| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 The Philips Institute, Virginia Commonwealth University, Richmond, VA, USA
2 Department of Microbiology and Immunology, Virginia Commonwealth University, Richmond, VA, USA
3 Department of Molecular Biology, University of Gdansk, Gdansk, Poland
4 University of San Luis, San Luis, Argentina
Correspondence
Janina P. Lewis
jplewis{at}vcu.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Supplementary data are available with the online version of this paper.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Rao et al., 1999; Schifferle et al., 1996). In addition, due to its abundance in the host, haemin serves as a valuable source of iron for most host-associated micro-organisms (Genco & Dixon, 2001). Haemin is not freely available in the human host; it exists bound with host scavenging proteins, forming complexes such as the haemoglobin–haptoglobin complexes (Hwang & Greer, 1980), haemin–haemopexin complexes (Hrkal et al., 1974) and haemin–albumin complexes (Beaven et al., 1974). Therefore bacteria must be able to acquire the haemin from the host proteins either by proteolytic degradation and/or by expression of high-affinity binding receptors capable of wresting the haemin from the host haemin-binding proteins (Lee, 1995; Olczak et al., 2005). Since haemin is too large to freely diffuse through the bacterial membranes, bacteria have also developed mechanisms to transport the haemin across the membrane(s). Several reports show that the genes encoding the haemin-uptake systems are organized in operons (Ochsner et al., 2000; Stojiljkovic & Hantke, 1994; Thompson et al., 1999; Vanderpool & Armstrong, 2001; Wyckoff et al., 1998), indicating that such organization of these uptake systems is a common theme.
Arranging genes into operons might seem the perfect way to ensure regulation of all genes by a specific regulator as well as ensuring the expression of equal amounts of each protein. However, although most of the bacterial protein-dependent transport systems are encoded by genes organized in operons, the stoichiometry of the products of the operon is not always 1 : 1 : 1. Differential expression within operons encoding various bacterial transport systems has been reported (Ehrmann et al., 1998; Khun et al., 2000; Ronpirin et al., 2001). This differential expression resulted in overexpression of promoter-proximal genes and was the result of either premature transcript termination or differential transcript stability. The importance of the stoichiometry of components involved in nutrient transport has been demonstrated by the Escherichia coli ferrichrome uptake system encoded by the fhuACDB operon (Mademidis & Koster, 1998). The overexpression of the outer-membrane receptor, FhuA, led to increased uptake of ferric hydroxamates, but no improved uptake was observed when the inner-membrane transport components, FhuCDB, were overproduced. However, the stoichiometry of the components encoded by the fhuACDB operon as well as the mechanism to achieve the stoichiometry remain unknown. Also, despite the well-documented clustering of the genes of the haemin-uptake systems, not much is known regarding expression of the operons or possible differential expression of individual genes within each.
Porphyromonas gingivalis, a Gram-negative anaerobic bacterium, is implicated as a major aetiological agent in adult-onset periodontal disease (Slots et al., 1986). This disease is a chronic inflammatory condition of the supporting tissues of the teeth affecting 10–15 % of adults (Fox, 1992). P. gingivalis lacks the complete protoporphyrin IX (PPIX) biosynthetic pathway and thus requires exogenous haemin for growth (Schifferle et al., 1996; Nelson et al., 2003). Three multigenic clusters encoding proteins thought to be involved in the haemin-acquisition pathway have been detected in the genome of P. gingivalis W83 (Nelson et al., 2003). The first locus, ihtABCDE, is composed of five ORFs encoding a TonB-dependent outer-membrane receptor (IhtA), lipoprotein (IhtB), periplasmic binding protein (IhtC), permease (IhtD) and cytoplasmic ATP-binding protein (IhtE) (Dashper et al., 2000). The second locus has a similar composition in that a TonB-dependent receptor Tla (also known as Tlr) is followed by a putative ATP-binding cassette haemin-transport system, htrABCD (Aduse-Opoku et al., 1997; Slakeski et al., 2000). Direct evidence for the role of the locus in haemin uptake was obtained using a Tla-deficient mutant. This mutant was significantly impaired in its growth on low levels of haemin. Our studies focused on the third locus, hmu. A portion of the locus was first reported in P. gingivalis 53977 by Karunakaran et al. (1997). This group demonstrated the presence of a haemin-repressed gene encoding a TonB-dependent outer-membrane protein, HemR. Upstream of hemR, a 429 bp ORF (orf1) was located. The two genes were independently transcribed, and while hemR was repressed by haemin, orf1 was upregulated in the presence of haemin. Using a different P. gingivalis strain, A7436, Simpson et al. (2000) demonstrated the presence of a homologue of this locus consisting of hmuY (99 % identical to orf1) and hmuR (identical at the 5' end to hemR but differing significantly at the 3' end). hmuY and hmuR are co-transcribed and repressed by iron. These potentially contradictory data provided valuable insight into the complexity of gene regulation within the locus. First, they demonstrated that genes within the locus were co-transcribed. However, the presence of multiple transcripts indicated that differential expression may take place within the locus. Ultimately they left the transcriptional organization and regulation of this system by iron and haemin in doubt.
