Heterogeneous regulation of bacterial natural product biosynthesis via a novel transcription factor.

Biological diversity arises among genetically equal subpopulations in the same environment, a phenomenon called phenotypic heterogeneity. The life cycle of the enteric bacterium Photorhabdus luminescens involves a symbiotic interaction with nematodes as well as a pathogenic association with insect larvae. P. luminescens exists in two distinct phenotypic forms designated as primary (1°) and secondary (2°). In contrast to 1° cells, 2° cells are non-pigmented due to the absence of natural compounds, especially anthraquinones (AQs). We identified a novel type of transcriptional regulator, AntJ, which activates expression of the antA-I operon responsible for AQ production. AntJ heterogeneously activates the AQ production in single P. luminescens 1° cells, and blocks AQ production in 2° cells. AntJ contains a proposed ligand-binding WYL-domain, which is widespread among bacteria. AntJ is one of the rare examples of regulators that mediate heterogeneous gene expression by altering activity rather than copy number in single cells.

Introduction

Phenotypic heterogeneity has recently developed into a major research focus in microbiology. Formerly, it was thought that genetically isogenic bacterial cell cultures coordinate gene expression collectively in a homogeneous environment. However, more and more examples arise showing that bacterial populations consist of genetically identical individual cells shaping different phenotypes, e.g. sporulation [1], bacterial competence [2], antibiotic production [3], or quorum sensing (see [4] for review). However, information about regulation mechanisms of bacterial phenotypic heterogeneity at the molecular level is rare to date.

Photorhabdus luminescens is an insect pathogenic enteric bacterium that maintains a mutualistic interaction with heterorhabditid nematodes, and can infect a variety of insect species. The bacteria colonize the gut of the infective juvenile stage (IJs) of the nematode Heterorhabditis bacteriophora. Upon entering insect larvae, the IJs release the bacteria by regurgitation directly into the insect’s hemocoel. Once inside, the bacteria rapidly replicate and quickly establish a lethal septicaemia in the host by production of a broad range of different toxins that kill the insect within two days. Bioconversion of the insect’s body by P. luminescens produces a rich food source for the bacteria as well as for the nematodes. The bacteria support nematode development and reproduction, probably by providing essential nutrients that are required for efficient nematode proliferation [5, 6]. Furthermore, the bacteria produce several antibiotics to defend the insect cadaver from other bacteria. When the insect cadaver is depleted, the IJs and bacteria re-associate and emerge from the carcass in search of a new insect host (see [7, 8] for overview).

P. luminescens is an ideal candidate to study phenotypic heterogeneity since the bacteria exist in two distinct phenotypically different forms, primary (1°) and secondary (2°) cells. During prolonged cultivation in insects or in culture medium, individual 1° cells switch to the 2° phenotype, so that in the end of the infection cycle approximately 20–50% of the cells have undergone the switching process [9, 10]. Both phenotypic variants are equally virulent towards insects. However, only 1° cells are known to associate with the nematodes and to trigger their development. Furthermore, 2° cells are unable to support nematode growth and development both in the insect cadaver as well as in culture. The reason why single cells of P. luminescens undergo phenotypic switching is not understood. It is assumed that the 2° variant is better adapted for a life in the soil, after being left in the insect cadaver while the 1° cells have emerged with the nematodes [11]. Evidently, 1° cells exhibit several phenotypic characteristics that are absent from 2° cells [9, 10]. One of the most distinct phenotypes of phenotypic heterogeneity in P. luminescens is pigmentation, which only occurs in 1° cells. Pigmentation is a result of the production of anthraquinones (AQ) (Fig. 1A). Among the natural products of P. luminescens, AQs have been extensively studied. AQs in general are common in plants, fungi and streptomycetes, but Photorhabdus is the only known Gram-negative AQ producer [12, 13, 14]. The biosynthetic gene cluster (BGC) for AQ production (antA-I) encodes a type II polyketide synthase (PKS), namely AntD (KSα), AntE (KSβ) and AntF (ACP), as well as modifying enzymes (AntG-I) [14]. For some AQs weak antimicrobial activity was detected and they are assumed to play an ecological role serving as deterrents against birds and scavenger insects [11, 15, 16]. The exact physiological role of the AQs has yet to be elucidated. The mechanism of regulation of the heterogeneous pigmentation phenotype, and therefore AQ production, was similarly unknown.

Fig. 1

AQ production of P. luminescens and schematic representation of the identification of unknown transcriptional regulators of the antA-I biosynthesis gene cluster via DNA protein pull down assay. A. Major AQs (1–4) produced by P. luminescens. B. Organization of the ant biosynthesis gene cluster of P. luminescens[14]. The promoter region of antA was amplified via PCR using 5′biotinylated forward primers yielding three overlapping, biotinylated DNA fragments (I–III). C. Immobilization of the DNA fragments I (internal control), II or III on streptavidin-coupled magnetic beads. Following incubation with cell lysate loose binding proteins were washed off by four washing steps with increasing NaCl concentrations. Finally, specific binding proteins were eluted in presence of 500 mM NaCl. D. SDS-PAGE, separated proteins of the DNA protein pull down assay. Coomassie blue stained polyacrylamide gel. 1–3: washing step 3 (175 mM NaCl) of fragments I–III, 4–6: washing step 4 (200 mM NaCl) of fragments I–III and 7–9: elution fraction (500 mM NaCl) of fragments I–III, M: protein marker. Protein bands smaller than 40 kDa and only visible in the elution fraction were cut out and identified via peptide mass fingerprinting. E. Domain structure of AntJ. The protein consists of two domains, an N-terminal HTH_11 domain, which is a putative DNA binding domain, and a C-terminal WYL domain, which is supposed to bind ligands [20].

Here we describe the identification of a regulator named AntJ that activates expression of the antA-I operon. AntJ is a novel type of regulator, harboring a putative DNA-binding domain that is essential for activation of AQ biosynthesis in P. luminescens 1° cells. The AQ promoter P is heterogeneously activated in 1° cells, whereas there is also a basal but homogeneous P activity present in 2° cells, although 2° cells do not exhibit AQ biosynthesis. Drastic overproduction of AntJ caused a homogeneous, strongly activated P activity in 1° as well as in 2° cells and led to AQ production by the usually non-pigmented 2° cells. Since both 1° and 2° cells contained comparable levels of the regulator AntJ, we suppose that AntJ directly mediates heterogeneous expression of antA-I by adjusting its activity rather than its copy number.

Results

Identification of AntJ as transcriptional activator of the AQ biosynthesis gene cluster

In order to identify a putative regulator of the AQ biosynthesis operon antA-I, we performed a DNA-protein pull down assay using the upstream region of antA. First, the 986 bp region upstream of antA (P) was subdivided into overlapping fragments (Fig. 1B, I–III), which were then immobilized on magnetic beads. Subsequently, the magnetic beads were incubated with P. luminescens 1° cell lysate and proteins binding to P were pulled out and identified by peptide mass fingerprinting (PMF) (Fig. 1C, D). Overall, we identified several proteins that specifically bound to P. The protein binding to fragment I (control) was identified as Plu2434 (CysB). In E. coli CysB is known to be a LysR-type family transcriptional regulator which positively regulates the cysteine regulon [17]. Three proteins corresponding to the bands between 40 and 25 kDa could be found in the elution fraction of fragment II as well as in the elution fraction of fragment III. Therefore, fragment III, a smaller internal fragment of II, must comprise the protein binding sites of the three proteins. The largest protein was identified as Plu2548, the intermediate protein band emerged as Plu0919 and the lowest band was determined to be Plu4185. The corresponding gene plu4185 mapped directly adjacent to antA-I and therefore was renamed as antJ.

