Bacterial strains and growth conditions

For targeted mutagenesis of the hemK/hemR operon in Leptospira biflexa, a suicide vector containing a kanamycin resistance cassette flanked by 600 bp of the 5’-end of hemK and 3’-end of hemR was introduced in L. biflexa serovar Patoc strain Patoc 1 by electroporation with a Biorad Gene Pulser Xcell, as previously described [27]. Electroporated cells were plated on EMJH agar plates supplemented with 50 μg/ml kanamycin. Plates were incubated for 1 week at 30°C and kanamycin-resistant colonies were screened for double homologous recombination events by PCR. The hemK^-/hemR^- double knockout mutant thus generated (named ΔhemKR in this study) was further confirmed by whole-genome sequencing. Leptospira biflexa wt and ΔhemKR strains were grown at 30°C in Ellinghausen-McCullough-Johnson-Harris (EMJH) liquid media [28]. When needed strains’ genotypes were confirmed by PCR with oligonucleotides HemK_Fw and HemR_Rev (S1 Table). For growth measurements, L. biflexa wt and ΔhemKR were grown at 30°C in standard EMJH medium with no agitation. When cultures reached exponential phase (OD₄₂₀ ~0.25), a 1:100 inoculum of each strain was subcultured in EMJH medium without supplemented iron, in the presence of 0, 35, 70 or 175μM 2,2’-dipyridyl (DIP), at 30°C with no agitation until reaching stationary phase (10 days). Bacteria motility and viability were examined using a dark field microscope (AxioImager A2.LED, Zeiss).

The shuttle expression plasmid pMaORI (Spc^R) [29] was used to express mutated HemK-encoding gene constructs in Leptospira. Cloning and pMaORI manipulation/preparation was done in E. coli strain π1 (ΔthyA) cells [30]. pMaORI was transformed into E. coli strain β2163 (ΔdapA) [30]. E. coli strains π1 and β2163 were grown in LB containing 50 μg/mL spectinomycin, the former strain supplemented with 0.3 mM thymidine, whereas the latter with 0.3 mM diaminopimelic acid (DAP) [30].

Cloning of complementation plasmids

To complement L. biflexa ΔhemΚR mutant, the wildtype hemKR allele harboring its own promoter region (349 bp upstream hemK’s start codon), was amplified from wt L. biflexa using primers HemR_pMaORI_Fw, HemR_pMaORI_Rev, HemK_pMaORI_Fw, HemK_pMaORI_Rev, p_hemKR_pMaORI_Fw, and p_hemKR_pMaORI_Rev (S1 Table). Site-directed mutagenesis was performed to generate a hemK_T102A mutant allele with primers HemK_T102A_Fw and HemK1_Rev (S1 Table) and inserted into pMaORI [29] by restriction-free cloning [31]. Correctly recombined pMaORI constructs were confirmed by BamHI/EcoRI digestion profile. Cloned pMaORI-hemKR and pMAORI-hemK_T102AR plasmids were then transformed into E. coli strains π1 and β2163 for downstream conjugation experiments.

Conjugation

E. coli strain β2163 cells harboring the pMaORI-hemKR and pMaORI-hemK_T102AR plasmids were co-incubated with L. biflexa cells in EMJH liquid media supplemented with 0.3 mM DAP (when required), until OD₄₂₀ 0.3, according to standard procedures [32]. Conjugants were incubated in solid EMJH agar plates supplemented with 50 μg/mL spectinomycin, at 30°C for 7–10 days. Selected colonies were grown in liquid EMJH medium supplemented with 50 μg/mL spectinomycin, and the presence of the plasmid was confirmed by PCR with primers HemK_Fw and HemK2_Rev (S1 Table).

