EbfC (YbaB) Is a New Type of Bacterial Nucleoid-Associated Protein and a Global Regulator of Gene Expression in the Lyme Disease Spirochete

Abstract

Nearly every known species of Eubacteria encodes a homolog of the Borrelia burgdorferi EbfC DNA-binding protein. We now demonstrate that fluorescently tagged EbfC associates with B. burgdorferi nucleoids in vivo and that chromatin immunoprecipitation (ChIP) of wild-type EbfC showed it to bind in vivo to sites throughout the genome, two hallmarks of nucleoid-associated proteins. Comparative RNA sequencing (RNA-Seq) of a mutant B. burgdorferi strain that overexpresses EbfC indicated that approximately 4.5% of borrelial genes are significantly impacted by EbfC. The ebfC gene was highly expressed in rapidly growing bacteria, but ebfC mRNA was undetectable in stationary phase. Combined with previous data showing that EbfC induces bends in DNA, these results demonstrate that EbfC is a nucleoid-associated protein and lead to the hypothesis that B. burgdorferi utilizes cellular fluctuations in EbfC levels to globally control transcription of numerous genes. The ubiquity of EbfC proteins in Eubacteria suggests that these results apply to a wide range of pathogens and other bacteria.

INTRODUCTION

Bacteria, like all organisms, face a major task in packing enormous amounts of genetic material into a very small space. Organization of bacterial nucleoids is facilitated, in part, by DNA-binding proteins (23, 84, 90, 95). Although prokaryotic nucleoid-associated proteins have sometimes been called “histone-like,” they are functionally and genetically distinct from eukaryotic histone proteins. Bacterial nucleoid-associated proteins preferentially bind certain sequences of DNA that are spread throughout the genome. As might be expected, the positions of nucleoid-associated protein binding sites relative to transcriptional promoters can have profound effects upon gene expression (9, 11, 42, 45, 46, 48, 58, 61, 78, 82, 88). Such effects can be due to physical interactions between nucleoid-associated proteins and transcriptional components or induction of DNA conformations that alter binding affinities of RNA polymerase and/or transcriptional activators and repressors (7, 49, 75, 82). Bacteria are under pressure to optimize genome structure while also ensuring that locations of binding sites for nucleoid-associated proteins result in beneficial effects on gene expression. One may logically conclude that the positive and negative effects of nucleoid-associated proteins on transcription are advantageous to the organism.

The classic model bacterium, Escherichia coli, produces at least 12 different nucleoid-associated proteins to manage its single chromosome (3, 23, 46). In contrast, the Lyme disease spirochete Borrelia burgdorferi, which possesses the most complex known bacterial genome, with as many as 25 distinct replicons per cell (17, 32), encodes only 4 putative nucleoid-associated proteins: Hbb, Gac, Dps (which has also been referred to as NapA), and EbfC. Until the present work, none of these proteins had been examined for interactions with the borrelial nucleoid. Hbb shares some structural and functional similarities with the E. coli IHF and HU proteins and is involved with transcriptional regulation of at least one borrelial gene (53, 58, 62, 89). B. burgdorferi Gac appears to be unique to Borrelia spp., is the product of an internal translational start site within the gyrA gene, and exhibits some similarities to E. coli HU (51, 52). Dps is likely to be involved with DNA compaction during stationary phase, similarly to E. coli Dps (20, 44, 55).

The fourth probable nucleoid-associated protein, EbfC, is a unique type of DNA-binding protein (21, 56, 74). Nearly every eubacterial species carries a homolog of EbfC, which has also been referred to as “Orf-12,” “Orf-107,” or YbaB (2, 21, 31, 56, 74). Prior to the studies by our laboratory, very little was known about the functions of these proteins, although the 3-dimensional structures of the Haemophilus influenzae and E. coli orthologs had been solved in hopes that the structures might suggest functions (Protein Data Bank [PDB] entry 1PUG and reference 56). Our studies were the first to identify that the EbfC/YbaB family members are DNA-binding proteins (21, 74). EbfC homodimers form a structure that has been described as a pair of tweezers (56, 74). Site-directed mutagenesis-based studies defined the extended alpha-helical “tweezers” as the DNA-binding domain, with the space between the “tweezers” being appropriate to fit around double-stranded DNA (74). The EbfC/YbaB orthologs of E. coli and H. influenzae bind DNA, but their preferred binding sequences are distinct from that of B. burgdorferi EbfC (21). B. burgdorferi EbfC preferentially binds to a palindromic DNA sequence, 5′-GTnAC-3′, where “n” can be any nucleotide (74). The B. burgdorferi genome contains a consensus EbfC-binding site approximately every 1 kb (74). Binding by EbfC induces bending in DNA (74). The protein primarily exists as a homodimer but can also form higher-ordered structures such as tetramers and octamers, giving EbfC the potential to aggregate/bridge DNA (74). Altogether, the functional features of EbfC resemble those of many nucleoid-associated proteins (23, 74).

Because the Lyme disease spirochete contains numerous replicons but very few probable nucleoid-associated proteins, we have continued to examine this bacterium as a model of EbfC/YbaB function. In the present work, we investigated the associations between EbfC and the nucleoid in vivo. We previously demonstrated that EbfC binds to the operator region that controls transcription of B. burgdorferi erp operons and that EbfC functions as an antirepressor to stimulate erp transcription (5, 13, 48, 74). Since many nucleoid-associated proteins exert global effects on gene expression, comparative RNA sequencing of a mutant B. burgdorferi strain that overexpresses EbfC was performed. In addition, transcriptional linkage was defined between ebfC and dnaX, which encodes subunits of DNA polymerase. Both ebfC and dnaX mRNA levels were found to be dependent upon the metabolic status of the bacterium.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

All studies used derivatives of the B. burgdorferi type strain, B31, cultured at 34°C in Barbour-Stoenner-Kelly II medium (47, 97). Chromatin immunoprecipitation (ChIP) analyses and studies on the effects of growth phase on ebfC and dnaX expression levels were performed using an infectious clone of the sequenced B31 culture, B31-MI-16 (17, 32, 60). For studies that required B. burgdorferi transformed with recombinant plasmids, clone B31-e2 was used (86). When necessary for maintenance of recombinant plasmids, kanamycin was added to medium at a final concentration of 200 μg/ml. Retention of native plasmids was evident from RNA sequence and reverse transcription-PCR (RT-PCR) results (see below).