Plu2548 was annotated as LysR-type family transcriptional regulator, and Plu0919 was identified as one of 35 PAS4-LuxR-type transcriptional regulators, whose putative function in P. luminescens has been discussed before [18, 19]. AntJ consists of an N-terminal HTH_11 domain (PF08279) and a C-terminal truncated WYL-domain (PF13280) and shows weak similarity to the DNA binding transcriptional regulator YafY (COG2378) [20] (Fig. 1E).

Due to the fact that all identified proteins are putative DNA binding transcriptional factors, their function in regulation of the AQ biosynthesis was analyzed further. Therefore, the corresponding genes were deleted separately. Neither the deletion of plu0919 nor of plu2548 caused an altered pigmentation phenotype (Fig. S1). However, the deletion of plu4185 (antJ) led to a complete loss of AQ production (Fig. 2A, c).

Fig. 2

AntJ activates the promoter of the antABCDEFGHI operon and the production of AQs. A. Comparison of the AQ production in different P. luminescens strains. HPLC-UV analysis (UV-chromatograms at 430 nm) of the ethyl acetate (EE) extracts of a: TT01−1°, b: TT01−2°, c: TT01−1°ΔantJ, d: TT01−1° + control 2 (pACYC_tacI/I), e: TT01−1° + antJ++ (pAKH05), f: TT01−2° + antJ++ (pAKH05), g: TT01−1° + control 1 (pCOLA_tacI/I), h: TT01−1°ΔantJ + antJ+ (pAKH04), B. Heatmap of changes in transcription level of antA-J of TT01−1°ΔantJ compared to TT01−1° wild type. C. Promoter activities of P in 1° and 2° cells with and without overproduction of AntJ after 8 h, 24 h and 48 h at the population level. Error bars represent standard deviation of at least three independently performed experiments. RFU, relative fluorescence units.

In order to elucidate the function of AntJ, we first examined the domain architecture. To date (October, 2016) the Pfam database counts 4519 protein sequences containing a predicted WYL-domain (PF13280) from 1659 different species. Nearly half of the proteins (2171) possess the same domain architecture as AntJ (N-terminal HTH_11 domain and C-terminal WYL-domain) including transcriptional regulators of human pathogens such as Pseudomonas aeruginosa. A phylogenetic tree, which includes the closest homologues of AntJ (based on blastp analyses), demonstrates that AntJ is only distantly related to its closest homologues (Fig. 3 and Table S5). AntJ also shows a low sequence identity to the predicted HTH_11/WYL regulators of the closely related genus Xenorhabdus (e.g. 44–43% sequence identity for orthologous HTH_11/WYL regulators from Xenorhabdus doucetiae, Xenorhabdus nematophila, Xenorhabdus szentirmaii, Xenorhabdus bovienii) (NCBI blastp). Photorhabdus temperata, another AQ producer contains an AntJ homologue with 91% sequence identity (NCBI blastp). The corresponding gene similarly maps directly downstream of the predicted AQ biosynthesis gene cluster. Photorhabdus asymbiotica does not produce AQs [21], and we could not detect any AntJ homologue in this species.

Fig. 3

AntJ is distributed among bacteria that produce secondary metabolites. Neighbour joining dendrogram of AntJ homologues. Sequences are listed in Table S5. Sequences were aligned using clustalW and the tree was generated using the Geneious tree builder incorporated into Geneious (v6.1.8) utilizing the Jukes-Cantor distance model. Scale bar represents amino acid substitutions per amino acid position.

AntJ positively regulates AQ production

In order to investigate a putative role of AntJ in the AQ biosynthesis, a marker-less deletion of the antJ gene in 1° cells was generated, yielding the strain TT01−1°ΔantJ. TT01–1° colonies as well as the liquid culture have an orange-red color as a result of AQ production. In contrast, TT01−1°ΔantJ colonies are beige to weak yellowish and the liquid culture is non-pigmented. HPLC-UV confirmed that the antJ deletion strain no longer produces any AQs (Fig. 2A, c). Complementation of ΔantJ was achieved in trans by introducing a low-copy number plasmid encoding antJ (pAKH04) (Fig. 2A, h). AQ production of both wild type 1° cells carrying the empty vector and the complemented 1°ΔantJ cells were quantitatively compared by using AQ peak areas (UV-chromatogram at 430 nm) of HPLC-UV analysis normalized to the OD600 at the time point of extraction (Fig. S2A) AQ-256 (1) was not detectable after 24 h of cultivation in the complemented strain (Fig. 2A, h) since 1 is the starting material for the production of all other AQs and is consumed when other AQs are produced. The amount of AQ-270a and b (2 and 3) reached almost 50% of wild type level, and AQ-284 (4) production corresponds to 73% of the wild type (Fig. S2A). It is important to compare the amount of 1–4 in total and not separately in order to avoid determining effects caused by small differences in growth of the cultures. Upon overexpression of antJ in 1° cells, the production of AQ-256 (1) was 46.42 ± 6.04 fold higher compared to wild type, 2 and 3 were present in comparable amounts, whereas 4 was three-fold lower compared to the wild type (Fig. 2A, e and Fig. S2A). Furthermore, a partial pigmentation due to AQ production could be restored in the usually non-pigmented 2° cells by a simple overproduction of AntJ (Fig. 2A, f and Fig. S2B). The production of 1 was 10.00 ± 0.73 fold higher than in the empty vector control of the 1° wild type, whereas 2 and 3 were clearly less produced (0.24 ± 0.01 fold and 0.21 ± 0.05 fold compared to the 1° wild type) and 4 was not detectable at all. Trace amounts of 2 were also detected for 2° cells without overexpression of antJ (Fig. S2B).

Moreover, RNA sequencing confirmed a drastic down-regulation of the transcripts antA-I, e.g. 195-fold (antG) and 72-fold (antA), upon deletion of antJ in 1° cells (Fig. 2B). Thus, we conclude that AntJ raises the AQ production by activating the transcription of the antA-I operon.

However, in the complemented antJ deletion strain the mRNA of antJ was still 18-fold down-regulated compared to the wild type. In accordance to this finding the transcription level of the antA-I genes was also found to be five- to tenfold lower compared to the wild type (Fig. 2B). Other genes, whose transcription was affected to a minor degree via AntJ, included a putative transcriptional regulator (plu0948) and a monooxygenase (plu0947). Several of the 24 up-regulated mRNAs belong to genes encoding enzymes of the primary metabolism e.g. three genes for fumarate metabolism, seven genes involved in flagella biosynthesis and two genes encoding proteins predicted to play a role in menaquinone biosynthesis of P. luminescens (Table S4).

In line with these findings, chemical analysis revealed that the amount of other known secondary metabolites [isopropylstilbene (IPS), GameXPeptides (GXP), photopyrones (PPY) and phurealipids (PL)] [13, 22, 23, 24] is not significantly affected when comparing the wild type with ΔantJ and the complemented strain ΔantJ + antJ+ (Fig. S3A). However, overexpression of antJ in the TT01−1° background leads to a reduction of photopyrones and PL A, indicating AntJ acts as some sort of repressor of the genes encoding the ketosynthase PpyS involved in PPY biosynthesis and the methyltransferase PliB responsible for the final PL biosynthesis step. The 2° cells produce less of the other known secondary metabolites compared to the 1° cells and antJ overexpression does not enhance the production of these secondary metabolites (Fig. S3B).

Next, we investigated the influence of AntJ on the P promoter activity in 1° as well as in 2° cells using a chromosomally integrated P-mCherry fusion [25]. Unexpectedly, 2° cells exhibited a P activity that was comparable to that in 1° cells, when analyzed at the population level. Presence of plasmid-encoded antJ increased P promoter activity up to twofold both in the 1° and 2° cell population compared to cells carrying the empty vector after 24 h of growth (Fig. 2C).