Expression and purification of recombinant proteins

pQE80L (Qiagen) plasmids harboring HemK_Δ80 (encoding a truncated version of the kinase, lacking the first 80 amino acids), HemK_Δ80,T102A (HemK_Δ80 with threonine 102 substituted by alanine), HemR_REC (receiver domain only) and HemR_FL (full-length) ORFs, each with an encoded Tobacco Etch Virus (TEV) protease cleavage site for N-terminal His-tag removal, were transformed into E. coli Rosetta-gami(DE3)—for HemK—or TOP10F’—for HemR—cells. Transformed Rosetta-gami(DE3) cells were grown with agitation (250 rpm) in 2xYT medium supplemented with 100 μg/mL ampicillin at 37°C until OD₆₀₀ ~ 1. Induction was carried with 1 mM IPTG at 30°C for 6 h. Transformed TOP10F’ cells were grown similarly, until OD₆₀₀ ~ 0.6. Induction was performed using 1 mM IPTG at 37°C for 3 h. Cells were harvested by centrifugation at 4000g, and pellets resuspended in 50 mM Tris pH 8, 500 mM NaCl, EDTA-free protease inhibitors (Roche) and 0.1 mg/mL DNase I (Sigma). Cells were disrupted with 1 mg/mL lysozyme followed by 1 cycle of freezing/thawing and high-pressure homogenization (Emulsiflex-C5, Avestin).

Soluble fractions were obtained by centrifugation at 30000g for 30 min, and supernatants subjected to Ni²⁺ affinity chromatography (HisTrap, Cytiva) in buffer A (50 mM buffer Tris pH 8, 500 mM NaCl and 120 mM imidazole for recombinant HemK_Δ80 and HemK_Δ80,T102A purification, or 20 mM imidazole for HemR_REC and HemR_FL). Elution was performed with a gradient of buffer B (buffer A with added 750 mM imidazole for HemK_Δ80 and HemK_Δ80,T102A purification, and 500 mM imidazole for HemR_REC and HemR_FL). Eluted fractions were pooled and incubated overnight with TEV protease in dialysis buffer (50 mM Tris pH 8, 500 mM NaCl). Digested pools were subjected to a second Ni²⁺ column purification step to separate cleaved His-tag and TEV protease from the sample. All samples were finally subjected to size-exclusion chromatography (Superdex 16/60 75 prep column, Cytiva), previously equilibrated with 25 mM Tris pH 8 and 500 mM NaCl. The peak fractions were pooled, concentrated, and stored at -80°C until use.

Autophosphorylation, phosphoryl-transfer and dephosphorylation catalysis assays

For autophosphorylation assays, 20 μM HemK_Δ80 or HemK_Δ80,T102A were incubated with 5 mM ATP, 10 mM MgCl₂. For phosphotransfer assays, 20 μM HemK_Δ80 or HemK_Δ80,T102A previously incubated with 5 mM ATP, 10 mM MgCl₂ for 60 min at room temperature (RT), were then incubated with equimolar concentrations of HemR_REC or HemR_FL. For phosphatase assays, 600 μM HemR_REC or HemR_FL were first phosphorylated by incubating with 5 mM acetyl phosphate, 10 mM MgCl₂, and then dimeric, phosphorylated HemR_REC or HemR_FL, were purified from monomeric, non-phosphorylated forms, by size-exclusion chromatography (Superdex 10/300 75 prep, Cytiva). 20 μM of phosphorylated HemR_REC or HemR_FL were incubated with equimolar concentrations of HemK_Δ80 or HemK_Δ80,T102A. At defined time intervals, reactions were stopped with SDS buffer and 25 mM dithiothreitol (DTT) for 10 min at RT, followed by further incubation with 50 mM iodoacetamide for additional 10 min. All samples were then separated by PhosTag-SDS-PAGE, as described below. Gels were stained with Coomassie blue, scanned with a CanoScan Lide 110 (Canon) scanner, and bands quantified by densitometry using ImageJ [33].

PhosTag-SDS-PAGE and Western blot of whole protein extracts

L. biflexa cultures were grown in 30 mL EMJH until OD₄₂₀ ~0.2. Prior to 5-aminolevulinic acid (ALA) or 2,2’-dipyridyl (DIP) incubation, an aliquot of 6.5 mL of untreated culture was harvested at 10000 g for 10 min at 4°C. When indicated, 300 μM ALA or 35 μM DIP was added to the cultures, and 6.5 ml aliquots were harvested at 1, 2 and 3 h post-incubation. Cell pellets were lysed resuspending with BugBuster and Lysonase according to the manufacturer’s protocol (Merck). Lysed cells were fractionated by centrifugation at 16000 g for 20 min at 4°C, and an aliquot of the soluble fraction was resuspended in SDS buffer with 25 μM DTT and incubated 10 min at RT, followed by 50 μM iodoacetamide for extra 10 min at RT. Samples were run in SDS-PAGE 12% bis-acrylamide, co-polymerized with 100 μM PhosTag (Wako) and 200 μM ZnCl₂. PhosTag in the presence of transition metals, shifts the electrophoretic mobility of phosphorylated proteins [34, 35], allowing to quantify P~HemR/HemR ratios. Electrophoreses were run at <10°C. Before Western blotting, PhosTag was eliminated by washing gels with SDS-PAGE running buffer and 2mM EDTA for 10 min, and then with no EDTA for an extra 10 min. Gels were blotted to nitrocellulose HyBond (Cytiva) membrane overnight with a TE 22 Mighty Small Transphor Tank Transfer Unit (Amersham) at 20 V.