GFP-tagged EbfC and fluorescent microscopy.

Our previous studies indicated that the amino terminus of EbfC projects away from both the core of the protein and the DNA-binding domain and that addition of 47 extra residues to the amino terminus does not interfere with DNA binding (5, 21, 48, 74). Based on those data, we hypothesized that an amino-terminally tagged green fluorescent protein (GFP)::EbfC fusion will retain the ability to bind DNA. Thus, a chimera that consists of the gfp3 open reading frame (ORF) fused with the B. burgdorferi ebfC open reading frame was created (Fig. 1). An in-frame spacer, encoding a 13-amino-acid unstructured tether, between the two genes was included to separate the two GFP and EbfC structures. Recombinant plasmid pSZW53-4 replicates in both B. burgdorferi and E. coli and contains a constitutively expressed tetR and the anhydrotetracycline (ATc)-inducible promoter Post (48, 96). The gfp::ebfC fusion gene was cloned into the multiple cloning site of pSZW53-4 such that it was under the transcriptional control of Post, producing pBLS709. Clonal B. burgdorferi strain B31-e2 was transformed with pBLS709, resulting in strain BJ13.

Fig 1.

Schematic of the relevant regions of pBLS709, which carries an inducible gene encoding the GFP::EbfC fusion protein. The fusion protein consists of GFP3 and EbfC, linked by a 13-amino-acid flexible “tether,” underlined with green, red, and blue, respectively. The tether consists of the 33 bp immediately 5′ of the B. burgdorferi ebfC gene plus 6 additional bp that form an MluI cleavage site. Additional restriction endonuclease recognition sites were built into the chimera and readily facilitate replacement of key components for production of alternative fusion proteins. The −35, −10, and tetO sites of Post are indicated.

Two additional strains were used as controls. KS20, which carries pBLS599 and constitutively produces untagged GFP (6), was a control to determine whether GFP by itself interacts with the bacterial nucleoid. KS50 carries pSZW53-4 (48) and was examined in the presence of ATc inducer to control for possible effects of the tetracycline analog.

Cultures of all four strains were incubated in the presence of 0.2 μg/ml ATc to mid-exponential phase (approximately 2 × 10⁸ bacteria per ml). Cells were gently pelleted by centrifugation, washed twice with ice-cold 0.5× phosphate-buffered saline (PBS), and then resuspended in 0.5× PBS. Hoechst 33582 (Invitrogen) stain was added to a final concentration of 2 μg/ml, and the mixture was incubated in the dark for 10 min at room temperature, to allow uptake of the stain. Aliquots of bacterial suspensions were placed onto glass slides and allowed to air dry in the dark.

Slides were visualized on a Zeiss AxioImager Z.1 stand using a 100× plan approach, a 1.4 numerical aperture, and an oil-immersion objective. Images were digitally captured using an AxioCam MR system and software in high-resolution grayscale mode. All images were generated via pseudocoloring and overlaying images, when appropriate, in Photoshop C5. Any adjustments made to an image with respect to contrast and color balance were performed on the entire image.

ChIP.

Antibodies were generated against the EbfC-derived sequence MSSVKSNIDNIKKEM in New Zealand White rabbits by NeoPeptide (Cambridge, MA). The same polypeptide was used to affinity purify EbfC-specific antibodies from collected serum. Immunoblot analyses using these purified antibodies confirmed that they recognized EbfC but not any other B. burgdorferi protein.

ChIP was performed essentially as previously described (48, 92, 93), with the following modifications. B. burgdorferi B31-MI-16 was cultured at 34°C to mid-exponential phase (approximately 2 × 10⁸ bacteria/ml). Formaldehyde was added to a final concentration of 1%, and then the mixture was incubated for 8 min at room temperature with shaking. Cross-linking was stopped by addition of glycine to a final concentration of 0.3 M. Bacteria were pelleted by centrifugation and washed twice with cold Tris-buffered saline (20 mM Tris [pH 7.5], 150 mM NaCl). Cell pellets were stored frozen at −80°C. As needed, bacterial pellets were thawed on ice and then resuspended in a 1:4 ratio of lysis buffer to IP buffer (lysis buffer: 10 mM Tris [pH 7.5], 20% sucrose, 50 mM NaCl, 10 mM EDTA; IP buffer: 50 mM HEPES [pH 7.5], 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) (92, 93). Lysozyme was added to a final concentration of 5 mg/ml, and the mixture was incubated at 37°C for 30 min. To shear the bacterial DNA, lysates were sonicated using a Branson 102C sonicator (Branson Ultrasonics Danbury, CT), with 6 pulses of 10 s each at 10% amplitude. Cellular debris was cleared by centrifugation at 10,000 × g for 10 min at 4°C.

Anti-EbfC or control anti-IgG (GE Healthcare) antibodies were bound to resin particles using immunoprecipitation kit-protein A magnetic Dynabeads (Invitrogen, Carlsbad, CA), according to the manufacturer's recommended protocol. Dynabeads without added antibody were used in experiments as an additional control.