AntJ specifically binds within the P promoter region

We were further interested in how AntJ positively influences AQ production. As a first step, we quantified the binding of AntJ to the P promoter region using surface plasmon resonance (SPR) spectroscopy. For that reason, a double-stranded biotinylated DNA fragment comprising approximately 400 bp upstream of the antA gene was immobilized onto a sensor chip that was previously coated with streptavidin. Different concentrations of purified AntJ were then injected over the surface. The sensorgrams revealed a strong interaction of AntJ to the DNA with an overall affinity (KD) of 100 nM defined by an association rate of k = 1.7 × 105/M*s and a dissociation rate of k = 0.015/s (Fig. 4A). However, the sensorgram shapes reveal that the binding mechanism does not reflect a true 1:1 binding event since the curves do not follow a final and linear saturation. This could be due to the fact that the DNA binding mechanism of AntJ is more complex as it has recently been shown for the bacterial response regulator YpdB [26]. Another possible explanation might be that a putative ligand of AntJ is missing, which would further stabilize the binding and end in higher association and slower dissociation rates. Next, the binding region was specified to a 50 bp region within P via electrophoretic mobility shift assays (EMSA). Six different 100 bp long DNA fragments were used for the binding assays that each shared a 50 bp overlapping region. Only two of these fragments (b and c) were shifted in the presence of AntJ suggesting that the DNA binding region of AntJ is localized within the overlapping 50 bp of DNA (Fig. 4B). This DNA fragment contained two DNA motif repeats “AATGCT” (m1 and m2), which are separated by a 28 bp spacer region. This region was predicted to reflect the specific DNA binding site of AntJ. To verify this hypothesis, we used a heterologous AntJ reporter gene assay. An E. coli reporter strain was used that harbors plasmid encoded antJ and different P-luxCDABE promoter fusions. Three different truncations (s1, s2 and s3) of the P promoter were tested for P promoter activity (Fig. 4C). The s1 truncation starts directly upstream of the first “AATGCT” motif, and exhibited enhanced P activity compared to the control P fragment revealing that the promoter is intact. However, P activity was drastically decreased if the m1 motif was absent (s2) and was further reduced if the spacer region in between m1 and m2 was missing (s3). If the sequence was either modified in motif m1 (mm1) or in motif m2 (mm2) no promoter activity was detectable revealing the specificity of each the m1 and m2 motif within the P region for AntJ binding. In contrast to other DNA binding proteins, the spacer region between the m1 and m2 repeat was of functional importance since AntJ was no longer capable of activating the promoter when the sequence of the spacer region was exchanged by a non-sense sequence (mspacer). This reveals that the complete region between m1 and m2 including the spacer is specific and essential for activation of P by AntJ.

Fig. 4

Characterization of the AntJ DNA-binding site. A. Identification of AntJ binding to the antA promoter via SPR-spectroscopy. The biotinylated P fragment (398 bp) was captured onto a streptavidin-coated sensor chip. Different concentrations of purified His-tagged AntJ (10 nM: purple line; 25 nM: dark blue line; 50 nM: light blue line; 100 nM green and red line; 250 nM: yellow line; 500 nM: orange line) were passed over the chip. AntJ binds to the promoter region of antA with an overall affinity (KD) value of 100 nM. B. Determination of a 50 bp long AntJ binding region within the antA promoter region via an electrophoretic mobility shift assay (EMSA). Six different 100 bp long fragments (a-f) were chosen comprising the whole promoter region. Each fragment shares a 50 bp overlapping region with its adjacent fragment(s). Fragment b and c contain the AntJ binding region. C. Identification of a potential DNA-binding motif of AntJ via reporter gene assays. In the upper panel the potential binding motif of AntJ comprising the two redundant sequences AATGCT (M1 and M2), which are separated via a 28 bp long spacer, is depicted. The lower graph shows a heterologous reporter assay with E. coli LMG194 harboring pBAD24-antJ as well as different P-luxCDABE fusions; control: the complete antA promoter, s1-s3: various truncations of the promoter region; mm1: altered potential binding motif 1, mm2: altered potential binding motif 2, mspacer: altered 28 bp long spacer region. The expression of antJ was induced via a final concentration of 0.0001% (w/v) arabinose. The graph shows P activities in the respective reporter strains after 6 h of growth. Error bars represent SD of at least three independently performed experiments.

P promoter activity is heterogeneous at the single-cell level

Since P activity was comparable in 1° and 2° cell populations and only 1° cells produce AQs, we investigated P activity at the single-cell level. For that purpose, cells harboring a chromosomally integrated P fusion were used, which were generated as described before [25]. Then, the noise value of this reporter strain meaning heterogeneity of P activity among single cells was determined. The noise value, also referred to as coefficient of variation (CV), describes the variance of fluorescence between a set of single cells and therefore reflects the overall degree of heterogeneity, whereby a noise value of 1 would reflect a high degree in heterogeneity, and a noise value approaching 0, total homogeneity. For P. luminescens, a noise value >0.38 was found to reflect heterogeneity [25]. It could be clearly observed that P activity is heterogeneous with a noise value of 0.70 in 1° cells at the single-cell level (Fig. 5). A distinct portion of the cells exhibited a very high promoter activity whereas the other cells showed only a basal, rather weak P activity. However, we found a clear antJ gene dose effect on the heterogeneous P activity. Upon introduction of one additional copy of antJ under the control of its native promoter, the heterogeneity was diminished by 39% to a noise value of 0.43 (Fig. 5). The mean fluorescence value was 1.4-fold increased, when compared to cells harboring only a single copy of antJ (Fig. 5). 1° cells carrying a chromosomal insertion of a P gene fusion, exhibited a drastically increased antJ expression due to the strong tac-promoter and showed a high P-activity in the presence and in the absence of the inducer IPTG (Fig. 5, Fig. S4). Upon addition of IPTG, the noise value was decreased to 0.25 indicating total homogeneity, whereas the overall mean fluorescence increased 5.7 fold. As expected, no P activity could be observed in 1° cells lacking antJ (Fig. 5). However, a heterogeneous P activity (noise value of 0.48) could be restored by simple integration of antJ at another site of the chromosome. Overproduction of AntJ also mediated a homogeneously distributed P activity in the ΔantJ background (Fig. 5, Fig. S9). In addition, AQ production correlated with the detected P activity in all the reporter strains (Fig. S5 and S6).

Fig. 5

Heterogeneity of P activity at the single-cell level. Single-cell microscopy imaging of P. luminescens TT01−1°(A), TT01−1°ΔantJ (B) and TT01−2° (C) cells with chromosomally integrated P-mCherry reporter gene fusion and different levels of AntJ. The pictures at the left panel do not contain additional levels of AntJ and reflect the native conditions. An additional copy of antJ was integrated into the genome, which is either under control of the native antJ promoter (second from left panel) or the inducible tac promoter (second from right panel). A control without a promoter upstream of mCherry served as negative control (right panel). The bars in the phase contrast microscopy pictures indicate a scale of 10 μm. The tac promoter was induced via the addition of 1 mM IPTG. Additionally, pictures were taken when no IPTG was added (Fig. S5). The pictures depicted are made via overlay of the phase contrast and the fluorophore channel. The fluorescence intensities of 450 single cells were individually measured for each experiment, being analyzed using the software Big Cell Brother [28]. The noise values were calculated as SD/mean from 450 cells after 24 h of growth. Representative images and values from one of three independently performed experiments are shown. For the separate phase contrast and fluorescence pictures see Fig. S8. Box plots that show the distribution of mCherry fluorescence signals and therefore the degree of heterogeneity are presented in Fig. S9. Additionally, AQ production of these reporter strains was measured and compared (Fig. S5 and S6). AU: arbitrary units.