A polyclonal monospecific anti-HemR antibody (αHemR) was produced in rabbits. Immunizations were done at Instituto Polo Tecnológico de Pando (Uruguay) respecting ethical guidelines for the use of animals in research. Briefly, 100 μg pure recombinant HemR (produced in E. coli, see [26] for construct and purification procedures) were injected subcutaneously with complete Freund’s adjuvant, followed by two boosters of 50 μg HemR in incomplete Freund’s at 14 and 28 days post-first immunization. Animals were bled for intermediate titration assays, and at ~60 days post-immunization for sera separation, stored at -20°C until use. Western blot membranes were pre-blocked with 3% bovine serum albumin in 0.1% PBS-Tween for 3.5 h, at RT and gentle agitation. αHemR was diluted 1/100 in blocking solution, and incubated with the membrane for 2 h at RT, with gentle agitation. Membranes were washed four times with 0.1% PBS-Tween 15 min/each, with strong agitation. Secondary anti-rabbit IgG (Sigma) antibody diluted 1/20000 in blocking solution was added to the membrane, with gentle agitation for 1 h RT. Four washing steps were repeated as before, and membranes were finally incubated with bioluminescent reagents (BM Chemiluminescence Western Blotting Kit, Roche) for 5 min. Protein bands were revealed and quantified using an ImageQuant^TM 800 (Cytiva) imaging system. Percentage of P~HemR (%P~HemR) was calculated by densitometry using ImageJ Fiji, subtracting background and subsequently quantifying total HemR signal (HemR_TOTAL), which is the sum of the signal corresponding to P~HemR plus the signal corresponding to non-phosphorylated HemR (HemR_NONP) (HemR_TOTAL = P~HemR + HemR_NONP). Then, %P~HemR = P~HemR*100/HemR_TOTAL.

RNA extraction

L. biflexa wt and ΔhemKR mutant strains were grown in liquid EMJH media with no shaking at 30°C until mid-exponential phase (OD₄₂₀ ~0.2). Untreated cultures and cultures treated with FeSO₄ 2 mM, δ-aminolevulinic acid (ALA) 300 μM, and 2,2’-dipyridyl (DIP) 35 μM were further incubated 3 h at 30°C with no shaking. RNA extraction was then performed as previously described [36].

RNAseq

RNAseq libraries were prepared with 3 replicates for each condition, using a TruSeq Stranded mRNA library Preparation Kit (Illumina, USA) following the manufacturer’s protocol. The libraries were sequenced on an Illumina NextSeq2000 platform, generating single-end reads with an average read length of 107 bp. Quality control and data pre-processing were conducted to ensure robust downstream analyses. The number of reads was between 12-20M reads per sample, with an average phred score of 33. Transcription of the hemK and hemR genes were analyzed in detail in the ΔhemKR mutant strain as a direct control of the knockout. hemR (LEPBIa1422) transcription was indeed abolished in all conditions. As for hemK (LEPBIa1423), standard differential gene expression (DGE) analyses did not detect a significant shift with respect to the wt; yet, a more detailed mapping on the genomic loci revealed that the reads corresponding to hemK, comprise only the initial 5’ fragment of the gene, upstream of the kanamycin-resistance cassette. Fold-change of individual mRNA species was an obvious figure to analyze DGE. The bioinformatic analyses of whole transcriptomes and differential gene expression were performed with Sequana [37]. Specifically, the RNAseq pipeline (v0.15.1) available at https://github.com/sequana/sequana_rnaseq was employed, built upon the Snakemake framework [38]. To prepare the data, reads were trimmed for adapter sequences and low-quality bases using fastp software v0.20.1 [39] and were subsequently mapped to the Leptospira genome using bowtie2 [40]. The nucleotide sequence and annotation were downloaded from the MicroScope platform [41] using L. biflexa serovar Patoc Patoc 1 entry (corresponds to NCBI genome RefSeq assembly GCF_000017685.1). The count matrix was generated using FeatureCounts 2.0.0 [42], which assigned reads to relevant features based on the previously mentioned annotation. To identify differentially regulated genes, statistical analysis was performed with the DESeq2 library 1.30.0 [43], and HTML reports were generated using the Sequana RNAseq pipeline. Key parameters for the statistical analysis encompassed significance, measured by Benjamini-Hochberg adjusted p-values with a false discovery rate (FDR) threshold of less than 0.05, as well as the effect size, quantified through fold-change calculations for each comparison. The hem-box DNA-binding motif was scanned with the FIMO Motif Search Tool (v. 5.5.1) [44].