The study reported here demonstrates that the hmuY variant in P. gingivalis W83 is much larger than the previously characterized variants and encodes a 23 kDa iron-repressible outer-membrane protein. Furthermore, the genomic locus is also larger than previously reported and is composed of six genes, hmuYRSTUV, encoding proteins potentially involved in the haemin acquisition process. We determined the transcriptional organization and regulation of the hmuYRSTUV locus, and examined its role in haemin uptake.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Porphyromonas gingivalis was maintained on blood agar plates (TSA II, 5 % sheep blood; BBL) in an anaerobic atmosphere composed of 80 % N2, 10 % H2, 10 % CO2, at 37 °C. Broth cultures of P. gingivalis were prepared using Brain Heart Infusion Broth (BHI) supplemented with haemin (5 µg ml–1; Sigma) and menandione (1 µg ml–1; Sigma) (enriched BHI). Escherichia coli strain TOP10 (Invitrogen) was used as the host for plasmid manipulations and was maintained on Luria–Bertani medium. Antibiotics (ampicillin, 100 µg ml–1, kanamycin, 50 µg ml–1) were used to select for E. coli transformants.
|
; Rishi et al., 2004). To deplete iron from the culture medium and the bacterial cells, high concentrations of the ferrous iron chelator 2,2'-dipyridyl (DP; Sigma) were used.
(i) For protein analysis.
An actively growing culture was used to start two 10 ml cultures in BHI supplemented with menandione (mBHI). One was grown in the presence of ferric chloride (200 µM) (to create iron-replete conditions) and the other was supplemented with 150 µM DP (to create iron-chelated conditions). Both cultures were incubated overnight. Cells were harvested and stored at –80 °C. Supernatant was used to prepare extracellular vesicles as described previously (Lewis & Macrina, 1998).
(ii) For RNA analysis.
Two flasks of 200 ml mBHI were inoculated with 5 ml of overnight culture. These cultures were grown until they reached mid-exponential phase (OD660 0.5). To one culture, ferric chloride (200 µM) was added to generate iron-replete conditions; the other was supplemented with 150 µM DP to create iron-chelated conditions. Cells were harvested after 1 h bacterial growth and stored at –80 °C.
Preparation of bacterial cells for analysis of the effect of iron and haemin on expression of hmuYR.
To determine the effect of various iron/haemin sources on expression of the hmu locus we used medium that has low iron content (1 µM) (Dashper et al., 2004). The remaining iron present in this medium was depleted by the addition of a low concentration of DP (final concentration 10 µM). P. gingivalis cells were depleted of haemin by four serial passages in Mycoplasma broth (Becton Dickinson) supplemented with menandione. This culture was then used to inoculate Mycoplasma broth and Mycoplasma broth supplemented with bovine haemin (1.5 µM), bovine haemoglobin (1.5 µM), ferric chloride (200 µM) and PPIX (1.5 µM). DP was included in all samples at 10 µM to chelate any free iron present in Mycoplasma broth. Samples were collected after 0, 1, 4 and 24 h of growth. Bacterial cells were collected by centrifugation (12 000 r.p.m., 2 min, 4 °C) and stored at –80 °C prior to RNA isolation.
Cloning and purification of recombinant HmuY.
We expressed a portion of the HmuY protein in E. coli using the T7 expression system (Novagen). We PCR-amplified a region (from bp 23 to bp 594) of the hmuY coding region using forward (23F) and reverse (23R) primers (see Supplementary Table S1, available with the online version of this paper). The PCR product was cloned into the pET22b expression vector (Novagen). The cloned fragment was rendered in-frame with six histidines encoded by the expression vector. This construct was introduced into E. coli BL21(DE3) cells and the expression of hmuY was induced with IPTG. Recombinant protein was purified on a nickel affinity column under denaturing conditions according to the manufacturer's instructions (Qiagen). Purity of this protein was assessed by SDS-PAGE (Laemmli, 1970). The remaining protein was dialysed against phosphate-buffered saline, pH 7.3, and stored at –80 °C.
Haemin-binding assay of HmuY.
Haemin binding to recombinant HmuY was examined using the peroxidase activity of haemoprotein (Francis & Becker, 1984). Thus 10 µl (0.5 mg ml–1) recombinant HmuY was incubated with 1 µl haemin solution (10 mM) at room temperature for 10 min. The LDS-PAGE gel containing the resolved protein was then washed twice for 30 min in TBS buffer (10 mM Tris/HCl, pH 7.5, 0.9 % NaCl) and then developed with 4-methoxy-1-naphthol in the presence of H2O2. To detect total proteins present in the gel it was stained with Coomassie blue. BSA, demonstrated to bind haemin, was used as a positive control and trypsin was used as a negative control. Bovine haemoglobin served as a control for the haemin detection method.
Antibody production to recombinant HmuY.
Polyclonal antibody to HmuY was generated in rabbits by Proteintech Group. Briefly, rabbits were immunized with 200 µg of the HmuY recombinant protein in the presence of complete Freund's adjuvant. The animals were boosted twice at 14 day intervals with 100 µg of the recombinant HmuY (mixed with Incomplete Freund's adjuvant) and 2 weeks later blood was collected. Sera were prepared and polyclonal HmuY antibody was affinity-purified using recombinant HmuY immobilized on a nitrocellulose membrane as described previously (Sexton et al., 2004) to remove background present when neat antiserum was used.
Western blot analysis.
Proteins from various P. gingivalis cell fractions – whole-cell lysates, outer membranes (prepared as described by Dashper et al., 2000) and outer-membrane vesicles (OMV) (Lewis & Macrina, 1998) – were solubilized in Laemmli buffer and separated by SDS-PAGE (Laemmli, 1970). The proteins were then transferred onto nitrocellulose membranes according to Towbin et al. (1992). The membrane was blocked for 40 min at room temperature in Tris-buffered saline (TBS)+1 % (w/v) BSA, and incubated overnight in TBS containing affinity-purified anti-HmuY serum at 1 : 500 dilution. After washing (three times, 10 min each) in TBS+0.05 % (v/v) Tween 20, the blot was incubated with a 1 : 2000 dilution of secondary antibody (anti-rabbit antibodies conjugated to horseradish peroxidase, Promega). After three washes, proteins were visualized with 4-chloro-1-naphthol in the presence of hydrogen peroxide.
