Klebsiella oxytoca strains and bacterial constructs used in this study are listed in Table 1. K. oxytoca MIT 09-7231 was used as the prototypic toxigenic strain and for construction of the isogenic Δcrp mutant (Darby et al., 2014). To determine gene expression of aroX and NRPS operons, different liquid bacteriological media were used: lysogeny broth (LB), tryptone soy broth (TSB), tryptone lactose broth (TLB), and Dulbecco’s modified Eagle’s medium (DMEM) with high glucose (4.5g/l). The transcription of aroX and npsA genes was analyzed under different environmental conditions in TLB medium at 37°C with shaking, and samples were harvested when an OD_600nm of 1.6 was reached for RNA extraction. TLB medium was prepared as previously described [17g/l tryptone, 10g/l lactose, and 2.5g/l dipotassium hydrogen phosphate (Tse et al., 2017)]. TLB medium was supplemented with 0.2% glucose, 0.2% glycerol, 0.3M NaCl, 1.0mM EDTA, 5.0mM CaCl₂, or 5.0mM MgCl₂. In order to examine the effect of lactose on the transcription of aroX and npsA genes, tryptone broth (TB) was used and gene expression results were compared with those obtained from cultures in TLB medium. Antibiotics [200μg/ml (ampicillin), 50μg/ml (kanamycin), or 10μg/ml (tetracycline)] were added to culture media when necessary.

K. oxytoca MIT 09-7231 was targeted for mutagenesis of crp and npsA genes by using the lambda-Red recombinase system (Datsenko and Wanner, 2000). Briefly, PCR fragments containing crp or npsA sequences flanking a kanamycin resistance gene were obtained by using gene-specific primer pairs (Table 2). Each purified PCR product was electroporated independently into competent K. oxytoca carrying the lambda-Red recombinase helper plasmid pKD119, whose expression was induced by the addition of L-(+)-arabinose (Sigma) at a final concentration of 1%. The mutations were confirmed by PCR and sequencing. Subsequently, the FRT-flanked kanamycin cassettes were excised from both Δcrp and ΔnpsA mutant strains after transformation with pCP20 plasmid, as previously described (Datsenko and Wanner, 2000).

A fragment with targeted mutations in the putative CRP-Box of intergenic regulatory region of the divergent aroX and npsA genes was generated using overlapping PCR (Ho et al., 1989) with specific primers (Table 2). Two fragments were generated separately in a first round of PCR: one with the 5΄ half and other with the 3΄ half of the intergenic regulatory region of aroX and npsA genes including the overlapping mutated region. Subsequently, the two fragments were mixed and amplified in a second round of PCR. DNA sequencing was carried out to verify the introduction of the point mutations.

The pTrc99K-CRP plasmid was constructed for complementation experiments by subcloning the crp gene from the pTrc99Acrp plasmid (Kurabayashi et al., 2017). The pTrc99Acrp vector was digested with NcoI and BamHI, and the fragment corresponding to the crp gene was then purified and ligated into pTrc99K previously digested with the same restriction enzymes. The identity of the insert was confirmed by DNA sequencing.

Total RNA was extracted from bacteria grown under different culture conditions using the hot phenol method (Jahn et al., 2008; Ares, 2012), with some modifications. Briefly, after the lysate was obtained, an equal volume of phenol-saturated water was added, mixed, and incubated at 65°C for 5min. The samples were chilled on ice and centrifuged at 19,000 × g for 10min at 4°C. The aqueous layer was transferred to a microtube, RNA was precipitated with cold ethanol, and it was incubated at −70°C overnight.

The RNA was pelleted by centrifugation at 19,000 x g for 10min at 4°C. Pellets were washed with cold 70% ethanol and centrifuged at 19,000 × g for 5min at 4°C. After careful removal of the ethanol, the pellets were air-dried for 15min in a Centrifugal Vacuum Concentrator 5,301 (Eppendorf). The pellets were resuspended in DEPC-treated water. Purification of RNA was performed using the TURBO DNA-free kit (Ambion, Inc.). Quality of RNA was assessed using the NanoDrop ONE (Thermo Scientific) and with a bleach denaturing 1.5% agarose gel, as previously described (Aranda et al., 2012). cDNA was synthesized using 1 μg of RNA, 5pmol/μl of random hexamer primers, and 200U/μl of RevertAid M-MulV-RT (Reverse transcriptase of Moloney Murine Leukemia Virus; Thermo Scientific).