qRT-PCR

Transcription levels of target genes hemA (LEPBIa1171) and hmuO (LEPBIa0669) were evaluated in L. biflexa wt and ΔhemKR cells by quantitative real time PCR 9qRT-PCR). RNA extracted as described above, was used for cDNA synthesis using iScript cDNA Synthesis Kit (BioRad). PCR amplification was done on cDNAs using a QuantStudio 3 thermocycler (Applied Biosystems, Thermo Fisher Scientific), using the primers listed in supplementary S1 Table, and followed in real time by using SsoFast EvaGreen Super-mix (BioRad), as follows: 15 min at 95°C; 40 x (0.25 min at 95°C, 1 min at 60°C); 0.25 min at 95°C; melting curves 1 min at 60°C; 0.25 min at 95°C with increment 0.1°C/s. The relative expression of target genes was calculated using the delta-delta Cq method and normalized against rpoB mRNA levels as described before [26].

The cytoplasmic catalytic region of HemK is trapped in the kinase-active state

A soluble construct of HemK (HemK_Δ80) lacking the first 80 amino acids that span the periplasmic and trans-membrane regions of the native protein, possesses ATP-dependent autokinase activity (Fig 1A), confirming previous observations [24]. Further characterizing the kinetics of the reaction, HemK_Δ80 achieved 50% maximal phosphorylation in ~60 min in the presence of excess ATP as phosphoryl donor (Fig 1A). P~HemK_Δ80 was also able to catalyze rapid phosphoryl-transfer to the phospho-receiver domain of HemR (HemR_REC) (Fig 1B), with undetectable back-transfer (Fig 1C), similarly to what occurs in other HK families [45]. Threonine 102 (T102) is predicted to play a key role in HemK’s phosphatase activity, as this is a conserved residue in this HK family (HisKA), engaged in the dephosphorylation of the RRs’ phospho-aspartate [4, 46]. Neither the autophosphorylation nor the phosphoryl-transfer reaction kinetics were affected by substituting HemK’s T102 by an alanine (Fig 1A and 1B). Even though most HisKA HKs exhibit specific phosphatase activity towards their cognate P~RRs [4], HemK_Δ80-dependent P~HemR dephosphorylation was not detected in vitro (Fig 1C). Different conditions were assayed, especially considering known requirements of other HisKA kinases to fully catalyze phosphatase reactions. Among these, the inclusion of ADP, and/or the use of full-length P~HemR as substrate (Fig 1C), did not result in HemK_Δ80 phosphatase enhancement.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Autokinase, phosphotransferase and phosphatase activities catalyzed by HemK.

(a) Autophosphorylation of HemK_Δ80 (left panel) and the phosphatase-null point-mutant HemK_Δ80,T102A (right), incubated for indicated times with 5mM ATP-Mg²⁺. (b) Phosphoryl-transfer kinetics from P~HemK_Δ80 (left) or P~HemK_Δ80,T102A (right) to HemR_REC, in the presence of 5mM ATP-Mg²⁺. (c) HemK-catalyzed phosphatase activity on P~HemR_REC, using HemK_Δ80 (left panel) or HemK_Δ80,T102A (center). The dephosphorylation of the regulator was not detectable under any condition. 5mM ADP-Mg²⁺ was included in the reactions, as some HisKA His-kinases are known to require nucleotide to exert their phosphatase activity (indistinguishable results were obtained with no nucleotide added). The right-most panel reproduces the previous panels’ assays but using a full-length construct of HemR (HemR_FL), ruling out the potential need of its DNA-binding domain to detect HemK phosphatase activity. The phosphorylated species of HemK and HemR are indicated (P~HemK and P~HemR) and suffer a mobility shift during electrophoresis due to the co-polymerization of PhosTag into the polyacrylamide gels. Each experiment was performed in duplicate, representative Coomassie-stained gels are shown.