Isolation of RNA.
Total cellular RNA was isolated from P. gingivalis W83 cultures using an RNeasy mini kit according to the manufacturer's (Qiagen) instructions. Contaminating DNA was removed by treatment of the preparations with RNase-free DNase (Qiagen). For RT-PCR, any remaining DNA was removed by additional treatment with RNase-free DNase (Ambion) at 37 °C for 1 h. Total RNA preparations were stored at –80 °C.
Northern blot hybridization.
Eight micrograms of total RNA was electrophoretically separated and transferred to positively charged nylon membranes (Boehringer Mannheim) using commercially prepared buffers (Ambion). The integrity of the transferred RNA was assessed by visualization of the 16S and 23S RNA using UV light. Hybridizations were performed using specific probes labelled with peroxidase as instructed by the manufacturer (ECL kit; Amersham). Hybridizing bands were visualized using blue-light-sensitive film (Eastman Kodak).
Relative RT-PCR.
Relative RT-PCR was performed in two steps. First, cDNA was synthesized with the SuperScript RNase H– reverse transcriptase (Invitrogen) and reverse gene-specific primers (Table S1) (50 pmol each) according to the manufacturer's instructions. Three hundred nanograms of RNA isolated from bacteria grown in iron-chelated conditions served as a template. The synthesis of cDNA was carried out for 60 min at 42 °C. The synthesis of cDNA through regions containing predicted stem–loop structures (hmuYR and hmuRS) was accomplished with thermostable reverse transcriptase (Thermoscript, Invitrogen) and higher temperature (60 °C). The cDNAs were then used for PCR amplifications with Platinum PCR Supermix HighFidelity mixtures (Invitrogen). Reactions were prepared according to the manufacturer's instructions with gene-specific primers. To determine the co-transcription within the hmu locus, the following primer sets were used: 1619 and 263 (for hmuYR), 267 and 265 (for hmuRS), 266 and 268 (for hmuST), and 270 and 269 (for hmuTUV) (Fig. 3a, Table S1). The thermal profile consisted of: one cycle at 95 °C for 3 min, 35 cycles of 95 °C for 45 s, 53 °C for 30 s and 72 °C for 2 min, and an extension step of 72 °C for 7 min. To demonstrate transcription and co-transcription within the hmuSTUV portion of the locus, two rounds of identical PCR were performed (using 1 : 10 dilution of the products from primary PCR as a template for the secondary PCR).
|
Nested PCR was done to amplify the circularized single-stranded cDNA. The ligated cDNA was amplified by primary PCR using primers MSR2 and MSF7 (Table S1, Fig. 4c). Thermocycling conditions were as follows: preincubation at 94 °C for 5 min, followed by 30 cycles of 94 °C for 30 s, 51 °C for 30 s, and 72 °C for 30 s, followed by an extension at 72 °C for 7 min. Secondary PCR was done using primary PCR product diluted 1 : 10 as a template and primers MSR6 and MSF2 (Table S1, Fig. 4c). The thermocycling conditions were as described above. Presence of amplicons was determined using agarose gel electrophoresis. PCR products were extracted from gels using the Qiagen Gel Extraction kit and subjected to sequencing.
|
Real-time, quantitative RT-PCR.
Quantitative TaqMan one step RT-PCR analysis was done with the TaqMan One Step PCR Master Mix Kit (Applied Biosystems). DNA-free total RNA isolated from bacterial cultures grown in iron-replete, iron-chelated and iron-limited conditions served as a template. Probes and primers (Table S1) were designed using Primer Express, version 2.0. The probes were labelled in the 5' end with the fluorescent reporter FAM (6-carboxyfluorescein) and in the 3' end with the fluorescent quencher TAMRA (6-carboxytetramethylrhodamine). Calibration curve plots for relative PCR were generated using serial fivefold dilutions of 0.2 ng genomic DNA and 100 ng total RNA as templates. For transcript quantification, calibration curves for each gene of the hmu locus were constructed using serial fivefold dilutions of 0.02 pg gene-specific PCR products (prepared as described above). For quantification of hmuY- and hmuR-specific messages, serial fivefold dilutions of 0.02 ng in vitro-synthesized hmuY and hmuR transcripts were also used (information regarding the generation of the in vitro-synthesized RNA is available as supplementary data with the online version of this paper). The experimental samples were tested in triplicate under the conditions recommended by the manufacturer using 20 ng total RNA (except for 16S rRNA, where we used 0.02 ng total RNA). Reverse transcriptase was omitted in samples serving as negative controls. The experiments were performed in the ABI Prism 7900 Sequence Detection System (Applied Biosystems) with the thermal profile consisting of one cycle of 30 min at 48 °C to synthesize cDNA; one cycle of 10 min at 95 °C; and 40 cycles each of 15 s at 95 °C and 1 min at 60 °C. The threshold cycle was determined to provide the optimal standard curve values (0.98 to 1.0). To control for the RNA quantity, a probe specific for 16S rRNA was used.
Construction of a mutant strain.
To construct an hmuY deletion mutant the entire ORF with its flanking sequences was PCR-amplified from W83 genomic DNA using 261 (forward) and 263 (reverse) primers (for primer positions and sequences see Table S1). The PCR product was subcloned into pCR2.1 (Invitrogen). A 437 bp ClaI–SspI fragment located between bp 127 and bp 618 of the hmuY gene was then replaced by the ermF-ermAM cassette (Fletcher et al., 1995). This recombinant plasmid was linearized with EcoRI, and electroporated into P. gingivalis W83. Double crossover between sequences flanking the resistance cassette and the wild-type gene rendered a mutant strain resistant to clindamycin. Southern blot analysis of clindamycin-resistant colonies was performed to determine the location of the ermF-ermAM cassette and to confirm the predicted structure of the allelic exchange product (results not shown). One mutant was selected for further analysis and was called V2794.