Quantitative real-time PCR was performed in a LightCycler 480 instrument (Roche) to quantify the gene expression levels. Specific primers (Table 2) were designed using the Primer3Plus software¹ (Untergasser et al., 2007). For LightCycler reactions, a master mix of the following components was prepared: 2.0 μl of PCR-grade water, 0.5 μl (10 μM) of forward primer, 0.5 μl (10 μM) of reverse primer, 5 μl of 2x Master Mix, and 2.5 μl of cDNA (~50ng). A multiwell plate containing all samples was loaded into the LightCycler 480 instrument. Amplification was performed in triplicate wells for each sample analyzed from three independent experiments. In each set of reactions, 16S rRNA (rrsH) was used as a reference gene for normalization of the cDNA amount. Real-time PCR analysis was performed using the following optimized assay conditions: (1) denaturation program (95°C for 10min); amplification and quantification programs were repeated for 45cycles (95°C for 10s, 59°C for 10s, 72°C for 10s with a single fluorescence measurement), (2) melting curve program (95°C for 10s, 65°C for 1min with continuous fluorescence measurement at 97°C), and (3) a cooling step at 40°C for 10s. The absence of contaminating DNA was tested by the lack of amplification products after 45 qPCR cycles using RNA as template. Control reactions with no RNA template and with no reverse transcriptase enzyme were run in all experiments. The relative gene expression was calculated using the 2^-ΔΔCt method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008). To ensure that 16S rRNA (rrsH) was an optimal reference gene for normalization of qPCR, absolute quantification was performed by obtaining a standard curve according to 10-fold dilutions of K. oxytoca 09–7231 chromosomal DNA (10³, 10⁴, 10⁵, 10⁶, and 10⁷ theoretical copies). Ct values were interpolated to standard curve to obtain gene expression (gene copies per μg RNA). Expression of 16S rRNA (rrsH) gene remained unaffected in all conditions tested (Supplementary Figure S1).

The His₆-CRP expression plasmid pQE80crp (Kurabayashi et al., 2017) was electroporated into competent E. coli BL21 (DE3). Bacteria containing recombinant plasmid were grown at 37°C to an OD_600nm of 0.5 in LB; 1.0mM IPTG was then added and cultured for 3h. Cells were then pelleted by centrifugation and resuspended in urea buffer (8M urea, 100mM Na₂HPO₄, and 10mM Tris-HCl, pH 8.0) and lysed by sonication. The lysate was centrifuged, and the supernatant was filtered through Ni-NTA agarose column (Qiagen) pre-equilibrated with urea buffer. After washing with buffer containing 50mM imidazole (200ml), the protein was eluted with 500mM imidazole (10ml). Fractions were analyzed by SDS-PAGE and Coomassie blue staining. Protein concentration was determined by the Bradford procedure. Aliquots of the purified protein were stored at −70°C.

To evaluate CRP binding to the promotor sequence, a 448-bp DNA probe containing the intergenic regulatory region of the divergent aroX and npsA genes was used. In addition, the probe ΔCRP-Box containing the mutation in the putative CRP-binding site on the intergenic regulatory region of the divergent aroX and npsA genes was employed. DNA probes from the regulatory region of the enterotoxigenic E. coli estA2 and eltA genes were used as positive and negative controls (Haycocks et al., 2015). The binding reaction was performed with 100ng of DNA probes and increasing concentrations of purified His₆-CRP with or without 200μM cAMP (Hirakawa et al., 2020) in a 20μl reaction mixture containing H/S 10X gel-shift binding buffer (400mM HEPES, 80mM MgCl₂, 500mM KCl, 10mM DTT, 0.5% NP40, and 1mg/ml BSA, De la Cruz et al., 2007). Samples were incubated for 20min at room temperature and then separated by electrophoresis in 6% non-denaturing polyacrylamide gels using 0.5X Tris-borate-EDTA buffer at 4°C. The DNA bands were stained with ethidium bromide and visualized under UV light.

HeLa cell line (ATCC CCL-2) was used to determine cytotoxic activity as previously described (Darby et al., 2014). Approximately 1×10⁶ cells were suspended in 900μl of DMEM high glucose (4.5g/l; Invitrogen) with 10% FBS (Gibco) and seeded into 24-well cell culture plates (Costar). To investigate cytotoxin production in the K. oxytoca strains (wild type, Δcrp, Δcrp pTrc99K-CRP, and ΔnpsA), 100μl of bacterial supernatants was filtered through a PVDF 0.22-μm sterile Millex-GV filter (Merck Millipore) and added to the wells containing HeLa cells. After 48h of incubation at 37°C under a 5% CO₂ atmosphere, the cells were visualized using a Nikon TE300 inverted microscope at 10X magnification. Cytotoxin production was defined as >50% cell rounding and detachment and<50% confluency, as compared to the negative control samples [(TLB medium only or supernatant of the non-toxigenic K. oxytoca ΔnpsA strain (Schneditz et al., 2014; Dornisch et al., 2017; Tse et al., 2017)]. Negative control samples had a monolayer with minimal cell rounding or detachment and>80% confluency.

