9tx supplemented with gluconate and sanguinarine. As expected, in the growth conditions of the assay (5 ml of cultures in glass tubes), the ΔgntP mutant could not grow. Instead, the other two strains formed a biofilm stuck to the tube bottom, which could not be detached by vortexing the tube (Figure 5A). We found that sanguinarine concentration in the spent medium of PAO1 and ΔkguT was halved with respect to that measured in the mock (i.e., without bacterial cells) samples. A 25% decrease was observed in the medium of ΔgntP cultures, in spite of very poor growth that reduced the number of cells potentially able to import sanguinarine in the cultures of this mutant with respect to PAO1 and ΔkguT ones (Figures 5A,B). These results suggest that sanguinarine may enter P. aeruginosa cells. In agreement with this hypothesis, PAO1 cells grown in the presence of sanguinarine are homogenously fluorescent (Supplementary Figure 3), a phenotype reasonably due to the entry in the cytoplasm of sanguinarine, which, like other alkaloids, is fluorescent (Janovská et al., 2009).

To test whether sanguinarine may inhibit the KguT-dependent 2-KG transport, preventing in this way the growth on this carbon source, we performed a 2-KG uptake assay. We measured 2-KG concentration in the growth medium with a colorimetric assay (Lanning and Cohen, 1951) before and after incubation with PAO1 cells in the presence or absence of sanguinarine. As negative control, the test was performed with the ΔkguT strain, which is unable to import 2-KG. As shown in Figure 6, the 2-KG concentration in the medium decreased upon incubation with PAO1 cells, whereas it did not change upon incubation with ΔkguT cells, as expected. The concentration of the 2-KG remaining in the medium upon incubation with PAO1 was only slightly higher (1.6 ± 0.6-fold) in the presence of sanguinarine than in its absence, suggesting that the 2-KG was transferred from the medium into PAO1 cells also in the presence of the alkaloid. This result contradicts the hypothesis that sanguinarine prevents 2-KG transport by KguT and suggests that sanguinarine may inhibit an intracellular factor specifically involved in 2-KG catabolism.

Five environmental and 15 clinical isolates, mainly derived from pulmonary infections of CF patients, were tested for growth in minimal medium with 2-KG as the unique carbon source in the presence or absence of sanguinarine. An environmental and seven clinical strains were unable to grow on 2-KG (data not shown) and were excluded from further analysis. Sanguinarine inhibited growth of all the other strains on 2-KG, but not on the other tested carbon sources, namely, succinate, glucose, and gluconate (Figure 7). In fact, the growth on succinate of almost all strains and on glucose and gluconate of three and four clinical isolates, respectively, was slightly stimulated in the presence of sanguinarine. Thus, sanguinarine inhibition of 2-KG utilization is not limited to PAO1 strain.