https://doi.org/10.1371/journal.pone.0311040.g001

Together with previous evidence exploring the effect of the key phosphorylatable histidine (H98) and aspartate (D53) residues on HemK and HemR, respectively [24], we now show that HemK_Δ80 specifically phosphorylates HemR in a unidirectional forward-transfer manner. Surprisingly, in vitro HemK_Δ80 is not able to dephosphorylate P~HemR, altogether suggesting that HemK_Δ80 is intrinsically trapped in a kinase-active state, in need of its sensory/transmembrane portion to shift to a phosphatase-competent state under appropriate signaling conditions (as demonstrated below).

The heme precursor 5-aminolevulinic acid triggers the stimulation of HemK phosphatase activity towards P~HemR

To discover signals sensed via the HemKR system, we asked whether the L. biflexa HemKR phosphorylation status was directly affected by exposition/deprivation of the cells to heme-synthesis building moieties (Fig 2). First attempts were done using heme itself. However, its toxicity at low micromolar concentrations, and the overwhelming global transcriptional effect we observed in L. biflexa cells exposed to <10μM hemin, hampered clearcut interpretations. Hence, a simpler scheme was followed using other heme-building blocks, such as 5-aminolevulinic acid (ALA) and iron. L. biflexa cells grown in EMJH medium until mid-exponential phase, were left untreated, or were supplemented with 300μM ALA, 35μM 2,2’-dipyridyl (DIP, a strong metal chelator) or 2mM FeSO₄ (Fig 2). Bacterial whole protein extracts were then separated with PhosTag-SDS-PAGE, and HemR was revealed by Western blot using an anti-HemR polyclonal antibody (Fig 2A). We observed that in normal growth conditions (untreated) L. biflexa P~HemR represents ~12–25% of the total pool of HemR in the cells, and that this ratio did not change throughout the course of the experiment (Fig 2B). Excess or deficit of iron also did not produce detectable changes of HemR phosphorylation. However, exposure to ALA provoked a sustained depletion of P~HemR, already observed at 1h of incubation (Fig 2A and 2B). To uncover the molecular mechanism of this ALA-triggered effect, we hypothesized that HemK directly or indirectly senses ALA (or a downstream metabolite derived from ALA, including fully synthesized porphyrins such as heme), leading to the activation of its phosphatase activity towards P~HemR. To test this hypothesis, we performed the same analysis using a L. biflexa ΔhemKR mutant strain [24] complemented with plasmid PhemK_T102A/hemR, encoding for HemK_T102A phosphatase-null mutant and wt HemR [4, 46] under the control of the operon’s native promoter. Notoriously, this strain exhibited comparable levels of HemR and P~HemR as the wt strain, yet did not trigger P~HemR dephosphorylation in response to ALA (Fig 2C and 2D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. 5-aminolevulinic acid, but not Fe³⁺, shuts down HemK-mediated phosphorylation of HemR in vivo.

(a) PhosTag-SDS-PAGE of wt L. biflexa whole protein extracts, followed by Western blot using an anti-HemR antibody. Bacteria were left untreated, or treated with 5-aminolevulinic acid (ALA), 2,2-dipyridyl (DIP), or an excess of iron (FeSO₄), as indicated. Black arrows indicate the electrophoretic mobility position of phosphorylated and unphosphorylated HemR species. (b) Quantification of P~HemR levels from panel (a) shown as percentage. Measurements were obtained from three biological replicates. (c) PhosTag-SDS-PAGE followed by Western blot as in panel (a). L. biflexa ΔhemKR mutant strain complemented with plasmid PhemK_T102A/hemR was left untreated or treated with ALA for 3h. (d) Quantification of P~HemR levels from panel (c), shown as percentage. Measurements were obtained from three biological replicates.

https://doi.org/10.1371/journal.pone.0311040.g002

Altogether, our results confirmed that HemK possesses phosphatase activity in vivo, leading to P~HemR dephosphorylation in a signal-dependent manner. Indeed, ALA, but not iron, was shown to be a potential signal to shut down the HemKR pathway.