Interestingly, a homogeneous P activity could be detected in 2° cells (noise value of 0.29). However, the overall mean fluorescence intensity of 1° and 2° cells were comparable (Fig. 5, Fig. S11). By duplicating antJ on the chromosome in 2° cells, the heterogeneity of P activity increased 1.9 fold, reaching a noise value of 0.54 without detecting any increase in the mean fluorescence values. After integration of an additional copy of antJ under control of P into the genome of 2° cells, the promoter activity of the anthraquinone operon could be enhanced with a homogeneous pattern and AQ production (only for 2) was detectable. Contrary to the results presented above (Fig. S2B) no production of 2 could be detected for 2° cells without an additional copy of antJ (Fig. S6C).

Relative quantification of the secondary metabolites production showed that all 2° reporter strains produce less of the tested secondary metabolites than the 1° cells (Fig. S7C). Neither one additional copy of antJ under the control of its native promoter, nor under the control of P, changed the production level of secondary metabolites compared to the respective strain without an additional copy of antJ (Fig. S7).

1° and 2° cells harbor similar AntJ copy numbers

To test whether heterogeneity of P activity is mediated by a different AntJ copy number within individual cells, we generated translational antJ-mCherry fusions that are under the control of the native antJ promoter. Then, the corresponding P-antJ-mCherry elements were integrated into the chromosomes of 1°, 1°ΔantJ and 2° cells. However, no significant differences in the fluorescence of the strains could be detected at the single-cell level (Fig. 6A). Additionally, the noise values of 0.19, 0.21 and 0.21 for 1°, 1°ΔantJ and 2° cells, respectively, showed a homogeneous distribution of AntJ-mCherry. Furthermore, no autoregulation of AntJ seems to occur since the fluorescence intensity of 1°ΔantJ cells did not differ significantly from 1° or 2° wild type cells. We additionally evaluated the transcription levels of antJ by quantitative real-time PCR at three different time points of the growth curve (early exponential, late exponential and stationary phase) of 1° and 2° cultures. In relation to the housekeeping gene recA, no differences of antJ-mRNA could be detected (Fig. 6B). This suggests that heterogeneous P activity is not simply mediated by different AntJ copy numbers, but must be directly caused by different AntJ activities in single 1° as well as in all 2° cells. Therefore, AntJ acts as a transcriptional activator of the AQ operon and is, to our knowledge, one of the rare examples causing heterogeneous gene expression by altering its affinity by a yet unknown mechanism rather than its copy number.

Fig. 6

AntJ levels in P. luminescens TT01−1°, TT01−1°ΔantJ and TT01–2° cells. A. Single-cell microscopy imaging of the three strains with chromosomally integrated P-antJ-mCherry constructs. The pictures depicted are made via overlay of the phase contrast and the fluorophore channel. For the separate phase contrast and fluorescence pictures see Fig. S10. The bars in the phase contrast pictures indicate a scale of 10 μm. The fluorescence intensities of 450 single cells were individually measured for each experiment, being analyzed using the software Big Cell Brother [28]. The noise values were calculated as SD/mean from 450 cells. The box plots represent the distribution of mCherry fluorescence signals from 450 cells after 24 h of growth. Representative images and box plots from one of three independently performed experiments are shown. AU: arbitrary units. B. The levels of antJ transcripts in TT01−1° and TT01–2° cells were determined via qRT-PCR after 9 h (exponential phase), 24 h (late exponential phase) and 30 h (stationary phase) of growth. Changes in transcript levels (expressed relative to recA) were calculated using the CT method [55]. Error bars represent SD of three independently performed experiments.

Discussion

P. luminescens exists in two phenotypically different forms called primary (1°) and secondary (2°). One of the predominant phenotypes specific for 1° cells is pigmentation due to anthraquinone (AQ) production. However, it was unclear how AQ production is regulated. Here we identified a novel type of regulator that we named AntJ, which activates transcription of the antA-I operon and therefore AQ production in 1° cells. AntJ is a member of the WYL-domain transcription factors. Within the closest AntJ homologues are transcriptional regulators from different bacteria that are pathogenic to eukaryotes and produce secondary metabolites, i.e., Legionella, Yersinia and Pseudomonas species. The WYL-domain (generally described as being around 170 aa, Pfam database) in AntJ is predicted to be C-terminally truncated (only 83 aa long). In most bacteria that have close AntJ homologues (i.e. Pseudomonas, Legionella, Achromobacter, Ochrobactrum and Xenorhabdus), the WYL-domain of the predicted DNA-binding proteins is also C-terminally truncated, indicating that this could be an adaptation for the binding of a similar ligand or binding partner. Interestingly, several species of Pseudomonas, Ochrobactrum and Xenorhabdus have been isolated from (infected) insects whereas the natural reservoirs for Legionella are probably amoeba. This is to the best of our knowledge, the first time that a function of a WYL-type regulator has been found that is involved in secondary metabolite production. The distribution of HTH_11/WYL proteins emphasizes the versatility and the importance of this type of regulators. Furthermore, the fact that the AntJ homologue of P. temperata, which shows pigmentation, shares a similarity of 93.6% with AntJ of P. luminescens whereas no AntJ homologue is found in P. asymbiotica implies a highly specific function of AntJ in regulating the pigment production at least in Photorhabdus species. The WYL-domain is predicted to be a putative ligand-binding domain whose ligand is still unknown. Until now the WYL-domain has only been found in bacteria [20]. We could show that the AntJ levels in 1° as well as in 2° cells is constant, demonstrating that transcriptional activation of the antA-I operon in single 1° cells is not mediated by a simple increase of AntJ. Therefore, a specific activation of AntJ must be mandatory to bind to the P promoter and to promote heterogeneity of AQ production. Since the WYL-domain is supposed to act as a ligand-binding domain, activation of AntJ could be due to binding of a specific metabolite or protein in single 1° cells, which does not exist in 2° cells. Homogeneous activation of AQ production due to a simple overexpression of antJ in 1° as well as in 2° cells must therefore lead to a ligand-independent activation caused by the enhanced AntJ copy number in the cells. On the other hand, a putative inhibiting ligand of AntJ cannot be excluded. Whereas one additional chromosomal copy of antJ under control of its native promoter did not lead to AQ production in 2° cells, it was indeed sufficient to decrease heterogeneity of P activity in 1° cells. Only upon plasmid based overexpression of antJ or chromosomal integration of antJ with a strong inducible promoter was AQ production detectable in 2° cells. Without additional expression of antJ only trace amounts of 2 are detectable in 2° cells. It is obvious that production of AQs in 2° cells can only be achieved upon drastic overproduction of AntJ, so that the influence of a putative inhibiting ligand might be out-competed by high AntJ copy numbers. The presence or absence of a putative ligand must therefore drive activation of AntJ in single cells and mediate heterogeneity of P activation as a noise generator. A similar system has already been described for Spo0A in Bacillus subtilis. Spo0A is a master regulator for spore formation and is only active when phosphorylated. The heterogeneous activation of Spo0A is achieved by a limiting phosphate flux. Only upon a certain threshold of phosphorylated Spo0A within the cell, sporulation is induced [27]. Another example for regulation of phenotypic heterogeneity by a noise of phosphorylation is bioluminescence in Vibrio harveyi [28]. The ratios between kinase and phosphatase activities of the three sensor kinases and hence the phosphorylation state of the response regulator LuxO is not only important for the signal output, but also for the noise of the system and therefore of light production in single cells. However, it is unlikely that AntJ activity is also driven by a heterogeneous post-translational modification like Spo0A or LuxO. Moreover, several examples for heterogeneity in quorum sensing exist, where the respective LuxR-type receptor binds the auto-inducer signal(s) and therefore regulates heterogeneous gene expression (see [4] for review). Therefore, we would suppose a similar mechanism for AntJ and the putative yet unknown ligand to mediate heterogeneous AQ gene expression.