[55Fe]Haemin binding and uptake assay.
Binding and uptake of [55Fe]haemin was evaluated in P. gingivalis strains grown in haemin/iron-replete and haemin/iron-depleted conditions. Bacteria were depleted of haemin by serial passage in Mycoplasma broth. The haemin-depleted cells were inoculated into Mycoplasma broth supplemented with haemin and ferric chloride (iron-replete conditions) and in Mycoplasma broth containing 20 µM DP (iron-depleted conditions) and grown to mid-exponential phase. For the uptake assay to ensure growth of bacteria the assay was performed in Mycoplasma broth (basal medium) serving as a source of nutrients and at 37 °C. Cells from overnight cultures were harvested and suspended to an OD660 of 0.6 in haemin-free basal medium. The MultiScreen assay system (Millipore) was used for the uptake experiments. The filtration 96-well plate was prewetted with 100 µl basal medium, then 100 µl (5x107 cells) of bacterial cells was added to selected wells. [55Fe]Haemin [specific activity 0.26 mCi mg–1 (9.6 MBq mg–1)] was purchased from RI Consultants LLC. A stock solution (0.5 mg ml–1) was prepared in 0.1 M NaOH. Two working solutions, one containing only [55Fe]haemin (100 µg haemin ml–1 with specific activity 20 µCi [740 kBq] ml–1) and the other containing [55Fe]haemin (50 µg ml–1 with specific activity of 10 µCi [370 kBq] ml–1 mixed with an equal amount of unlabelled haemin (50 µg ml–1), were prepared. Ten microlitres of the working solutions was added to designated wells and samples were incubated for 20 min at 37 °C, anaerobically. As control for energy-independent binding and uptake, the assay was also performed at 4 °C. Liquid was then removed by vacuum, and cells were washed three times with 200 µl TBS supplemented with 0.1 % (v/v) Tween 20 (TBS-T). Filters were air-dried and collected in polyethylene scintillation vials (Fisherbrand; Fisher). Two millilitres of scintillation fluid was added to each vial and radioactivity was determined by scintillation counting with ‘wide open’ window setting, using a Beckman LS6500 scintillation counter. The active binding and uptake were determined by subtracting c.p.m. values obtained from experiments performed at 4 °C from those obtained at 37 °C.
Bioinformatics approaches.
To determine the presence and stability of secondary structures within the mRNA we used the Mfold server as described by Zuker (2003) (http://www.bioinfo.rpi.edu/applications/mfold/rna/form1.cgi).
Sequence similarity searches were done using the BLAST Program (NCBI, http://www.ncbi.nlm.nih.gov/). Comparison of the P. gingivalis hmu locus to regions similar in other bacteria was done using the Comprehensive Microbial Resource available at The Institute for Genomic Research (http://www.tigr.org/).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Since these vesicles are much easier to prepare compared with preparations of outer membranes, they were used in our initial studies. The SDS-PAGE analysis revealed drastic upregulation of a 23 kDa protein (Fig. 1a). Several attempts to determine the N-terminal sequence of this protein were unsuccessful. It was noteworthy that after incubation of the protein preparations at 37 °C in reducing conditions the size of the protein band upregulated in iron-deficient conditions was reduced to 21 kDa (data not shown). We determined the N-terminal sequence of the 21 kDa protein. This sequence, DEPNQ, was used to search the P. gingivalis W83 translated sequence (Nelson et al., 2003; available at http://www.tigr.org/) using the tBLASTN program. This sequence matched a single ORF, potentially encoding a protein with an approximate size of 23 kDa (Fig. 1c). This ORF was identical at its 3' end to the previously reported HmuY protein, but contained 74 codons 5' to the HmuY start codon, including an ATG start codon. Because the DEPNQ sequence was located upstream from the reported HmuY sequence, this suggests that the upstream ATG is the true start codon for HmuY.
|
; Lewis et al., 1999). This hypothesis was verified by Western blot analysis of OMV derived from the parental strain (W83) and a Kgp-deficient strain (V2577) (Lewis & Macrina, 1998) using anti-HmuY antibody. For proteolytic processing of the proteins, the vesicles were incubated for 1 h at 37 °C prior to gel electrophoresis. A 23 kDa band was observed in vesicles from the Kgp-deficient strain while this band was missing in the vesicles from the parental strain; instead the parental strain displayed a 21 kDa band (results not shown).
To confirm the identity of the 23 kDa protein, its cellular location, and its upregulation in iron-depleted media, we performed Western blot analysis using affinity-purified polyclonal antisera raised against a recombinant form of the protein, constructed from the longer version of the HmuY ORF (Fig. 1b
). A more intense 23 kDa protein band was detected in the lane containing proteins derived from P. gingivalis grown in iron-depleted conditions (Fig. 1b, lane 3) compared with the lane containing proteins from bacteria grown in iron-replete conditions (Fig. 1b, lane 2). This confirms that HmuY is the 23 kDa iron-regulated protein seen in Fig. 1(a). Also, only one protein 23 kDa in size was detected in P. gingivalis, which is in agreement with the size predicted from the HmuY protein sequence. Lastly, to confirm that HmuY is located in the outer membrane, we analysed protein preparations from outer membranes and compared them to those of whole-cell lysates and vesicles. An iron-repressed 23 kDa protein was detected in vesicles, whole-cell lysates and outer membranes (Fig. 1b, lanes 2, 3, 4, 5, 6, 7), thus further supporting the location of the protein in the outer membrane of P. gingivalis.