HemKR-dependent gene regulation is consistent with a response effecting heme/iron homeostasis

Previous efforts to define the HemR regulon had pinpointed two targets, which code for enzymes involved in heme biosynthesis and degradation [24, 26]. To further our understanding, we performed whole transcriptome analysis to identify HemKR-dependent differential gene expression (DGE) in L. biflexa, using ALA as an effective signal to shut down the pathway (S2 Table). We also used the wild-type strain (wt) and a double knockout with both hemK and hemR genes disrupted (ΔhemKR), to discriminate direct implications of HemKR in the ALA-triggered response (S2 Table). L. biflexa single gene knockout strains (ΔhemK and ΔhemR) had been previously characterized as auxotrophic for heme, a deficiency that could also be rescued by adding ALA to the growth medium [24]. Intriguingly, L. biflexa ΔhemKR strain grew in the absence of ALA/heme, even though no evident differences could be identified among single and double knockout strains’ whole genome sequences, other than the targeted gene disruption sites. The growth of the ΔhemKR strain was also not affected by washing thoroughly and passing the cells eight consecutive times in EMJH medium without ALA/heme supplementation, altogether confirming that this double knockout mutant is capable of de novo heme biosynthesis.

Transcriptomic analyses confirmed that transcription of hemK and hemR genes was indeed abolished in the mutant strain (S2 Table). The kanamycin cassette introduced to disrupt the hemK/hemR operon locus, sits at base-pair 286 starting from the 5’ end of hemK’s open reading frame. Hence, only short transcripts mapping onto this short 5’ hemK fragment were detected, confirming that HemK protein cannot be expressed(see Experimental Procedures for further details). DGE analysis of ALA-treated vs untreated wt Leptospira cells, revealed a significant downregulation of 128 genes and upregulation of 51 (log₂ fold change (FC) > |1|). This contrasted with the effect in the ΔhemKR mutant, which showed only 6 downregulated and 24 upregulated genes under the same experimental conditions (Fig 3A), consistent with HemKR being engaged in mediating the ALA-triggered response. Many ALA-responsive genes encode hypothetical proteins, for which functional roles require future studies (S3 Table). Yet, several key heme and iron metabolism-related genes were found to be differentially expressed in the wt strain and not in the ΔhemKR mutant (Fig 3A and S3 Table). Among such targets, the heme biosynthesis operon hemA was down-regulated, as well as exbB1/D1 and LEPBIa3432, the latter two linked to TonB-dependent iron/siderophore/porphyrin outer-membrane transport. On the other hand, the heme-degrading hmuO heme oxygenase, displayed modest yet significant overexpression. Two representative targets, up- and down-regulated, were further confirmed by qRT-PCR (Fig 3B and S4 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. 5-aminolevulinic acid induces a largely HemKR-dependent transcriptional regulation response.

(a) Transcriptomic differential gene expression (DGE) analysis of ΔhemKR and wt L. biflexa strains, comparing cells that were treated with 300μM 5-aminolevulinic acid (ALA) vs untreated (see S1 Table for full data). The plot is a dual log₂ fold change (log₂FC), comparing the DGEs of ΔhemKR vs wt strains. Note the stronger regulation in the wt strain compared to the ΔhemKR mutant (flat pattern of log₂FCs distribution; the red dashed line is the linear least squares regression curve with slope 0.1954 ± 0.011, and coefficient of determination R² = 0.084). Blue dots represent genes with P-values (Padj)≤0.05 in both strains; green dots, genes with Padj≤0.05 only in L. biflexa wt; yellow dots, genes with Padj≤0.05 only in L. biflexa ΔhemKR; grey dots, genes with Padj>0.05. Black arrows label selected genes with functional annotation or clearcut homology to functionally characterized orthologs involved in heme/iron metabolism; purple arrows, correspond to hypothetical genes or non-protein-coding RNAs. (b) Confirmation of DGE effects triggered by ALA, performing qRT-PCR of two target genes, one repressed by ALA (hemA) and one induced (hmuO). Statistical significance of differences was calculated by Student’s t test using three replicas (ns = P>0.05; * = P≤0.05; ** = P≤0.01;*** = P≤0.001). (c) Volcano representation of significance vs fold change analyzing the transcriptomic DGE between L. biflexa ΔhemKR vs wt strains grown under standard EMJH culture conditions. Labeled genes are the same as in panel (a), except for hemA (here not significant with log₂FC -0.4191). (d) Dual log₂FC plot comparing ‘wt ± ALA treatment’ vs ‘ΔhemKR vs wt’. Note the diagonal pattern of distribution (least squares regression in red dashed line, slope 0.4732 ± 0.009, and coefficient of determination R² = 0.301) revealing a significant proportion of coincident differentially regulated genes in wt+ALA compared to ΔhemKR.