A global regulator that mediates phenotypic heterogeneity in Photorhabdus is HexA, which was identified as global transcriptional regulator and repressor of secondary metabolism [29, 30]. The influence of HexA on symbiosis was valued when exploring that 2° cells of P. temperata restore the ability to support nematode development upon inactivation of hexA. Therefore, HexA seems to be responsible for the lack of symbiosis factors in 2° cells [29]. However, there are no differences in heterogeneity of P activity in 1°ΔhexA cells detectable (our own observation), so that variable HexA copy numbers in single 1° cells as well as enhanced HexA levels in 2° cells do obviously not influence AntJ heterogeneity of AQ production.

The absence of the global regulator HdfR (Plu4688) harms symbiosis due to the incapability of P. luminescens cells to realize the transmission of the symbiont [31]. Interestingly, deletion of hdfR altered the transcription level of 124 genes including the antA-I operon, which was down-regulated in 1°ΔhdfR cells. However, we could not co-elute HdfR with the P promoter, suggesting that HdfR indirectly influences antA-I expression, possibly by regulation of the putative ligand that interacts with AntJ. Other regulators binding to the P region were the LysR-type regulator Plu2548 and the LuxR-like protein Plu0919. However, a deletion of the respective genes did not show any alterations in the pigmentation, so that these regulators are not supposed to be responsible for heterogeneous P activation.

In other AQ producing organisms, little is known about the regulation of AQ synthesis. Cultures of Morinda citrifolia plant cells are capable of using chorismate as starting material for the production of AQs. However, no specific regulator has yet been identified. It is known that in M. citrifolia AQ production is dependent on the activity of the enzyme isochorismate synthase, which leads chorismate to AQ synthesis [32]. Comparable to that a correlation between a higher concentration of the starting material malonyl-CoA within the cell and enhanced production of AQs has also been proposed in P. luminescens [11]. Similar as in P. luminescens it was shown in Aspergillus fumigatus that a global regulator of secondary metabolism, LaeA, regulates the biosynthetic genes for AQ production, but the detailed regulatory mechanism is not understood [33]. However, heterogeneous AQ production has neither been described in Aspergillus fumigatus nor in Morinda citrifolia.

It has been demonstrated before that both stochasticity, inherent in the biochemical process of gene expression, and fluctuations in other cellular components can contribute to overall variation of gene expression in single cells, that were referred to as intrinsic and extrinsic noise [34]. Since AntJ copy number is constant in 1° as well as in 2° cells it can be concluded that noise of AntJ activity is mainly caused by extrinsic noise of a specific ligand. A basic principle of phenotypic heterogeneity has been described due to noise based on molecules that exist in low copy numbers within a cell [35]. Consequently, the cellular state corresponding to the molecular composition of a cell and its gene expression phenotype varies over time and between individual cells. Specific sources of this variation can include stochastic gene expression [36] and stochastic partitioning of molecules at cell division [37]. Another molecular mechanism that can result in phenotypic heterogeneity independent of a specific ligand is the epigenetic modification of the DNA caused by differences in DNA modification like methylation. Those epigenetic modifications are an important source of phenotypic heterogeneity in eukaryotic microorganisms [38, 39]. It could be possible that 1° and 2° cells differ in their overall modification of the DNA, but however this has not been described yet. Moreover, such a phenomenon could indeed explain the absence of AQ biosynthesis in 2° cells, but not the heterogeneity of P activation in single 1° cells. The heterogeneity in AQ production of 1° cells is a typical example of division of labor [35] since production of AQ and secondary metabolites in general can require a large fraction of the celĺs genomic and metabolic resources [40] and is therefore cost-intensive. Since the AQs are secreted it is sufficient that only a part of the population invests the energetic costs for AQ biosynthesis, whereas the whole population can benefit from it.

In summary, we propose that the presence of a putative ligand of AntJ at a certain concentration promotes noise and therefore heterogeneity of AQ production in 1° cells, whereas the predominant presence or the total absence of this ligand is the reason for the lack of AQ production in 2° cells. However, the mechanism of a ligand activating AntJ, which is from our point of view more likely, is illustrated in Fig. 7. Overall, AntJ is a novel type of transcriptional activator and one of the rare examples yet that mediates heterogeneous gene expression by altering its activity rather than its copy number. The identification of the putative ligand that might bind to AntJ remains to be elusive.

Fig. 7

Model of the regulation of AQ production in P. luminescens 1° and 2° cells by AntJ. 1° cells are depicted in yellow, 2° cells in cyan. The putative activating ligand of AntJ is shown in red.

Material and methods

Strains, plasmids and oligonucleotides

All strains used in this study are listed in Table S1, all plasmids in Table S2 and oligonucleotides in Table S3.

Molecular biological methods

DNA manipulation was performed following standard procedures [41]. For isolation of genomic DNA from P. luminescens TT01 Gentra Puregene Yeast/Bact Kit (Qiagen, Hilden) or Ultra-Clean Microbial DNA Isolation Kit (Mo Bio Laboratories Inc., Carlsbad) were used. Polymerase chain reaction (PCR) was done with Phire Hot Start II DNA polymerase (Thermo Scientific, Darmstadt), OneTaq polymerase (New England Biolabs, Frankfurt) or Q5 polymerase (New England Biolabs, Frankfurt) according to the manufacturer’s instructions. Oligonucleotides were purchased from Eurofins Genomics. The MinElute PCR Purification Kit (Qiagen, Hilden) or HiYield® PCR DNA Fragment Extraction Kit (Südlabor, Gauting) was used for DNA purification from agarose gels. Plasmid isolation was performed with the Invisorb® Spin Plasmid Mini Two Kit (Stratec, Birkenfeld) or HiYield® Plasmid Mini Kit (Südlabor, Gauting). DNA restrictions were performed following the manufacturer’s protocol using endonucleases from Thermo Scientific (Darmstadt) or New England Biolabs (Frankfurt).

Bacterial strains and culture conditions

P. luminescens and E. coli were aerobically grown in LB medium [41] at 30 °C (P. luminescens) and 37 °C (E. coli). For agar plates 1.5% (w/v) agar was added to the media. If necessary, media were supplemented with chloramphenicol (20 μg/ml), kanamycin (50 μg/ml), gentamycin (15 μg/ml), ampicillin (100 μg/ml) and/or rifampicin (50 μg/ml). The plasmid-based antJ overexpression in the P. luminescens TT01−1° + antJ++ (pAKH05) and TT01−2° + antJ++ (pAKH05) was induced with 0.1 mM IPTG. Expression of chromosomal P was induced upon addition of 1 mM IPTG. In order to avoid overexpression of antJ in P. luminescens TT01−1°ΔantJ + antJ+ (pAKH04), P was not induced and antJ expression was achieved as result of the basal promoter activity. For growth of E. coli ST18 the medium was supplemented with 5-aminolevulinic acid (50 μg/ml) [42]. Pre-cultures were grown overnight and the following day appropriate volumes of the pre-cultures were used for inoculation of the main cultures to an OD600 = 0.05 (fluorescence microscopy) or OD600 = 0.1 (chemical analyses). All strains are listed in Table S2.

Generation of the plasmids

To complement antJ in the TT01−1°ΔantJ deletion strain and for overexpression of antJ in the wild type, the entire coding sequence of antJ was amplified from genomic DNA of P. luminescens using primers AKHp14 and AKHp15. Using these primers on both ends of the insert a 24 bp overhang homologous region to pCOLA_tacI/I or pACYC_tacI/I respectively, was added to enable Gibson assembly. The backbones of pCOLA_tacI/I and pACYC_tacI/I were amplified by PCR using primers AKHp12 and AKHp13. Assembled products (pAKH04 and pAKH05) were used to transform E. coli DH10B. Plasmids were isolated and transferred into strains TT01−1°ΔantJ and TT01−1° or TT01−2°, respectively.

For the purification of AntJ used for SPR spectroscopy, the plasmid pBAD24-antJ-His was generated via amplification of antJ (728 bp) using primers antJ-NheI_fwd and antJ-NdeI-ostop_rev. The insert and the vector were cut with restriction enzymes NheI and NdeI. The correct insertion was verified via PCR using primers pBAD24-seq_fwd and pBAD24-seq_rev.