Sonicated borrelial supernatants (900 μl) were incubated with antibody-bead complexes, or control beads alone, for 20 min at room temperature, followed by a second incubation at 4°C overnight. Bead complexes were washed 3 times with IP wash buffer (Invitrogen protein A Dynabead kit) and then resuspended in IP buffer. After transfer to clean microcentrifuge tubes, bead complexes were washed 4 times with IP buffer supplemented with 500 mM NaCl, followed by 2 washes with Tris-EDTA (TE) buffer (10 mM Tris, 1 mM EDTA). Beads were resuspended in TE and incubated at 65°C for 18 h. Eluted DNA was purified using the DNeasy blood and tissue kits (Qiagen), and antigens were eluted from antibody using elution buffer (immunoprecipitation kit; Invitrogen). Following antigen elution, immunoprecipitation of EbfC was verified by immunoblot analysis using EbfC-specific antiserum.

Agarose gel electrophoresis indicated that immunoprecipitated DNAs primarily consisted of fragments 300 to 600 bp in size. Each sample preparation was treated with Tsp509-I (FastDigest; Fermentas) for 1 h and purified using th e Wizard PCR cleanup system (Promega). Digested ChIP fragments were ligated into the compatible EcoRI site of pUC19 and transformed into E. coli. Forty-three clones were selected at random, and their inserts were sequenced.

EbfC overexpression by B. burgdorferi.

Plasmid pBLS704 is a derivative of pSZW53-4 that contains wild-type ebfC under the transcriptional control of Post (48). Strain KS51 was produced from B. burgdorferi B31-e2 by transformation with pBLS704. Strain KS50, described above, served as a control. Both strains were cultured to early exponential phase (approximately 10⁶ bacteria per ml), and then ATc was added to a final concentration of 0.5 μg/ml. Cultures were then incubated at 34°C for 68 h, to mid-exponential phase (approximately 2 × 10⁸ bacteria per ml). Bacteria were harvested by centrifugation and stored at −80°C. To confirm that EbfC was overexpressed in strain KS51, lysates of each culture were subjected to immunoblotting using EbfC-specific antibodies and by quantitative reverse transcription-PCR (qRT-PCR) (59).

Bacterial RNA was purified as previously described (59), followed by a secondary purification and DNase treatment using the MasterPure RNA isolation kit (Epicentre). RNA purity and integrity were assessed using a Bioanalyzer nanochip for prokaryotes (Agilent). Samples that yielded RNA integrity numbers (RINs) above 9.0 were used for both RNA sequencing and qRT-PCR validation.

RNA sequencing (RNA-Seq).

High-throughput, flow cell, RNA Illumina sequencing was performed by AGCT Inc. (Wheeling, IL). Total RNA extracted from each strain and condition was analyzed at least once, with a minimum of 2.4 million reads per sample. Burrows-Wheeler Aligner (http://bio-bwa.sourceforge.net/bwa.shtml) was used to map the reads to the reference sequences under the pair-end configuration. The resulting SAM files were converted into BAM and subsequently sorted by samtools (http://samtools.sourceforge.net/). The mapping statistics for each BAM file were extracted by Picard tools (http://picard.sourceforge.net/). Pair-end reads were mapped to the B31 reference genomes separately with Tophat (http://tophat.cbcb.umd.edu/), a spliced read mapper based on Bowtie (http://bowtie-bio.sourceforge.net/index.shtml). The resulting alignment results were processed by Cufflinks (http://cufflinks.cbcb.umd.edu/) for transcript assembly and gene expression analysis. The list of transcripts and gene expression levels were expressed as fragments per kilobase of exon per million fragments mapped (FPKM), which is analogous to single-read reads per kilobase of exon per million fragments mapped (RPKM). Cufflink output, genes.fpkm.tracking, reports the transcript coordinates and expression levels expressed in FPKM for each sample; gene_exp.diff reports sample-based pairwise gene expression level comparison with statistical parameters fixed at P < 0.05. Numbers of reads per sample were normalized relative to the 5S, 16S, and 23S rRNA reads per sample.

qRT-PCR analyses of gene expression levels.

qRT-PCR was used to examine the validity of RNA-Seq results and to determine expression of ebfC and dnaX in cultured B. burgdorferi. The procedures were performed as previously described (59), using oligonucleotide PCR primers specific for each gene being examined. Results were compared with those for the constitutively expressed flagellar gene flaB. All qRT-PCRs were performed in quadruplicate, and runs were performed at least twice. Following each run, primer specificity and amplicon purity were assayed via melting-curve analysis (36, 59).

Characterization of the dnaX-ebfC operon.

Transcriptional linkage between dnaX and ebfC was determined by RT-PCR of total RNA purified from cultured strain B31-MI-16, as previously described (73). One oligonucleotide primer was complementary to a sequence in dnaX, and the other was complementary to a sequence in ebfC.

pBLS590 contains a promoterless gfp gene, preceded by a multiple cloning site, and replicates autonomously in both B. burgdorferi and E. coli (6). The 843 bp immediately 5′ of the dnaX open reading frame was PCR amplified such that BamHI and KpnI sites were added to the 5′ and 3′ ends, respectively, and then cloned into pBLS590. The resultant plasmid, pAAB200, was transformed into strain B31-e2. The insert of pAAB200 was truncated by overlap extension PCR (40) to produce pAAB203 and pAAB206 (see Fig. 5C). Similar methods were used to produce pAAB100, which contains the entire dnaX open reading frame 5′ of gfp and derivatives pAAB102 and pAAB104 (see Fig. 5D).

Fig 5.