Analysis of hmuYRSTUV indicates that it is a novel haemin-uptake locus
Our study demonstrates that the HmuY protein is larger by the N-terminal 74 aa (Fig. 1c, underlined sequence) than its previously reported variants (Karunakaran et al., 1997; Simpson et al., 2000). Also, the originally described HmuY led to incorrect annotation of the hmuY gene on the genome of P. gingivalis W83 at both Oralgen (http://www.oralgen.lanl.gov) and TIGR (http://www.tigr.org). Further analyses revealed that the N-terminus of the 23 kDa protein (HmuY) contained the typical prokaryotic membrane lipoprotein attachment site: vsLTSC (aa 16–21; see Fig. 1c, Box 1). In addition, the N-terminal sequence upstream of the lipoprotein motif is typical for a prokaryotic leader sequence; the positively charged N-terminus (Met followed by two Lys residues) is followed by a region of hydrophobic residues (typical for an intermembrane region). The acylation of the protein and presence of a leader sequence would be consistent with the outer-membrane location of HmuY and thus are further evidence that the original HmuY annotation was mistaken.
Analysis of the sequence surrounding hmuYR in the genome of P. gingivalis W83 revealed that there is another portion of the locus, which we designated hmuSTUV, that may have been missed in previous reports as being part of the same operon (Karunakaran et al., 1997; Simpson et al., 2000) (Fig. 2). These genes within this portion of the locus are separated by short intergenic regions which lack promoter signature sequences, consistent with the possibility that they are transcribed from a common promoter located upstream of hmuY. The last gene of the putative operon, hmuV, is separated from its downstream gene by 937 bp. This gene, gcvT, encodes an aminomethyltransferase and is transcribed in the opposite orientation, indicating that it does not belong to the hmu locus.
|
), the structural composition of the proteins encoded by the hmu locus is novel for a haemin-uptake locus.
Prevalence and regulation of the hmu locus
To further assess the significance of our study we examined the presence of the locus in commonly used P. gingivalis laboratory strains and in various clinical isolates. Using PCR amplification with primers 261 (–576 to –555 bp upstream of hmuY) and 23R (594–573 bp of hmuY), designed to amplify the hmuY gene with the upstream region, we detected the presence of the locus in the genomes of all but one (strain 381) of the P. gingivalis strains listed in Table 1. These data indicate that the hmu locus is very common among P. gingivalis strains. The 1.2 kb PCR products obtained using genomic DNA from the various strains as a template were sequenced and the sequences were aligned with that from P. gingivalis W83. Notably, the additional 1 bp present after bp 105 in the previously reported variant of hmuY, orf1 (Karunakaran et al., 1997), was missing in all strains examined, in agreement with the sequence reported in this study (see Fig. S1, available as supplementary data with the online version of this paper). These data suggest that the longer version of hmuY encoding a 23 kDa protein described in our study is found in most P. gingivalis strains and that the previously reported shorter versions of the gene resulted from rare mutations or sequencing errors.
We also searched other bacterial genomes for the presence of a locus similar to P. gingivalis hmu. Interestingly, similar loci were found in other anaerobic bacteria: Bacteroides fragilis, Bacteroides thetaiotaomicron and Prevotella intermedia (Fig. 2). These loci encode proteins that exhibited significant identity and similarity through their entire lengths (with the exception of HmuT). Importantly, growth of these bacteria is also dependent on exogenously provided haemin and it is possible the hmu locus plays a role in haemin acquisition in these bacteria as well. Our studies indicate that, similarly to the P. gingivalis hmu locus, the Prevotella intermedia hmu locus is also drastically upregulated in iron-depleted conditions (data not shown), suggesting that a similar regulatory mechanism governs the gene expression within the loci.
The fact that most bacterial haemin-uptake encoding genes are organized in operons led us to study the possibility of the hmu genes having an operonic organization as well. Furthermore, the previously reported putative regulatory sequences (Fur-binding sites; Karunakaran et al., 1997) are within the coding region of hmuY and as such may not play a role in regulation of the locus. Therefore the regulatory mechanism of the locus remains to be determined. To gain an insight into possible regulatory mechanisms, we examined the role of iron and haemin in regulation of the hmu locus and determined the transcriptional start site. Then we examined the possibility of differential gene expression within the locus. Finally, we determined the role of the locus in haemin uptake and the biological significance of the of the iron regulation.
Secondary structures may form within hmuYRSTUV mRNA
Secondary structures may form within operons and contribute to differential gene expression within the operons by either impeding the transcription process or stabilizing the upstream mRNA (Newbury et al., 1987a). To investigate the possibility of differential gene expression within the hmu locus, we analysed the predicted mRNA for the presence of secondary structures, such as stem–loops. First, we analysed the intergenic sequences for the presence of secondary structures. No secondary structures were detected. Then, we analysed the coding regions of hmuR and hmuS for the presence of repeated sequences capable of forming secondary structures. Since local structures (formed within relatively short sequences) are more likely to form in vivo we analysed the sequences progressively using 60 bp regions. Notably, the 5' region of the hmuR gene (complementary to bp 18–43) contained a 10 bp inverted repeat (Fig. 3a). Secondary structure analysis and computations of that region using Mfold (Zuker, 2003) have shown that the sequence has the potential to form a stem–loop structure with predicted stability equivalent to free energy of dissociation of –16.9 kcal mol–1 (–70.7 kJ mol–1) (Fig. 3a). A second inverted repeat capable of forming a less stable structure (–7.7 kcal mol–1; –32.2 kJ mol–1) was detected adjacent to it (bp 80–98) (not shown in Fig. 3a). A third repeat capable of forming an equally stable stem–loop structure (–17.8 kcal mol–1; –74.5 kJ mol–1) was detected in the middle of the hmuR coding region (bp 899–934) (Fig. 3a). Finally, a fourth 13 bp inverted repeat was detected within 5' coding region of the hmuS gene (bp 24–54). The computed stability of this region was –20.8 kcal mol–1 (–87.0 kJ mol–1) (Fig. 3a). The presence of the stem–loop structures within the hmu locus suggests that differential expression may take place within the locus.