For the purification of AntJ used in EMSA assays the plasmid pCATI4_antJ was constructed. For that purpose antJ was amplified with the primers AKHp44/45 and to both ends of the insert an overhang homologous to pCATI4 was attached for Gibson assembly. The vector pCATI4 was amplified by PCR using primers AKHp42/43. E. coli DH10B was transformed with the assembled vector. The vector insert was verified via sequencing.

For reporter assays in E. coli LMG194, the plasmid pBAD24-antJ was generated via amplification of antJ with the primers antJ-NheI_fwd and antJ-NdeI_rev (731 bp) and subsequent cleavage of the insert and the vector via NheI and NdeI. The different constructs of the antA promoter region were cloned into the plasmid pBBR1-mcs5-tt-lux. The complete promoter was amplified with the primers PantA-XbaI_fwd and PantA-XmaI_rev (389 bp). For generation of the antA promoter derivatives s1 (241 bp), s2 (239 bp) and s3 (203 bp), the primers PantA-s1-XbaI_fwd, PantA-s2-XbaI_fwd and PantA-s3-XbaI_fwd were used and each primer was combined with PantA-XmaI_rev for PCR. The construct mm1 (398 bp PCR) was generated via an overlap-PCR (PantA-XbaI_fwd, PantA-XmAI_rev) of the products that had been amplified with the primers pairs PantA-XbaI_fwd, PantA-mm1-ol_rev and PantA-mm1-ol_fwd and PantA-XmaI_rev. The construct mm2 (398 bp) was made via an overlap-PCR (PantA-XbaI_fwd, PantA-XmAI_rev) of the products that had been amplified with the primer pairs PantA-XbaI_fwd, PantA-mm2_rev and PantA-mm2_fwd, PantA-XmaI_rev. For construction of the mspacer (398 bp) two PCRs with the primer pairs PantA-XbaI_fwd, PantA-spacer-ol_rev and PantA-spacer-ol_fwd, PantA-XbaI_rev and a subsequent overlap-PCR as described above were performed. All of the PCR products were cut with restriction enzymes XbaI and XmaI and subsequently ligated with equally treated vector pBBR1-mcs5-tt-lux. Correct insertions were verified via control-PCR using the primers check-pBBR_fwd and check-pBBR_rev.

A PCR was performed to amplify eyfp with a size of 757 bp by using the primers eyfp-mcs_fwd and eyfp-EcoRI_rev and the plasmid pEYFP as a template. The PCR product and the vector pPINT-mCherry [25] were cut with the enzymes ApaI and EcoRI and the resulting plasmid was called pPEINT-mCherry. The control-PCR was performed using the primers check-mcherry-ins_rev and check-eyfp2_rev with a resulting PCR size of 1106 bp.

In order to generate the eyfp fusions three different constructs were inserted into pPEINT-Pless-mCherry-Pless-eyfp. P (519 bp) was generated via PCR using primers PantJ-NheI_fwd and PantJ-Not_rev and for generation of P (1220 bp) primers PantJ-NheI_fwd and antJ-ostop-NcoI_rev were used. For generation of both PCR products chromosomal DNA of P. luminescens was used as template.

P was amplified via Ptac-antJ-NheI_fwd and antJ-ostop-NcoI_rev, containing the lacI repressor gene and antJ under the control of the tac-promoter. The plasmid pAKH05 was used as template and the product size was 2110 bp.

A PCR was performed to generate P fusions for all three plasmids. P was amplified by a PCR with the primers PantA-NheI_fwd and PantA-NheI_rev and the PCR product was cut with NheI. The proper insertion within the correct direction was verified via DNA sequencing. The resulting plasmids were called pPEINT-P-mCherry-P-eyfp, pPEINT-P-mCherry-P-antJ-eyfp and pPEINT-P-mCherry-P-antJ-eyfp. A control PCR with the primers check-mcherry-ins_rev and check-eyfp _rev was performed.

For generation of the plasmids pPINT-P-mCherry and pPINT-P-mCherry the following procedures were performed. P was amplified via the primers PantJ-NheI_fwd and PantJ-BamHI_rev resulting in a 516 bp product and the 1215 bp long P product was generated with the primers PantJ-NheI_fwd and antJ-ostop-BamHI_rev. For both PCR products, chromosomal DNA of P. luminescens was used as template. In order to insert P and P upstream of mCherry, the vector pPINT-mCherry as well as the inserts were cut with NheI and BamHI. The resulting plasmids were called pPINT-P and pPINT-P. The correct insertions were verified by a PCR with the primers check-mcherry-ins_rev and check-eyfp-ins_rev. Furthermore, all generated plasmids were sequenced (Genomics core facility, LMU Biozentrum) to verify correctness.

Competent cells and transformation

E. coli cells were made chemically competent and transformed as described elsewhere [43].

Generation of knockout mutants

The gene encoding AntJ (plu4185) was deleted using pCKcipB [24], which is a derivative of pDS132 [44]. For the in frame deletion of antJ an upstream (926 bp) and a downstream fragment (788 bp) of antJ were amplified by PCR using primers AKHp06–AKHp09 introducing a PstI and a BamHI restriction site to the 5′end of the upstream fragment and the 3′end of the downstream fragment, respectively. Splice overlap extension was used to fuse the two PCR products, which were subcloned into pJET1.2, digested using PstI and BamHI restriction sites and cloned into pCKcipB via the PstI and BglII restriction sites. The resulting plasmid pAKH01 was transferred by electroporation into E. coli S17-1 λpir. Transformants were confirmed via plasmid extraction and restriction analysis before using for conjugation with P. luminescens TT01. Conjugation and plasmid excision via a second homologous recombination were performed as described before [14]. In order to confirm the 702 bp markerless deletion of antJ primers AKHp10 and AKHp11 were used. For the deletion of the genes plu0918-0925 and plu2548 a similar strategy was applied. Upstream and downstream regions were amplified using primers AKHp16-19 and AKHp22-25, respectively. In this way both upstream/downstream fragments were each extended with a 20 bp/25 bp stretch homologous to the vector and a 12–13 bp overlap homologous to the other fragment. Related up- and downstream fragments were cloned into pCKcipB, linearized with PstI and BglII using Gibson Cloning (Gibson Assembly Cloning Kit, New England Biolabs GmbH). Deletion of 7498 bp or 887 bp were confirmed by PCR using primers AKHp20-21 or AKHp26-27.

Reporter gene assays

For promoter activity assays in E. coli LMG194, cells were inoculated at an OD600 of 0.05 and aerobically grown at 37 °C in microtiter plates within a Tecan Infinite F500 system (Tecan). OD600 and luminescence were measured every 10 minutes for 8 h. Arabinose was added at a final concentration of 0.0001% (w/v) after the cells reached an OD600 of 0.2. Data are reported as relative light units (RLU) in counts per second per milliliter per OD600.

For reporter activity assays with P. luminescens, the OD600, which initially was adjusted to 0.05, the cells were aerobically grown at 30 °C in microtiter plates with a Tecan Infinite F500 system (Tecan). The OD and fluorescence intensity of mCherry (560 nm excitation, 612 nm emission, 20 nm bandwidth) were measured after 8, 24 and 28 h of incubation. Data are reported as relative fluorescence units (RFU). Raw fluorescence data were normalized with the optical density (OD600) of the respective culture.