Defining the dnaX-ebfC operon of B. burgdorferi. (A) Schematic of the dnaX-ebfC operon, shown to scale. Hatched arrows show orientations of dnaX, ebfC, and the genes immediately 5′ and 3′ of the dnaX-ebfC operon. Solid arrows indicate transcriptional start sites. Open arrows below the line indicate locations of sequences complementary to oligonucleotide primers used for RT-PCR linkage analyses. (B) Evidence that dnaX and ebfC are cotranscribed. Lane 1, RT-PCR using purified borrelial RNA as the template and oligonucleotide primers complementary to sequences found in dnaX and ebfC; lane 2, reaction mixture lacking reverse transcriptase; lane 3, reaction mixture without cDNA-producing random hexamer oligonucleotides; lane 4, reaction using purified B. burgdorferi genomic DNA as the template; lane 5, reaction using water in place of template. (C) Mapping the promoter 5′ of dnaX. At top are schematics of plasmids containing DNAs 5′ of dnaX transcriptionally fused to gfp. Below are flow cytometry data of GFP expression by B. burgdorferi carrying the fusion plasmids. Plasmids pAAB200 and pAAB206 contain a borrelial promoter that is absent from pAAB203. (D) Mapping the promoter within the dnaX open reading frame. At top are schematics of plasmids containing fragments of the dnaX ORF, all lacking the noncoding DNA 5′ of dnaX, fused to gfp. Below the diagrams are flow cytometry results of GFP production by B. burgdorferi carrying each fusion construct. Plasmids pAAB100 and pAAB104 contain a promoter, which is absent from pAAB102.

Mid-exponential-phase cultures of each transformed B. burgdorferi strain were assayed for GFP expression using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA), with excitation at 488 nm and detection at 530 nm. Each experiment involved measuring a minimum of 75,000 events. Each experiment was performed using at least two distinct cultures to ensure reproducibility.

The promoter within the dnaX open reading frame was determined by rapid amplification of 5′ cDNA ends (5′ RACE; Invitrogen) analysis of total RNA purified from B. burgdorferi that contained pAAB100. Oligonucleotides complementary to sequences within gfp were used for gene-specific primer extension. This approach was taken so that only transcripts originating within the dnaX open reading frame would be detected. Controls consisted of reaction mixtures that lacked reverse transcriptase, hexamer oligonucleotide cDNA primers, or template RNA. All primer extension reactions were repeated at least twice using separate B. burgdorferi cultures and RNA preparations to ensure reproducibility.

RESULTS

EbfC is a nucleoid-associated protein.

Spirochetes are much longer than most other bacteria, with Borrelia cells having lengths as great as 25 μm (41). In contrast, E. coli cells tend to be only 2 μm in length (54). A previous study of the relapsing fever spirochete Borrelia hermsii identified numerous nucleoids in each bacterium (50). Similarly, we observed multiple centers of dense DNA in the Lyme disease spirochete (Fig. 2).

Fig 2.

Colocalization of GFP-tagged EbfC with B. burgdorferi nucleoids. Epifluorescence microscopic images of B. burgdorferi nucleoids (blue, Hoechst 33582 DNA stain) and GFP. (A) Wild-type parental B. burgdorferi strain B31-e2. (B) Derivative KS20, which constitutively expresses soluble GFP. (C) Derivative KS50, which contains a recombinant plasmid with an AT-inducible promoter without an open reading frame (empty vector). (D) Derivative BJ13, which produces GFP-tagged EbfC from an inducible promoter.

Localization of EbfC in borrelial cells was investigated by use of a fusion protein between EbfC and green fluorescent protein (GFP). Production of the GFP-tagged EbfC protein in B. burgdorferi resulted in dense clusters of green fluorescence that colocalized with the bacterial nucleoids (Fig. 2D). Wild-type bacteria or bacteria carrying an empty expression vector did not exhibit any green fluorescence (Fig. 2A and C). B. burgdorferi strains expressing GFP alone were uniformly green fluorescent throughout the bacterial cells, indicating that GFP by itself does not colocalize with DNA (Fig. 2B).

Consensus EbfC-binding sequences occur throughout the B. burgdorferi genome (74), another characteristic of nucleoid-associated proteins. Previous chromatin immunoprecipitation (ChIP) of EbfC binding to live B. burgdorferi confirmed that this protein is bound to the GTnAC-containing erp operators in vivo (48). A variation of our earlier method was utilized to determine whether EbfC also binds in vivo to additional sites. Briefly, proteins and DNA were cross-linked in live B. burgdorferi, and EbfC-containing complexes were immunoprecipitated. Eluted DNA fragments were digested and ligated into a plasmid cloning vector, and the inserts of 43 randomly chosen clones were sequenced. All contained fragments of B. burgdorferi genomic DNA (Table 1). Control studies using nonspecific IgG or without any antibody failed to yield clones of borrelial DNA.

Table 1.

Genomic origins of 43 randomly chosen clones of EbfC chromatin-immunoprecipitated DNAs^a

Genetic element	Sequence coordinates	Coding/noncoding	Corresponding region
Main chromosome	12728–12756	NC	priA (BB0014)
Main chromosome	37940–38053	C	BB0038
Main chromosome	93805–93864	C	BB0096
Main chromosome	93809–94042	C/NC	BB0096-BB0097
Main chromosome	120058–120147	C	rpsB (BB0123)
Main chromosome	141097–141198	C	acrB (BB0140)
Main chromosome	201525–201607	C	murE (BB0201)
Main chromosome	319529–319649	NC	cheW
Main chromosome	336976–337084	C/NC	oppAI/II/III (BB0327-329)
Main chromosome	414930–415072	C	proS (BB0402)
Main chromosome	422715–422765	C	nucA (BB0411)
Main chromosome	564132–564193	NC	ligA (BB0552)
Main chromosome	621782–620368	NC/C	murB (BB0598)
Main chromosome	683512–683630	C	ptsG (BB0645)
Main chromosome	735520–735578	NC/C	xylR-1 (BB0693)
Main chromosome	773553–773591	C	pbp3 (BB0732)
Main chromosome	784311–784355	C	ABC transporter ATP-binding protein (BB0742)
Main chromosome/cp26/lp38	880015–880025/8000–8010/22073–22063	NC/C	xylR-2 (BB0831)/BBB10/BBJ29
cp9	4290–4431	NC	eppA (BBC06)
cp26	8160–8227	C	BBB10
cp32-1	2504–2574	C/NC	BBP03/BBP04
cp32-1	24104–24201	C	bppA (BBP38)
cp32-3	25746–26046	C/NC	erpG (BBS41)
cp32-4	19467–19568	NC	BBR30/BBR31
cp32-4	22988–23182	C	bppA (BBR36)
cp32-4	25747–25777	NC	erpH (BBR40)
cp32-4, cp32-9, lp56	5608–5696/5614–5702/9068–9156	NC/C	BBR08/BBN08/BBQ16-17
cp32-6	12112–12256	C	BBM18
cp32-6	25593–25635	C/NC	erpK (BBM38)
lp17/lp56	1208–1332/51606–51730/1237–1361	NC	BBD03/BBQ85-86
lp28-1/lp36	8608–8699/22131–22294	NC	BBF17/BBK34-35
lp28-1	9766–9844	C	BBF18
lp28-2	1484–1543	C	BBG02
lp28-2	16466–16562	NC	BBG20
lp28-3	26022–26138	C	BBH37
lp36	17437–17496	C/NC	BBK25
lp38	10448–10545	NC	BBJ15-16
lp54	2975–3105	C	Antigen S1 (BBA05)
lp54	10232–10383	NC	ospA/ospB (BBA15-16)
lp54	12924–13044	C/NC	BBA19
lp54	17189–17268	NC	dbpB (BBA25-26)
lp54	41406–41497	C	BBA61
lp56/main chromosome	15775–15788/731099–731112	NC/C	erpX (BBQ47)/BB0689