Genes within the hmu locus are co-transcribed
There are six genes within the hmu locus encoding potential haemin-acquisition proteins (Fig. 3a). The structure of the locus indicates that genes within the locus are co-transcribed and suggests the presence of a transcript approximately 9 kb in size that would be expected in the absence of differentially abundant messages. The first approach we used to detect an hmu-specific transcript was Northern blot analysis. To test the iron-dependent transcription of the hmu locus, we performed analyses on total RNA isolated from P. gingivalis W83 grown in iron-replete and iron-chelated conditions. We detected hmuY- and hmuR-specific transcripts that were more abundant in iron-chelated conditions (Fig. 3b). However, no hybridization signals were obtained using probes specific for the hmuS, hmuT, hmuU and hmuV genes with RNA from bacteria grown either in iron-replete or in iron-depleted conditions (data not shown).
The Northern blot analysis verified the co-transcription within the hmuYR sublocus. Unexpectedly we also detected a 1 kb transcript specific for hmuY. These data suggest that differential expression takes place within the operon and that hmuY mRNA is more prevalent than the hmuYR mRNA. The absence of the expected 9 kb transcript also indicated limitations of the Northern blot technique in detecting sparse messages that may be contiguous with upstream more abundant transcripts.
Since we were not able to detect messages for all genes within the locus, we performed a more sensitive assay: RT-PCR. The locations of primers used in this study are shown in Fig. 3(a); sequences of the primers are available as supplementary data (Table S1). The results are shown in Fig. 3(c). To confirm the co-transcription of the hmuYR locus, we performed RT-PCR using primers 1619 and 263. The PCR product of expected size confirmed the co-transcription of the two genes (Fig. 3c, lane 3). To test the co-transcription of hmuR with hmuS we used primers 267 and 265. Again, the PCR product of expected size was detected (Fig. 3c, lane 7). The co-transcription of hmuST was established by observation of an expected 1.1 kb PCR product using primers 266 and 288 (Fig. 3c, lane 11). Lastly, the co-transcription of hmuTUV was demonstrated by a 1 kb PCR product obtained using primers 270 and 269 (Fig. 3c, lane 16). Interestingly, the PCR products through the hmuYR and through the hmuRS regions were obtained only when cDNA synthesized at a higher temperature (60 °C vs 42 °C) was used as a template (Fig. 2c, lanes 3 and 7).
To further examine the operonic organization and the possibility of the presence of internal promoters we constructed the isogenic mutant V2794, containing a deletion of most of the first gene of the locus, hmuY. Since the antibiotic-resistance cassette used in this study (Fletcher et al., 1995) contains a transcriptional terminator, this lesion would be expected to have a polar effect if the genes within the locus are co-transcribed. Thus, we quantified the transcript level for all genes present in the locus using total RNA isolated from the parental and mutant strains by real-time RT-PCR. While messages for all genes were present in the parental strain, the hmu-specific messages were barely detectable in the mutant strain (results not shown), indicating that the mutation has a polar effect on the genes downstream as well as further confirming RT-PCR data indicating that the genes within the locus are transcribed as a single unit. In addition, this experiment suggests that there is no internal promoter within the operon.
The hmuY promoter is located 292 bp upstream of the translational start site
hmuY is separated from an upstream gene prtC by an intergenic region of 855 bp. Although prtC is transcribed in the same orientation as hmuY, it likely is not a part of the hmu operon. This is based on the size of the transcripts obtained from our Northern blot analysis, which are too short to contain prtC- and hmuYR-specific transcripts. Thus, we searched for an hmuY promoter within the intergenic sequence. The location of the hmuY 5' end mRNA was determined using 5' RACE (Fig. 4). Analysis of 5' RACE products from primary PCR revealed two bands with approximate sizes of 70 bp and 290 bp (Fig. 4a, lane 2). No amplification products were detected in reactions in which templates where reverse transcriptase was omitted during the cDNA synthesis step were used (Fig. 4a, lane 3). The secondary PCR produced two bands with approximate sizes of 180 bp and 100 bp (Fig. 4b, lane 2). No amplification product was obtained using the primary PCR from the negative control as a template (Fig. 4b, lane 3). Since the primers used for the secondary PCR analysis should result in a product smaller by approximately 100 bp than that from the primary PCR the 180 bp PCR product was of expected size. Thus, DNA from the gel slice containing the 180 bp fragment was extracted and its sequence determined. Alignment of the sequence to a sequence upstream of hmuY demonstrated ligated product and thus presence of a transcriptional start site at –292 bp (G) upstream of the hmuY start site (Fig. 4c).
Analysis of sequence upstream of the transcriptional start site identified sequences similar to the typical E. coli RNA polymerase 
70 consensus promoter sequences: centred at –35, TTGACA, and at –10, TATAAT, relative to the transcription initiation site (Hawley & McClure, 1983) (Fig. 4c). Interestingly, we also located a 24 bp sequence capable of forming an inverted repeat structure that was positioned 148 bp upstream of the transcriptional start site (Fig. 4c). Lastly, we found a putative ribosome-binding site with the sequence AAGGT located 9 bp upstream of the translational initiation codon, ATG (Fig. 4c).