DNA-protein pull down assay

For the DNA-protein pull down assay the upstream region of antA (986 bp) was amplified by PCR into three overlapping fragments using the primers AKHp01-05. All forward primers were biotinylated. The PCR products were excised after gel electrophoretic separation. Following agarose digest and isopropanol/glycogen precipitation, the DNA was re-dissolved in distilled water. Dynabeads® M-280 Streptavidin were used according to the manufactureŕs protocol (Invitrogen, Life Technologies). 100 μl of the Dynabeads® were loaded with 3 μg of the purified DNA fragment. P. luminescens was grown to an OD600 of 5.4 and harvested cells were stored at −20 °C. A cell pellet from 100 ml culture was resuspended in 5 ml lysis buffer [50 mM tris(hydroxymethyl)aminomethane Tris/HCl, pH 7.4, 5 mM ethylenediaminetetraacetic acid (EDTA) disodium salt, 100 mM NaCl, 1 mM dithiothreitol (DTT), 0.05% (v/v) Triton X–100, 1:200 protease inhibitor mix (Protease Inhibitor Cocktail SetIII, EDTA free, Calbiochem®) a spatula tip lysozyme]. For completion of cell lysis as well as destruction of cellular DNA the cell suspension was sonicated. Unsolvable cell fragments were removed by centrifugation. 3 μg promoter fragment (immobilized on the Dynabeads®) were incubated together with 10 μg salmon sperm DNA with 1.4 ml cell lysate. After 15 min of incubation, washing steps (15 min) with increasing salt concentration were performed [125/150/175/200 mM NaCl in 50 mM Tris/HCl, pH 7.4, 1 mM DTT, 0.05% (v/v) Triton X–100, 1:1000 protease inhibitor mix]. Final elution of the proteins was realized by incubation with 500 mM NaCl in 50 mM Tris/HCl, pH 7.4 for 30 min. The proteins in the cell lysate, in the washing steps (W1-W4) and the elution fraction were harvested using chloroform/methanol precipitation [45].

Peptide mass fingerprint

The dried protein pellets were solved in protein loading dye [20 mM Tris HCl, 4% glycerin, 2% (w/v) SDS, 0.3 M β-mercaptoethanol, 0.04% (w/v) bromophenol blue, pH 6.8] and proteins were size depended separated using SDS-PAGE. Bands of interest were excised from Coomassie Blue G-250 stained gels and in-gel digested with slight modifications as described earlier [46]. Reduction and alkylation were performed as described in [46] (step 4.A). Prior to overnight digestion at 37 °C in 60 μl digestion buffer [10 ng μl−1 trypsin, 10% (v/v) ACN, 10 mM NH4HCO3] the gel slice was washed three times with 50% 100 mM NH4HCO3/50% acetonitril (ACN) for 10 min, and then with ACN for 10 min. Peptide extraction was performed for 30 min at 37 °C with 30% ACN/0.1% trifluoroacetic acid (TFA) followed by a second extraction with 50% ACN/0.1% TFA. Both extracts were mixed and dried under reduced pressure. Afterwards the peptides were dissolved in 20 μl 30% ACN/0.1% TFA and measured with MALDI MS.

MALDI MS analysis and Mascot database search

MALDI MS analysis and sample preparation was described earlier [47]. The samples were 1:1 mixed with the matrix [20 mM 4-Chloro-α-cyanocinnamic acid (ClCCA) solved in 70% (v/v) ACN] on a MALDI-Target and air-dried. Spectra were analyzed with the Qual Browser (Version 2.0.7; Thermo Fisher Scientific, Inc., Waltham) and used for a database search on the mascot server (matrixscience.com) applying the following parameters: peptide mass fingerprint database: NCBInr; enzyme: trypsin (3 missed cleavages allowed); fixed modification: carbamidomethyl (C); variable modification: oxidation (M); peptide tolerance: 5 ppm; mass values: MH+, monoisotopic.

Integration of reporter genes into the genome of P. luminescens

In order to integrate P-mCherry as well as the P, P and P-antJ constructs into the genome of P. luminescens, the donor strain E. coli ST18 [42], which requires the addition of 5-aminolevulinic acid for growth, was first transformed with the respective plasmids. The conjugative plasmid transfer was subsequently performed by the filter mating method [42]. For that reason, the donor as well as the recipient strain was cultivated up to an OD600 of 0.8–1. The donor strain was washed in LB medium for 3 times and then mixed with the recipient strain in a ratio of 1:5 in a final volume of 1/10 of the donor‘s initial volume. A nitrocellulose filter was set onto an LB agar plate and the mixed cells were then dropped onto the filter without dispersion. After the incubation at 30 °C over-night, the cells were re-suspended in 500 μl LB and spread onto LB agar plates containing kanamycin and incubated for up to two days at 30 °C. The genomic DNA of single colonies were isolated and used as a template to check for chromosomal integration of the plasmid via PCR (primers: check-rpmE_fwd, oriT_fwd, gmR-pNPTS_fwd, check-mcherry-ins_rev, check-glmS_rev). As all the strains yielded additionally P, P and P (no induction) that had no fluorescence effect, the reporter strains are just referred to as P, P and P.

Transcription analyses using quantitative real-time PCR (qPCR)

P. luminescens 1° and 2° cells were cultivated as described above. Samples were taken at 9 h, 24 h and 30 h of growth before the total RNA was isolated using the Quick RNA Mini prep kit (Zymo Research, Freiburg im Breisgau). The RNA was then re-transcribed to cDNA via the random-primed first-strand cDNA synthesis kit (Thermo Scientific, Dreieich). Quantitative real-time PCR (iQ5 real-time PCR detection system, BioRad, München) was performed using the synthesized cDNA, a SYBR green detection system (BioRad, München) and specific internal primers for antJ and recA. Samples from three independently performed experiments were used and the Ct value (cycle threshold) was determined after 40 cycles using the iQ software (BioRad, München). Values were normalized with reference to recA and relative changes in transcript levels were calculated using the comparative ΔΔCt method [48].

RNA extraction and sequencing

For RNA sequencing, the respective P. luminescens strains were inoculated in LB broth with an OD600 of 0.3 with pre-cultures grown in the same medium. After 8–10 h of aerobic cultivation at 30 °C cells were harvested (OD600 = 5) and for extraction of total RNA, cells from 1 ml culture were re-suspended in 800 μl AE-buffer [20 mM NaOAc, 1 mM EDTA (pH 5.2) made in nuclease free water], mixed with 500 μl acidified phenol:chloroform and after transfer to a homogenization tube (lysing matrix B: 0.1 mm silica spheres, MP Biomedicals Germany GmbH) disrupted using a cell homogenizer (FastPrep-24R 5G, MP Biomedicals Germany GmbH) with two cycles of 30 s at 6 m s−1. After phase separation the aqueous phase was extracted two times with 500 μl of chloroform. Finally, the RNA was precipitated by adding 600 μl isopropanol and 120 μl 3 M sodium acetate at −20 °C overnight. The RNA was pelleted and washed twice with ice cold 70% (v/v) ethanol, dried on ice and re-suspended in 35 μl nuclease free water. Then DNase treatment (TURBO DNA-free™ Kit, Life Technologies) was performed following the manufacturer’s instructions. Eurofins Genomics GmbH (Ebersberg, Germany) performed RNA quality control and rRNA depletion immediately before sequencing the RNA samples. Data is available at the European Nucleotide Archive (ENA) using accession number (ERP018162). During sequencing, directionality of the reads was maintained and all sequencing data was mapped to the P. luminescens TT01 genome [49] using the bowtie function as a part of Nesoni (v0.128). The count function from Nesoni (v0.128) was subsequently used to count before analysing with Limma/Voom, which is part of Degust (v0.19 available at http://www.vicbioinformatics.com/degust/). A false discovery rate (FDR) of <0.01 was used and only sequences which were at least 1.5-fold up-/down-regulated compared to the WT were considered significant. Transcripts with more than 2.5-fold altered expression level in the empty vector control (TT01−1° + control1) compared to the WT were excluded (Table S4).