https://doi.org/10.1371/journal.pone.0311040.g003

That the dual log₂FC plot of differential expression results in a horizontally flat distribution (Fig 3A), uncovers the stronger transcriptional response of L. biflexa wt to ALA treatment in comparison to the ΔhemKR mutant strain. The behavior of the hemA and exbB1 operons, expected to be activated by P~HemR, was consistent with ALA triggering HemKR shutdown. However, fold-change figures were modest. To obtain further confirmatory evidence, we then asked whether the removal of HemR, in the ΔhemKR strain, could be a relevant proxy of HemR dephosphorylation. Transcriptomic DGE was performed, comparing L. biflexa ΔhemKR vs wt strains grown in standard EMJH conditions (Fig 3C and S2 Table). The elimination of HemKR resulted in overall similar effects to those visualized with the wt strain under ALA exposure. Moreover, the ΔhemKR knockout exhibited even higher fold-change figures (Fig 3A and 3C). The dual log2FC plot in Fig 3D shows an overall linear relationship in the direction of differential expression suggesting a similar transcriptomic effect of the exposure to ALA and hemKR deletion, both leading to HemKR shutdown. The hemA operon was a notable exception, repressed in the wt after exposure to ALA while not significantly affected in the ΔhemKR mutant, it strongly suggests a composite control by different regulatory elements, which has been reported for this gene cluster in other bacteria [47].

Several of the differentially regulated genes, both in wt+ALA and in the ΔhemKR mutant, comprise hem-box motifs within their promoter regions, such as hmuO and hemA, the expression of which is indeed known to respond to HemR control [26]. We now demonstrate that exbB1/exbD1 and LEPBIa3432 are also part of the HemR regulon. While the former had been previously anticipated [26], we have now identified a new motif TGACAGTACTGTGACA on the minus strand of the genome (nucleotide coordinates 3,550,754–3,550,769) [44], compatible with the hem-box consensus, and just upstream of LEPBIa3430/LEPBIa3431. These two genes were indeed downregulated, similarly to LEPBIa3432, altogether suggesting they may form an operon. Several other ALA-responsive targets do not display easily discernible hem-box motifs, yet the majority correspond to genes encoding hypothetical proteins, justifying focused studies to unveil their regulatory mechanism and functional assignments.

In sum, ALA–and/or some derived metabolite along the porphyrin biosynthesis pathway–shuts down the HemKR system, anticipating heme synthesis inhibition (due to HemA decrease) and simultaneous stimulation of heme degradation (due to HmuO heme oxygenase’s surge).

The phenotype of the ΔhemKR knockout strain revealed marked tolerance to iron deficit

Notoriously, LEPBIa1883 lies among the most significantly upregulated genes in the L. biflexa ΔhemKR strain compared to wt (Fig 3C and 3D). This gene codes for the outer-membrane TonB-dependent ferric citrate receptor/transporter FecA. The expected effect would be for a larger pool of iron to become available in this mutant that lacks the HemKR system, a convergent effect with the one induced by heme oxygenase, which also liberates iron from heme. We thus conjectured that the ΔhemKR strain would be more tolerant to environmental iron deficit than the wt. To test this hypothesis, we assessed cell growth under increasing amounts of the iron-chelator DIP, producing mild to severe iron deficiencies. While the wt strain displayed the expected sensitivity to iron deprivation, L. biflexa ΔhemKR was significantly more resistant (Fig 4). Complete depletion of iron can be achieved by further increasing the concentration of DIP (175 μM), at which point both wt and ΔhemKR cultures arrested growth, only to be reverted–in both strains–by adding the xenosiderophore deferoxamine. All evidence considered, we confirmed that the L. biflexa ΔhemKR mutant, lacking the HemKR TCS, is indeed more tolerant to iron deficiency than the wt strain.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. L. biflexa ΔhemKR strain is more tolerant to iron deficiency.