Genes within the hmu operon are iron repressed
In view of the contradictory reports regarding the effect of iron and haemin on expression of the hmuY and hmuR genes (Karunakaran et al., 1997; Simpson et al., 2000), we examined the effect of iron, haemin, haemoglobin and PPIX on expression of both those genes by real-time RT-PCR. Data from 60 min incubation are shown in Table 2. A 31-fold reduction of hmuY-specific transcript was observed after addition of ferric chloride, and a 35-fold reduction of the hmuR-specific transcript was noted. Similar results were observed after longer times of incubation (up to 24 h; results not shown). Thus, we conclude that the hmu locus is repressed by iron.
|
). No major changes in transcript levels were noted after addition of PPIX to the haemin-starved cells, indicating that the increased gene expression was haemin-specific (Table 2).
Genes within the hmu operon are differentially expressed
Our Northern blot analysis indicated differential expression within the operon. Thus we determined the amount of transcript for each gene of the hmu operon in the presence and absence of added iron by real-time RT-PCR. Several levels of differential expression were noted (Table 3). First, the genes encoding the outer-membrane proteins were overexpressed when compared to the full-length transcript. The hmuYR mRNA was expressed at approximately 25- to 50-fold higher level than the full-length transcript (Table 3). The overexpression of hmuYR transcript was observed using RNA from cells grown with and without iron (generated by addition of DP). Second, the hmuYRSTU-specific transcript was 20- to 30-fold more prevalent than the full-length transcript hmuYRSTUV (Table 3). Lastly, differential expression within the hmuYR suboperon encoding the outer-membrane proteins was observed. The expression of hmuY-specific transcript was approximately 7- to 20-fold higher than the hmuR mRNA (Table 3). Similar results were obtained using in vitro-synthesized hmuY- and hmuR-specific transcripts for the generation of calibration curves (see Table S2, available as supplementary data with the online version of this paper).
|
; Olczak et al., 2001). Thus we examined the ability of HmuY to bind haemin. For these studies recombinant HmuY protein was used. This protein was purified to homogeneity (Fig. 5a) and then incubated with haemin (Fig. 5b, c). We noted that HmuY heated and resolved under denaturing conditions forms monomers (Fig. 5a). However, when not heated and run under native conditions this protein formed oligomers in addition to the monomeric form (Fig. 5b). Notably, the various forms of HmuY bound haemin (Fig. 5c, lane 2). In our assays haemin also bound to BSA; however, no haemin-binding activity was detected using trypsin, indicating that the haemin binding to HmuY was specific (Fig. 5c, lanes 4 and 5).
|
Haemin binding was determined using cells that were incubated on ice. Inhibition of [55Fe]haemin binding in the presence of equal amounts of unlabelled haemin demonstrated that the binding examined in this study was specific (results not shown). As shown in Fig. 6(a), significant haemin binding occurred in iron-replete conditions; however, no significant difference in the amount of bound haemin between the parental and hmu-deficient strains could be detected, indicating that this binding was not dependent on the HmuYR receptor and confirming the presence of other haemin-binding components in P. gingivalis. It is noteworthy that in iron-depleted conditions the amount of bound haemin detected in the parental strain was twofold elevated compared to that observed in the parental strain grown under iron-replete conditions and to that detected in the mutant strain, indicating that the HmuYR receptor plays a role in haemin binding (Fig. 6a). Also, the observation that the parental strain accumulated more haemin in iron-depleted conditions is consistent with the upregulation of the locus in those conditions.
|
the haemin taken up by the parental strain grown in iron-depleted conditions was fourfold higher than that detected in the strain grown in iron-replete conditions. Furthermore, the haemin uptake in iron-depleted conditions was also approximately fourfold reduced in the mutant strain compared to the parental strain. No haemin uptake by the mutant strain was detected in iron-replete conditions. The inhibition of [55Fe]haemin uptake in the presence of equal amounts of unlabelled haemin demonstrated that the uptake examined in this study was specific (results not shown). These data indicate that the hmu locus plays a role in haemin uptake and also are consistent with the upregulation of the locus in iron-depleted conditions. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
)] the HmuSTUV proteins may be involved in processing and transport of the haemin molecule across the inner membrane. However, this functional assignment will need to be experimentally verified.
Although several multigenic loci encoding haemin-uptake proteins are present in P. gingivalis, not much is known regarding the transcriptional organization and expression within the loci. Such knowledge has implications when examining the operonic organization of the loci using Northern blot analysis and is also important when constructing mutant strains using antibiotic-resistance cassettes whose insertion results in termination of the transcript. Also, it may give an insight into the stoichiometry of the products encoded by the operon. The genetic structure of the hmu locus suggested that the genes within the locus are co-transcribed. Our studies demonstrate that indeed the genes within the hmu locus are transcribed as a single unit, and in agreement with that, insertion of the erm cassette containing a transcriptional terminator (Fletcher et al., 1995) into the first gene of the locus results in abrogation of all of the hmu transcripts located downstream of the insertion.
To gain a better understanding of the mechanism underlying regulation of the hmu operon, we examined the effect of various iron and haemin sources on expression of genes within the locus. We showed that expression of the genes is drastically elevated in iron-limited conditions. These data are in agreement with results reported by Simpson et al. (2000). However, no significant alterations in expression of hmuYR were observed when haemin was added to iron-depleted cultures. These data contradict findings of Liu et al. (2004) and Karunakaran et al. (1997), who reported that hmuR is down-regulated by haemin. Karunakaran et al. (1997) also reported that a variant of hmuY, orf1, was upregulated in the presence of haemin, whereas we show that hmuY is regulated in the same fashion as hmuR, which is consistent with the operonic structure of the locus. The difference between our data and those of Karunakaran et al. (1997) may result from a nucleotide sequence difference between the strains used for the studies. Analysis of sequence upstream of orf1 (Karunakaran et al., 1997) revealed a 1 bp insertion (cytidine) located after bp 105 of the hmuY sequence reported in this study (Fig. S1). That change would result in a frame-shift and thus a different gene product. However, since our sequencing data demonstrate the presence of the hmuY variant from P. gingivalis W83 in most P. gingivalis strains, we expect that our findings regarding the length and regulation of the gene can also be applied to the majority of P. gingivalis strains.