The LDH Cytotoxicity Assay Kit (Invitrogen) was used according to the manufacturer’s instructions to measure lactate dehydrogenase (LDH) released from HeLa cells after damage of plasma membrane integrity. 1×10⁴ HeLa cells were cultivated in 100μl DMEM high glucose (4.5g/l; Invitrogen) with 10% FBS (Gibco), seeded into 96-well flat-bottom culture plates (Costar), and incubated at 37°C under a 5% CO₂ atmosphere for 48h. Subsequently, the medium was removed, and the cells washed with PBS. Then, 10μl of negative control (PBS), culture medium control (TLB), positive control (lysis buffer), and filtered bacterial supernatants (wild type, Δcrp, Δcrp pTrc99K-CRP, and ΔnpsA) as described before was added to DMEM medium without FBS for 48h. After treatment, an aliquot of 50μl each sample medium was transferred to a new 96-well plate, and kit solutions were added into each well. The absorbance was measured at 490nm and 680nm with a spectrophotometer (Multiskan Ascent, Thermo Scientific). All samples were tested in triplicate on three independent biological replicates, and the mean results were expressed as LDH cytotoxicity by subtracting the 680nm absorbance background value from the 490nm absorbance value.

Statistical analysis was performed using Prism 7.0 (GraphPad Software, Inc., San Diego, CA, United States). Data represent the mean±standard deviation (SD). The mean differences were determined using one-way ANOVA followed by Tukey’s comparison test. Values of p<0.05 were considered statistically significant.

To determine the optimal conditions of expression of the aroX and NRPS operons of K. oxytoca MIT 09-7231, the bacteria were cultivated in different culture media, such as LB, TSB, TLB, and DMEM transcription analyzed by RT-qPCR. The conventional LB medium was used as a reference to determine the basic levels of expression of genes encoded by the aroX and NRPS operons. The lowest levels of transcription of the aroX and NRPS operons occurred when the bacteria were cultivated in DMEM; this was rather surprising since expression of most virulence factors of pathogenic E. coli strains occurs upon growth in DMEM (Leverton and Kaper, 2005; Platenkamp and Mellies, 2018). In comparison with the growth of MIT 09-7231 in LB, the expression of aroX and NRPS operons of MIT 09-7231 cultivated in TSB and TLB media enhanced ~5- and~150-fold, respectively (Figure 1A). Unlike TSB medium, TLB contains lactose instead of soy. The expression levels of TV genes were very similar in the different culture media, supporting the notion that they are genetically organized in operons. Since aroX and npsA are the first genes of the aroX and NRPS operons, respectively, only the expression of these two genes was evaluated throughout the study. When K. oxytoca MIT 09-7231 was grown in TLB at 37°C for 12h, the highest levels of expression of aroX and npsA genes occurred at 9h and were maintained for 12h, which corresponds to stationary growth phase (Figure 1B). Our data indicate that growth in TLB medium, which contains lactose, favors the expression of K. oxytoca aroX and NRPS operons during stationary growth phase.

The influence of nutritional factors in the expression of genes involved in TV biosynthesis was quantified by RT-qPCR when K. oxytoca was grown in TLB supplemented with 0.2% glucose, 0.2% glycerol, 0.3M sodium chloride, and 1.0mM EDTA. The expression of aroX and npsA genes was repressed (~4-fold) and activated (~3-fold) by glucose and glycerol, respectively (Figure 1C). In high osmolarity (0.3M NaCl), the transcription of aroX and npsA decreased ~8-fold (Figure 1C). Transcription of these genes required divalent cations because the addition of EDTA dramatically diminished their expression (~50-fold), and this effect was partially and fully reverted by the addition of CaCl₂ and MgCl₂, respectively (Figure 1C). Of note, depletion of divalent cations by the addition of EDTA to the culture medium affected negatively bacterial growth (Figure 1D). This effect was reversed by the addition of CaCl₂ and MgCl₂ to TLB containing EDTA (Figure 1D). The data indicate that glucose, high osmolarity, and depletion of divalent cations from the growth medium repress aroX and npsA genes.