L. biflexa wt and ΔhemKR were grown in EMJH medium with no supplemented iron. Cultures were left untreated or treated with increasing concentrations of the iron chelator 2,2’-dipyridyl (DIP) and incubated at 30°C with no agitation for 10 days. One representative experiment is shown from three biological replicates.

https://doi.org/10.1371/journal.pone.0311040.g004

Iron-deficiency triggers a transcriptional response that implicates a complex regulatory network beyond HemKR

The tolerance of L. biflexa ΔhemKR to iron deprivation suggested a crosstalk between heme- and iron-responsive regulatory pathways. We studied the systemic effect of iron on gene expression, in the wt and ΔhemKR strains. Iron-starving conditions were brought about by adding 35μM DIP to the culture, whereas iron excess was attempted by using 2mM FeSO₄-supplemented medium (instead of 330 μM in standard EMJH). Unfortunately, iron overabundance hindered RNA-purification in the wt strain and could not be further analyzed (S2 Fig).

RNAseq analysis of L. biflexa under iron-restricted conditions revealed a strong regulatory response in both wt and ΔhemKR mutant (Fig 5A and S2 Table). Largely consistent with reports studying L. interrogans [14] and other bacteria [48] under similar stress, L. biflexa wt cells responded to Fe³⁺ deficit by downregulating 151 genes and upregulating 105 (considering DGEs of log₂FC>|1|).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Iron deficit induces transcriptional regulation with mixed HemKR-dependent/independent responses.

Transcriptomic differential gene expression analysis comparing L. biflexa cells treated vs untreated with 35 μM DIP for 3 h at 30°C with no agitation. (a) Dual log₂ fold change (log₂FC) plot, comparing differential gene expression (treated with DIP vs untreated) of ΔhemKR vs wt strains. Note the same trend in regulation behavior between wt and mutant strains (diagonal pattern of log₂FCs distribution; the red dashed line is the linear least squares regression curve with slope 0.5349 ± 0.009, and coefficient of determination R² = 0.456). Coloring and labeling schemes as in Fig 3A. (b) Confirmation of DGE effects triggered by DIP, performing qRT-PCR of the same target genes as in Fig 3B (hemA, hmuO). Statistical significance of differences was calculated by Student’s t test using three replicas (ns = P>0.05; * = P≤0.05; ** = P≤0.01;*** = P≤0.001; **** = P≤0.0001).

https://doi.org/10.1371/journal.pone.0311040.g005

Although many of the iron-responsive genes code for hypothetical proteins, annotated genes linked to heme and/or iron metabolism were significantly affected (S5 Table), and two representative targets were further confirmed by qRT-PCR (Fig 5B and S4 Table). Even though the pattern of the iron-deficit response was globally similar in wt vs ΔhemKR (upper right and lower left areas along the diagonal in Fig 5A), a significant number of genes displayed fold-change differences between both strains, strongly suggesting a synergistic overlap of HemKR and additional iron-responsive regulatory systems. For instance, a few genes encoding hypothetical proteins were over-expressed (e.g. LEPBIa1691, LEPBIa0249, among ~25 others), whereas most were downshifted (LEPBIa2210, LEPBIa2180, among ~100 others) only in the wt strain, and not in ΔhemKR (Fig 5A). Also of note, 17 tRNA-encoding genes (spanning tRNAs that bind to 13 different amino acids) were intriguingly overexpressed, only when the HemKR system is present. Importantly, the absence of the HemKR system, abolished the wild-type iron-sensitive behavior of two operons that play key functions in heme homeostasis (hemACBLENG and exbB1/exbD1), which are repressed in the wt strain under iron-deficit (Fig 5A). Instead, in the ΔhemKR mutant exbB1/exbD1 was no longer repressed and, more dramatically, the heme-biosynthesis operon hemACBLENG was induced ~2.5-fold. This means that the ΔhemKR strain exhibited ~8-fold increase of hemACBLENG expression compared to the wt under iron deprivation (S2 Table), strikingly indicating a higher heme biosynthetic activity when the metal is restricted, that requires the HemKR system to be shut down.

All the evidence considered, a synergistic effect between HemKR and some other regulatory systems, facilitate an adaptive heme-homeostatic response when iron is scarce.