^aFor some DNAs, the identical sequence occurs in multiple locations, so all possible locations are indicated. Sequence coordinates of each cloned DNA fragment in the B. burgdorferi strain B31 genome are listed (17, 32). ChIP indicated in vivo EbfC binding to 4 loci that were also identified by RNA-Seq as being regulated by EbfC, indicated in bold.

Consistent with our earlier results (48), 4 of the 43 plasmid clones contained erp operator DNA: erpG on cp32-3, erpH on cp32-4, erpK on cp32-6, and erpX on the defective prophage lp56 (Table 1) (17, 85, 86). Other sequences bound in vivo by EbfC were distributed all through the genome, both on the main linear chromosome and on many of the smaller replicons (Table 1). All of the cloned DNAs contained either a full or a half consensus EbfC-binding sequence. Although not fully comprehensive, this survey provided evidence demonstrating that EbfC binds in vivo to sites throughout the borrelial genome.

In addition to the erp loci noted above, four other cloned EbfC-bound DNAs contained genes whose transcript levels were significantly altered in an EbfC-overexpressing B. burgdorferi mutant (see below). Thus, 8 of the 43 (19%) randomly sampled DNAs that were bound by EbfC in vivo were derived from genes that are also significantly affected by cellular EbfC levels.

Global effects of EbfC on borrelial transcript levels.

As noted above, nucleoid-associated proteins often exert genome-wide impacts upon gene expression. B. burgdorferi regulates its cellular levels of EbfC: ebfC is transcribed at significantly greater levels by bacteria during mammalian infection than by bacteria within the midguts of unfed ticks (57), and as we demonstrate below, ebfC is transcribed at high levels in rapidly growing bacteria but is undetectable in stationary-phase cells. B. burgdorferi EbfC positively affects transcription of the infection-associated erp operons, which show the same in vivo expression patterns as does ebfC (48, 57). Global impacts of EbfC on genome-wide B. burgdorferi gene expression were assessed by use of comparative RNA sequencing (RNA-Seq). Repeated attempts by us and others to delete ebfC from the B. burgdorferi genome have failed (48, 74), suggesting that its gene product is essential for this bacterium. As an alternative method to assess the effects of EbfC on borrelial gene expression, an inducible system which causes B. burgdorferi to overexpress EbfC was utilized (48, 96). This approach has been utilized in other studies of bacterial nucleoid-associated proteins (25, 77, 83). Transformed B. burgdorferi bacteria were cultured under conditions that led to either EbfC overexpression or normal levels of EbfC. Bacterial RNA was then purified, quantitatively sequenced, and compared. RNA-Seq results were validated by qRT-PCR of independent cultures.

For the cultures analyzed by RNA-Seq, ebfC transcripts were 28-fold more abundant in the induced bacteria than in the uninduced control bacteria. Enhanced production of EbfC did not cause any measurable effects on bacterial growth rate. As described below and in reference 57, cellular levels of EbfC vary manyfold during culture and the natural tick-mammal infectious cycle. Increased EbfC levels led to statistically significant (P < 0.05) changes in the transcript levels of 52 genes, some of which were positively influenced by EbfC while others were repressed (Table 2). This corresponds to approximately 4.5% of the genes in this strain of B. burgdorferi (17, 32).

Table 2.