Analytical scale culture extraction and HPLC-UV/MS analysis

P. luminescens strains were inoculated in LB media in 100 ml Erlenmeyer flasks with an overnight culture to an OD600 of 0.1 and aerobically grown for 24 h at 30 °C. For ethyl acetate (EE) extractions 2 ml samples of the cultures were extracted for 1 h at RT with 2 ml EE + 0.1% FA. 1 ml of the organic supernatant was dried under nitrogen flow. The extracts were re-dissolved in 250 μl methanol and measured by HPLC-UV/MS. HPLC-UV/MS analysis was done as explained before [50]. We used an ACN/0.1% FA in H2O/0.1% FA gradient from 5 to 95% in 12 min at a flow rate of 0.6 ml/min at 30 °C. Chromatograms were analyzed using Bruker Compass DataAnalysis 4.2. For AQ quantification peak areas of the HPLC-UV analysis (UV-chromatograms at 430 nm) were used. For SM quantification, peak areas of the HPLC-MS chromatogram were determined using Bruker Compass TargetAnalysis Version 1.3. The respective peak areas were normalized against the cell density (OD600) when cultures were harvested.

Fluorescence microscopy and single-cell analysis

The P. luminescens fluorescence reporter strains were investigated using a fluorescence microscope (Leica, Bensheim, Germany). An excitation wavelength of 546 nm and a 605 nm emission filter with a 75 nm bandwidth was used to measure mCherry. The cultures were inoculated at an OD600 of 0.05 with the appropriate volume of the respective over-night cultures, and were then grown at 30 °C for 24 h in the presence of appropriate antibiotics. An agarose pad [0.5% (w/v) agarose in PBS buffer, pH 7.4] was prepared onto a microscope slide, 7 μl of the grown cultures were dropped onto the slide and covered with a coverslip. Three independent biological experiments were performed. For each experiment, the fluorescence intensities of 450 cells were measured using the single-cell analysis tool Big Cell Brother [28].

Purification of AntJ

For purification of AntJ for SPR analysis, E. coli BL21 was transformed with plasmid pBAD24-antJ-His. The cells were cultivated up to an OD600 of 0.5 and then gene expression was induced by the addition of 0.5 mM IPTG. After 3–4 h of aerobic cultivation at 30 °C the cells were harvested by centrifugation (6500 rpm, 15 min, 4 °C). After harvesting the cells were disrupted via using a Cell disruptor (Constant Cell Disruption Systems, Northants, UK) at 1.35 kbar and 4 °C in lysis buffer [50 mM Tris/HCl pH 7.5, 10% (v/v) glycerol, 10 mM MgCl2, 1 mM dithiotreitol, 0.5 mM phenylmethylsulfonylfluoride, and 0.03 mg/ml (w/v) DNase]. After removal of intact cells and cell debris by centrifugation (9.000 × g, 10 min), the cytosol was frozen at −80 °C. Purification was carried out as described before [51]. The equilibration and washing steps were performed in buffer E [50 mM Tris/HCl, pH 7.5, 10% (v/v) glycerol, 200 mM NaCl, 2 mM β-mercaptoethanol, 10 mM imidazole], and 250 mM imidazole was added for the elution of the protein from the column.

For purification of AntJ for EMSA experiments, E. coli BL21 Star (DE3) was transformed with plasmid pCATI4_antJ. Cells were aerobically grown at 30 °C in autoinduction medium [LB medium supplemented with 3.3 g/l (NH4)2SO4, 6.8 g/l KH2PO4, 7.1 g/l Na2HPO4, 5 g/l glycerol, 0.5 g/l glucose and 2.0 g/l α-lactose] [52]. After 24 h cells were harvested (30.000 x g, 15 min, 4 °C) and pellets were stored at −20 °C. Cell lysis and protein purification were performed as described before [53]. Apart from this all buffers were adjusted to pH 7 and contained additionally 10% (v/v) glycerol and 0.1% Triton X–100 in order to improve solubility of AntJ. Purified AntJ was immediately used for EMSA experiments.

Electrophoretic mobility shift assay (EMSA)

Labeling of DNA regions for EMSA experiments was performed as described before [54]. The antA upstream region was subdivided into overlapping fragments a-f (100 bp). Each fragment shares a 50 bp overlapping region with its adjacent fragment(s). The fragments were amplified and expanded to 140 bp using oligonucleotides AKHp30-41, which add 20 bp stretches to both sides of the fragment homologous to 5′Cy3-labeled oligonucleotide AKHp28 respectively AKHp29. AKHp28 and AKHp29 were used in a second PCR. EMSAs were performed with minor modifications as described before [54]. Briefly 0.125 pmol 5′-Cy3 labeled DNA fragments, together with 2.5 μg salmon sperm DNA and 0.5 pmol heterologously expressed and purified AntJ were incubated for 5 min at RT in shift buffer [25 mM Tris-acetate, 0.5 mM DTT, 50 mM KC2H3O2, 8 mM Mg(C2H3O2)2] before loading on native 4.5% polyacrylamide gels. Protein-DNA complexes were separated from unbound DNA on the gel in TAE buffer (40 mM Tris-acetate pH 7.8, 1 mM EDTA) applying a voltage of 3.5 V/cm. 5′-Cy3 labeled DNA fragments on the gel were detected as described before [54].

Surface plasmon resonance (SPR) spectroscopy

SPR assays were performed in a Biacore T200 (GE Healthcare, München) using carboxymethyl dextran sensor chips pre-coated with streptavidin (XanTec SAD500L, XanTec Bioanalytics GmbH, Düsseldorf). Biotinylated fragments of the antA promoter were generated by amplification of the P region using chromosomal DNA of P. luminescens as template, and using the biotinylated primer PantA-Btn_fwd and PantA-XmaI_rev. Before immobilizing the DNA fragment, the chip was equilibrated by three injections using 1 M NaCl/50 mM NaOH at a flow rate of 10 μl/min. 10 nM of the respective double-stranded biotinylated DNA fragment was injected using a contact time of 420 s and a flow rate of 10 μl/min. As a final wash step, 1 M NaCl/50 mM NaOH/50% (v/v) isopropanol was injected. Approximately 300 RU of the DNA fragment was captured onto flow cell 2 of the chip. AntJ was diluted in HBS-EP+ buffer [10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.05% (v/v) Surfactant P20] and passed over flow cells 1 and 2 in different concentrations (0 nM, 10 nM, 25 nM, 50 nM, 100 nM, 125 nM, 250 nM, and 500 nM) using a contact time of 180 s followed by a 240 dissociation time before the next cycle started. The experiments were carried out at 25 °C at a flow rate of 30 μl/min. After each cycle the surface was regenerated by injection of 2.5 M NaCl for 60 s at 30 μl/min flow rate. Sensorgrams were recorded using the Biacore T200Control software 2.0 and analyzed with the Biacore T200 Evaluation software 2.0. The surface of flow cell 1 was used to obtain blank sensorgrams for subtraction of bulk refractive index background. The referenced sensorgrams were normalized to a baseline of 0. The 1:1 binding algorithm was used for calculation of the binding affinity. Peaks in the sensorgrams at the beginning and the end of the injection emerged from the runtime difference between the flow cells of each chip. Experiments were performed in the Bioanalytics core facility of the LMU München.

Phylogenetic analysis

For the identification of the 100 most similar protein sequences of AntJ, a BLASTp search against the non-redundant protein database from NCBI was performed. For the phylogenetic analysis ClustalW was used to align the sequences. Three sequences were deleted from the alignment due to the occurrence of unusual amino acids. The tree was generated with the Geneious tree builder incorporated into Geneious (v6.1.8) (Biomatters Ltd., New Zealand) utilizing the Jukes-Cantor distance model.

Declarations

Author contribution statement

Antje K Heinrich, Angela Glaeser: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

Nicholas J Tobias, Ralf Heermann, Helge B Bode: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.

Funding statement

Research in the Bode and the Heermann laboratories were both supported by the Deutsche Forschungsgemeinschaft (SPP1617).

Competing interest statement

The authors declare no conflict of interest.

Additional information

No additional information is available for this paper.