As described above, carbon sources such as lactose, glucose, and glycerol affect the transcription of aroX and npsA genes. CRP is a global regulator that senses the fluctuations of carbon sources and controls the transcription of some enzymes involved in metabolite biosynthesis (Bai et al., 2011; Gao et al., 2012; Soberón-Chávez et al., 2017; Jeong et al., 2021). Hence, we sought to investigate the role of CRP in the regulation of aroX and npsA genes. Growth rates and expression of these genes in the wild type and its derivative ∆crp isogenic mutant were compared after growth in TLB at stationary phase (OD_600nm=1.6) by RT-qPCR. In the absence of CRP, growth was significantly attenuated and expression levels of both aroX and npsA genes were diminished ~10-fold. Both, growth rate and expression were reversed by the complementation of this mutant with the pTrc99K-CRP plasmid (Figures 2A,B). These results indicate that CRP positively regulates expression of aroX and npsA genes.

We wanted to know whether the regulation exerted by lactose on TV genes implicated CRP. Thus, we quantified the transcription of aroX and npsA genes in the wild-type and ∆crp mutant strains growing in TLB (medium with lactose) and TB (tryptone broth), which is TLB without lactose. Growth of the wild-type strain in the absence of the lactose decreased TV gene transcription by ~5-fold (Figure 2C). This effect was CRP-dependent because transcription of aroX and npsA was not altered in the ∆crp mutant strain (Figure 2C). As an expression control of a lactose-regulated gene, we quantified the transcription of crp in the wild-type growing in the presence of lactose and found ~2-fold increase in expression. We hypothesized that lactose is involved in the CRP-dependent aroX and npsA transcription; thus, we analyzed TV-mediated cytotoxicity on HeLa epithelial cells using the supernatants of the wild-type and ∆crp mutant cultures grown in the presence and absence of lactose (Figure 2D). The supernatant recovered of the wild-type strain grown in TLB (with lactose) presented a higher cytotoxic effect on HeLa cells than the supernatant obtained from the wild-type strain grown in TB (without lactose). The supernatants from the ∆crp mutant grown with/without lactose caused a slight cytotoxic effect on HeLa cells (Figure 2D). These data support our hypothesis that lactose induces CRP-mediated aroX and npsA genes transcription and consequently triggers TV-mediated cytotoxicity on HeLa epithelial cells.

Sequence analysis of the intergenic region of aroX and npsA genes identified a putative CRP-binding site (Figure 3A). This sequence (CGTGA-N₆-TCTAA) shared 7 of 10bp with the E. coli consensus sequence (TGTGA-N₆-TCACA; Figure 3B). Electrophoretic mobility shift assays (EMSA) were performed using a recombinant His₆-CRP protein and DNA probes. Indeed, CRP bound to the aroX-npsA intergenic region since CRP-DNA complexes were observed using 100nM of His₆-CRP and this DNA-binding activity was dependent of the presence of cAMP (200μM; Figure 3C). To demonstrate whether the putative CRP-box was required for CRP binding to the aroX-npsA intergenic region, site-directed mutagenesis of the CRP-box was performed (Figure 3B). CRP did not bind to the aroX-npsA intergenic region containing the mutation of CRP-Box (Figure 3D). As positive and negative controls, estA2 and eltA regulatory regions were used (Figures 3E,F; Haycocks et al., 2015). These results clearly show that CRP binds directly to the aroX-npsA intergenic region through recognition of a specific binding site.

To corroborate the role of CRP on the aroX and npsA genes transcription, we determined the TV-mediated cytotoxicity on epithelial cells using the supernatants from K. oxytoca wild type, ∆crp mutant, and complemented ∆crp mutant. Cytotoxicity was severely affected in the ∆crp mutant as compared to the wild type and the complemented mutant. As expected, the supernatant of the non-toxigenic ∆npsA strain did not cause any cytotoxic effect on HeLa cells (Figure 4A). Further, the LDH release activity assay showed that wild-type strain supernatant induced death of HeLa cells (~14-fold) as compared to the PBS control. In contrast, the ∆crp mutant supernatant significantly reduced the death of HeLa cells by ~6-fold as compared to the wild-type strain supernatant. Levels of released LDH by HeLa cells treated with the supernatant of the complemented ∆crp mutant were similar to that of wild-type strain supernatant. Neither the TLB medium nor the non-toxigenic ∆npsA supernatant induced death of HeLa cells (Figure 4B). These phenotypic data support the role of CRP global regulator as a transcriptional activator of genes involved in the K. oxytoca TV biosynthesis.