B. burgdorferi genes whose transcript levels were significantly changed (P < 0.05) by elevated cellular levels of EbfC^a

Expression type and gene	No. of reads		Fold change (log2 transformed)	Protein function/conserved domains
	Control	EbfC elevated
Increased
lp54:BBA72	1	99.5	99.5	Putative outer membrane protein family P35
lp54:BBA65	1	32.4	32.4	Putative outer membrane protein family P35
Chrom:BB0160	1	15.2	15.2	Alanine racemase
lp54:BBA25	1	14.3	14.3	DbpB, decorin-binding protein B
Chrom:BB0634	1	12.4	12.4	Exodeoxyribonuclease V subunit gamma
lp28–3:BBH13	1	10.5	10.5	Unknown
lp54:BBA37	2	19.1	9.5	Unknown
lp54:BBA36	1	8.6	8.6	Putative lipoprotein protease
lp54:BBA71	7	57.2	8.2	Putative outer membrane protein family P35
Chrom:BB0635	2	15.2	7.6	Nicotinate phosphoribosyltransferase metabolism
Chrom:BB0577	31	223	7.2	Unknown
lp54:BBA73	13	85.7	6.6	Putative outer membrane protein family P35
Chrom:BB0403	3	19.1	6.4	Predicted secreted protein
Chrom:BB0680	179	1,070	6.0	Putative chemotaxis
Chrom:BB0585	13	66.7	5.1	MurD, UDP-N-acetylmuramoyl-l-alanine:d-glutamate ligase
Chrom:BB0163	67	332	4.9	Unknown
Chrom:BB0406	10	45.7	4.6	Unknown
Chrom:BB0681	178	779	4.4	Methyl-accepting chemotaxis protein
lp54:BBA66	15	61.0	4.1	Putative outer membrane protein family P35
Chrom:BB0683	100	402	4.0	3-Hydroxy-3-methylglutaryl-coenzyme A synthase
Chrom:BB0661	25	97.2	3.9	PriA, primosomal protein
Chrom:BB0776	11	40.0	3.6	Unknown
Chrom:BB0144	114	404	3.5	Putative glycine/betaine/l-proline transporter
Chrom:BB0569	62	215	3.5	Putative chemotaxis protein
Chrom:sBB0653	22	76.2	3.5	SecF, secretion
Chrom:BB0693	134	461	3.4	XylR-1, putative regulatory protein
Chrom:BB0636	16	51.4	3.2	G-6-P-1-dehydrogenase
Chrom:BB0245	59	183	3.1	Putative cytoskeletal/structural protein
Chrom:BB0238	107	320	3.0	Putative DNA replication
Chrom:BB0749	62	179	2.9	Metabolism
Chrom:BB0168	542	1,520	2.8	DnaK, molecular chaperone
Chrom:BB0066	9	47.6	2.4	Unknown
lp54:BBA55	6	51.4	2.4	Putative RNA recognition/binding motif
Decreased
lp54:BBA31	46	1.91	24.1	Putative bacteriophage-related protein
lp54:BBA50	31	1.91	14.2	Unknown
Chrom:BB0823	27	1.91	14.2	Putative lipoprotein
cp26:BBB14	24	1.905	12.6	Unknown
cp26:BBB07	64	5.72	11.2	Putative outer surface protein/integrin binding motif
lp54:BBA61	126	11.43	11.2	Putative signal transduction protein
lp54:BBA40	38	3.81	10.0	Unknown
lp54:BBA41	71	7.62	9.3	Metabolism
lp17:BBD14	63	7.62	8.3	Unknown
cp26:BBB23	92	11.4	8.1	Unknown
cp9:BBC01	61	7.62	8.00	Unknown
lp54:BBA74	522	78.1	6.9	Oms28, outer membrane protein
Chrom:BB0472	152	26.7	5.7	UDP-N-acetylglucosamine 1-carboxyvinyltransferase
lp54:BBA62	711	130	5.5	Putative lipoprotein
lp54:BBA38	201	45.7	4.4	Metabolism
Chrom:BB0461	184	47.6	3.9	DnaX
Chrom:BB0424	88	22.9	3.9	Metabolism
cp26:BBB17	224	61.0	3.7	Putative inositol-5-monophosphate dehydrogenase

^aGenes are listed according to genomic location and B. burgdorferi strain B31 ORF locus tags (17, 32). Chrom, main chromosome. Loci that were also demonstrated to be bound by EbfC in vivo are boldfaced. Numbers of RNA-Seq reads were adjusted relative to the average numbers of reads of rRNA transcripts per sample, which resulted in fractions for the EbfC-elevated sample. It is probable that additional genes are affected by cellular EbfC levels, as not all B. burgdorferi genes are transcribed during growth in laboratory medium, and the analyzed bacteria do not contain every DNA element that is naturally present in B. burgdorferi.

The validity of the RNA-Seq data was tested by qRT-PCR of 9 apparently regulated genes, using different cultures of induced and uninduced bacteria. Of the 9 loci, 8 were also found to be differentially expressed in the second culture set (Table 2; Fig. 3). The variability in transcript levels of the remaining gene, dnaX, may be due to its complex regulatory system; as noted below, transcription of dnaX can be dramatically affected by bacterial metabolic activity. The constitutively expressed flaB gene (59) and all the borrelial rRNAs were unaffected by EbfC. Altogether, the RNA sequencing results indicate with very high probability that the genes listed in Table 2 are, indeed, regulated by cellular levels of EbfC. Moreover, since a large proportion of B. burgdorferi genes are not transcribed during cultivation in laboratory medium (1, 10, 65, 71), it is likely that the Lyme disease spirochete EbfC regulon includes additional loci.

Fig 3.

qRT-PCR validation of RNA sequencing results. Eight randomly selected loci that were identified by comparative RNA sequencing as being influenced by cellular EbfC levels were also examined by quantitative reverse-transcription PCR analysis (qRT-PCR). Values indicate the fold change of each gene when EbfC was overexpressed relative to that gene under normal, background EbfC levels. All qRT-PCRs were performed two times, each in quadruplicate. Mean abundance of mRNA from each sample was normalized against transcript levels of the constitutively expressed flaB gene.

EbfC affected a wide variety of borrelial genes, encoding both structural and metabolic proteins (Fig. 4A). The distribution of EbfC-influenced genes was heavily biased toward the smaller replicons. The ca.-54-kb linear replicon lp54 is home to 34% of those genes, while the ca.-980-kb main chromosome carries 52% (Fig. 4B). The smaller replicons carry the majority of infection-associated and other contingency genes, while the main chromosome is largely comprised of genes for housekeeping proteins (17, 32).

Fig 4.

Graphical representation of genes that were significantly (P < 0.05) affected by cellular levels of EbfC. (A) Putative functional groupings of the EbfC-influenced genes. (B) Genomic distribution of loci influenced by EbfC.