The different results regarding the effects of haemin on regulation of the locus might be due to the inclusion of DP (in low concentrations) in our assays. Our studies were done in Mycoplasma broth containing low levels of iron and thus the 10 µM DP that we used was sufficient to chelate the free iron present in the medium, as demonstrated by previous work (Bramanti & Holt, 1991; Dashper et al., 2004). In addition, previous studies demonstrated haemin-dependent regulation. It is probable that the regulation was mediated by iron that was released from haemin, and thus DP was included in our study to chelate any haemin-derived iron so as to examine the effect of haemin alone on expression of the locus as opposed to haemin and iron.
Our data unequivocally show iron-dependent repression of the hmu locus. The most common regulator of iron-dependent expression in bacteria is Fur (Andrews et al., 2003; McHugh et al., 2003). A fur homologue has been identified in P. gingivalis (GenBank accession no. AF171224; TIGR annotation PG0456) (Nelson et al., 2003; Olczak et al., 2005). Initial studies in Genco's laboratory have shown that P. gingivalis Fur can partially complement the functional activity of the E. coli Fur (Liu et al., 2006). Thus we analysed the sequence upstream of the hmuY transcriptional start site for a consensus Fur-binding site. However, no such sequences were identified. It is possible that the sequences recognized by P. gingivalis Fur differ from the consensus Fur box sequences. The absence of E. coli-like Fur-binding consensus sequences has been observed in other bacteria (Friedman & O'Brian, 2003; Palyada et al., 2004). It is also possible that the genes are regulated by a different mechanism (indirectly by Fur or independently of Fur). Another putative iron-dependent repressor is also encoded on the genome of P. gingivalis (PG1044). Finally, we identified an inverted repeat capable of forming secondary structure upstream of the transcriptional start site that may have a regulatory function.
The role of iron in regulation of haemin-uptake systems is well established; however, the transcript levels of individual genes within the multigenic transcription units encoding the components of the uptake systems remain largely unknown. Thus, it is noteworthy that we observed differential expression within the hmu locus, resulting in higher levels of transcript for promoter-proximal genes than for promoter-distal genes. Since gene expression often correlates with protein expression in bacteria we can hypothesize that the transcript ratio also extends to the ratio of proteins encoded by the locus. The larger amount of HmuY compared with other proteins encoded by the locus would thus explain our observation of only one protein (HmuY) detected to be drastically upregulated on our SDS-PAGE gel despite the fact that all genes within the locus are upregulated in the absence of iron. Explanations for the observed differences in gene expression within the hmuYRSTUV operon include differential transcript expression or stability. There is a precedent for differential expression within bacterial operons and usually it is a consequence of secondary structures (stem–loop structures) located within mRNA (Ehrmann et al., 1998; Khun et al., 2000; Newbury et al., 1987a, b; Ronpirin et al., 2001). Sequences capable of adopting the secondary structures are usually located in the intercistronic regions between the coding regions of genes within the operon. Interestingly, we identified inverted repeats with the potential to form stable stem–loop structures within the coding regions of the hmuR and hmuS genes. The first stem–loop (in the 5' end of hmuR) may account for the termination of transcription after hmuY, thus resulting in the 1 kb transcript (if the transcriptional start site was located approximately 400 bp upstream of the locus). Repeat sequences capable of forming a secondary structure with a slightly higher calculated energy of dissociation (–20.8 kcal mol–1; –87.0 kJ mol–1) are located in the 5' end of the hmuS coding region. The predicted stability of this structure is similar to that of one observed in another differentially regulated operon, the fbpABC locus of Neisseria meningitidis (Khun et al., 2000). The fact that the size of hmuY (with upstream sequences) is 1 kb and that of the hmuR gene is 1.9 kb would be consistent with the 3 kb transcript corresponding to hmuYR detected in our Northern blot analysis. We also detected a putative stem–loop structure in the middle of the hmuR transcript (bp 899–934). The location of this putative secondary structure could explain the presence of the 2 kb transcript in our Northern blot analysis. The higher temperatures needed for the cDNA synthesis through regions containing the sequences capable of forming the stem–loops imply that the secondary structures indeed form in the analysed RNA. However, further studies are required to definitely establish the role of the structures in the differences in transcript abundance observed within the hmu locus.
The differential expression within the hmu operon may have functional implications. We have shown that there is excess of hmuYR-specific message compared to the full-length hmuYRSTUV mRNA. HmuY and HmuR are outer-membrane proteins and as such would serve as haemin receptors. HmuY is a novel protein with no significant similarity to other characterized proteins. However, three peptides derived from a P. gingivalis 381 cell envelope haemin-binding protein identified by Kim et al. (1996) show 90 %, 90 % and 89 % identity, respectively, to internal regions of the protein characterized in this study, suggesting that these are the same proteins. Also, we have shown that HmuY binds haemin, and the haemin-binding property of HmuR has been demonstrated previously (Liu et al., 2006; Olczak et al., 2001). In agreement with the haemin-binding ability of the proteins we observed reduced haemin binding in the mutant devoid of the receptor. Also we show that the hmu locus is a major haemin transporter in P. gingivalis. Since haemin is not freely available within the human host, capture of this nutrient at the bacterial cell surface would be a rate-limiting step. Also, an increased concentration of haemin receptors would be beneficial to efficiently compete for haemin binding with host haemin-scavenging proteins such as