Four of the EbfC-affected genes were also identified in the EbfC-ChIP survey: priA and xylR-1, both located on the main chromosome, and dbpB and open reading frame (ORF) BBA61, which are located on the small linear replicon lp54 (Tables 1 and 2) (32). The positively affected dbpB gene encodes decorin-binding protein B, an outer surface protein involved with adherence to host tissues during mammalian infection (26, 27, 28, 37, 38, 79, 80). B. burgdorferi encodes two proteins that show similarity to the E. coli xylose-responsive regulatory protein, XylR-1 and XylR-2 (32). However, B. burgdorferi does not utilize xylose (91), so the actual functions of those proteins are not known. If they are actually regulatory proteins, then some of the effects of EbfC on borrelial gene expression may be indirect, through a hierarchy of interwoven regulatory networks.

Among the EbfC-repressed genes were flhB, encoding a flagellar export/assembly protein (33), and ORF BBB07 (8). EbfC also repressed the gene encoding Oms28/BBA74, an outer membrane-associated protein that is produced during tick colonization but repressed during mammalian infection (19, 63, 64, 81).

B. burgdorferi ebfC and dnaX form a complex operon.

In a large proportion of Eubacteria, including B. burgdorferi, the ebfC ortholog is located on the main chromosome, immediately 3′ of dnaX (Fig. 5A). The dnaX gene encodes the τ and γ subunits of DNA polymerase (29, 30). E. coli cotranscribes dnaX and ebfC (18, 31). To determine whether this is also true of the Lyme disease spirochete, reverse transcription-PCR (RT-PCR) was performed on purified B. burgdorferi RNA, using one oligonucleotide primer complementary to a sequence within dnaX and another primer complementary to a sequence within ebfC (Table 3). A single, appropriately sized amplicon was obtained, indicating that a single mRNA contains both ebfC and dnaX (Fig. 5B, lane 1). Control reaction mixtures lacking either reverse transcriptase or random hexamer primers or in which template was replaced with water all failed to yield amplicons, indicating that the qRT-PCR product was not the result of contaminating DNA (Fig. 5B, lanes 2, 3, and 5).

Table 3.

Oligonucleotide primers used in this work

Oligonucleotide name	Sequence (5′–3′)	Target
DNAX-EBFC RT F	GTGAGTTTGAATATAATGAGCTTCA	dnaX-internal
DNAX-EBFC RT R	TTGTCAATATTATTCTTAACGCTAGAC	ebfC-internal
DNA QRT F	GCATATAAAAGTAGTGGTAGCG	dnaX
DNA QRT R	ATTGCTCATAAGACACTCCAG	dnaX
EBFC QRT F	GGCAGTAAATCCGTTAGATTT	ebfC
EBFC QRT R	CTTTAACCTTAGAGACAGCATC	ebfC
0020-F	GCTGCTATTCTTGCAGAG	bb0020
0020-R	GCTCCTGCCCATAAGTTTG	bb0020
0144-F	GCATCTGGAAAGATAGACGG	bb0144
0144-R	GGAACATAGCTTGGCACCAC	bb0144
0330-F	GGTAGATGAGACAATAGGAGC	bb0330
0330-R	GATTTTGTTATCCCTTCAGCGG	bb0330
0661-F	GCGAGCAATGTGGCGAAAAAG	bb0661
0661-R	GGACTGTAGTGACAATTAGGAC	bb0661
0680-F	GCGCAAGTAAGAAGAGCAG	bb0680
0680-R	GCTCACCTTTACAGAGTC	bb0680
0693-F	CAAGAACAGACCTGGCTC	bb0693
0693-R	CTCCCATTGAATACGCATAATC	bb0693
BBB07-F	CCTCTTTGCTCTATATTAACGG	bbb07
BBB07-R	CTAATCTAATTGCCCAGGCG	bbb07
BBA25-F	CTATTGGTCGGATGTAGTATTG	bba25
BBA25-R	CGGTTTTAAGACCTGTAAAAGC	bba25
BBA66-F	GCCTTAAACTACAGCTTTAG	bba66
BBA66-R	CTAGCAGTTCAAATGCAG	bba66
Flh-F	GCCATCAAGCTTCCTGAAG	flhB
Flh-R	GAATTTTTTGCCCATCTGG	flhB
Fla3	GGGTCTCAAGCGTCTTGG	flaB
Fla4	GAACCGGTGCAGCCTGAG	flaB

In order to map the operon's promoter, DNAs 5′ of dnaX were cloned into plasmids as a series of transcriptional fusions with gfp. B. burgdorferi bacteria were transformed with each chimeric plasmid, and production of GFP was assessed by flow cytometry. This method mapped the transcriptional promoter of the dnaX-ebfC operon to between −166 and −247 of the dnaX translational start codon (Fig. 5C).

Bacteria as diverse as E. coli and Corynebacterium glutamicum possess promoters within the dnaX open reading frame that can independently transcribe their ebfC orthologs (18, 24, 31, 72). The possibility that B. burgdorferi might behave likewise was addressed by cloning the dnaX gene, lacking any 5′ noncoding DNA, directly in front of gfp, to create pAAB100. B. burgdorferi transformed with that dnaX::gfp plasmid produced GFP, indicating that a transcriptional promoter is located within the dnaX ORF (Fig. 5D). GFP was also produced from an operon fusion consisting of the 560 bp 5′ of the ebfC translational start codon, but not from a smaller fusion containing 497 bp 5′ of ebfC (Fig. 5D). Primer extension analyses of B. burgdorferi containing pAAB100 indicated a transcriptional start site located 501 bp upstream from the ebfC initiation codon.

B. burgdorferi maximally expresses ebfC during periods of rapid growth.

Cellular levels of many nucleoid-associated proteins vary during different stages of bacterial growth, while levels of others may remain constant (4). To investigate the effects of growth stage on B. burgdorferi ebfC expression, cultured bacteria were sampled from lag phase through stationary phase, RNA was isolated, and qRT-PCR analysis was used to assess mRNA levels. Levels of ebfC message peaked at the time of maximal bacterial growth rate (Fig. 6). As the culture entered stationary phase, the amount of ebfC mRNA plummeted to levels that were not detectable. RNA extracts of stationary-phase B. burgdorferi contained appreciable levels of the flagellum-encoding transcript flaB, indicating that those bacteria were transcriptionally active.

Fig 6.

B. burgdorferi differentially expresses ebfC and dnaX according to growth phase. A culture of wild-type B. burgdorferi was cultured from lag phase through stationary phase. Culture density was measured daily, and bacterial aliquots were removed for analysis every 2 days. qRT-PCR was used to determine bacterial levels of ebfC and dnaX, normalized to message levels of the constitutively expressed flaB. The solid line illustrates changes in bacterial density over time (left y axis). The dotted line indicates relative levels of ebfC mRNA, while the dashed line indicates levels of dnaX mRNA.

Additionally, since dnaX and ebfC can be both cotranscribed and transcribed independently of one another (see above), levels of dnaX transcription were assessed. Expression of dnaX also peaked when B. burgdorferi was at maximal growth rate, although differences in relative levels of the two transcripts were observed (Fig. 6). Studies of the regulatory mechanisms controlling the apparently complex dnaX-ebfC operon are ongoing.

DISCUSSION

The above results indicate that the EbfC directly associates with the B. burgdorferi nucleoids, by binding sequences throughout the genome. Due to the high degree of structural conservation among EbfC orthologs (21, 56, 74), we hypothesize that this feature will be conserved across bacterial taxa. Nucleoid organization is not a trivial task for any bacterium and is especially complex for the Lyme disease spirochete with its 20+ distinct replicons. In addition to EbfC, only three other proteins that might function as nucleoid-associated proteins have been identified in B. burgdorferi. Either more, unique B. burgdorferi proteins await discovery, or this spirochete manages its complex genome with a very limited protein repertoire. The complicated, yet simple, Lyme disease spirochete thus presents itself as an intriguing model organism for studies of genome maintenance.

Nucleoid-associated proteins often have profound effects upon gene expression, through direct interactions with transcriptional machinery and through alterations in DNA structures than can influence transcription across great distances. The dynamic natures of multimolecular structures, changes in numbers of potential binding sites as the genome replicates, and regulated production of nucleoid-associated proteins during bacterial growth phases can all combine to produce fluctuating effects of nucleoid-associated proteins on global gene expression. A large proportion of Eubacteria contain adjacent dnaX and ebfC genes. Considering the strong selective pressures on organisms to maximize the efficiency of life processes, there appear to be benefits from linking DNA replication with this particular nucleoid-associated protein. In the present study, B. burgdorferi was found to produce maximum levels of ebfC transcript during the rapid, exponential growth phase, while ebfC mRNA was not detectable during late stationary phase. B. burgdorferi bacteria within the midguts of unfed ticks are relatively somnolent and probably do not divide for months on end, whereas tick feeding induces rapid borrelial growth and division times on the order of 1 to 2 h per cycle (12, 22, 67, 68, 69, 87). It is probably not a coincidence that several genes associated with mammalian infection are both positively influenced by EbfC and upregulated during transmission (26, 27, 34, 35, 36, 38, 39, 48, 60, 66, 79, 80). Conversely, the current studies found that EbfC repressed transcription of oms28/ORF BBA74, a gene that is expressed during tick colonization but not during vertebrate infection (64). Studies are under way to elucidate the mechanisms behind the stimulatory/repressive effects of EbfC on these genes.

The natural infectious cycle of B. burgdorferi is extremely complex, not only involving persistent infection of the physiologically distinct vertebrate host and tick vector but also requiring efficient bacterial transmission between host and vector. For example, a spirochete within a tick must recognize when the tick is feeding, traverse the tick midgut lining, pass through the hemocoel, penetrate the salivary glands, and enter the salivary ducts, just to be delivered to the host. The bacterium must then balance the need for adherence to host tissues with its drive to disseminate through the host, to eventually establish persistent infection of various organs and tissues. A number of studies have demonstrated that the Lyme disease spirochete utilizes an alternative sigma factor, named RpoS, to control production of several proteins required for early stages of mammalian infection (14, 15, 16, 43, 76). Data from the current report and others (6, 48, 57) indicate that the EbfC regulon may also be involved with borrelial transmission. While overlap between the RpoS and EbfC regulons is probable, the two mechanisms are also distinct, since expression of EbfC is known to be independent of RpoS (57), and rpoS levels were not detectably affected by EbfC (this work). In addition, the Lyme disease spirochete possesses numerous other regulons that overlap to varied degrees (70). The understanding of bacterial gene regulation has come a long way since the initial E. coli lac operon-based concept of one operon-one regulator. It is now generally well accepted that gene regulation can be as complicated in prokaryotes as it is in eukaryotes, with multiple factors cooperating/competing to fine-tune transcription levels in response to a diversity of stimuli (46). The apparent use of EbfC and other borrelial factors in controlling transmission-associated genes calls for further studies into these regulons, of both how target genes are affected and how the regulators are themselves controlled.

Nearly every species of Eubacteria carries a recognizable homolog of EbfC. The structures of the H. influenzae and E. coli orthologs have been solved and are essentially identical (reference 56 and PDB entry 1PUG). Even though the spirochete and proteobacterium phyla diverged many millions of years ago, the sequence of B. burgdorferi EbfC was easily modeled on the H. influenzae ortholog (74), indicating extremely strong constraints on these proteins' structures. The present studies determined that B. burgdorferi EbfC is a nucleoid-associated protein. As additional evidence, while the manuscript was in review, another group published that the EbfC homolog of Deinococcus radiodurans is associated with the nucleoid of that bacterium (94). The conservation of ebfC genes across the domain Eubacteria suggests that the orthologs of other species serve the same function and may similarly participate in global regulation of gene expression.