The Fur homologue BosR requires Arg39 to activate rpoS transcription in Borrelia burgdorferi and thereby direct spirochaete infection in mice

Abstract

Borrelia burgdorferi is the causative agent of Lyme disease. In B. burgdorferi, RpoS controls the expression of virulence genes needed for mammalian infection. The Fur homologue BosR regulates the transcription of rpoS and therefore BosR determines, albeit indirectly, the infection status of the spirochaete. Transcription of rpoS in B. burgdorferi is complex: rpoS can be transcribed either from an RpoD-dependent promoter to yield a long transcript or from an RpoN-dependent promoter to yield a short transcript. This study shows that BosR repressed synthesis of the long transcript while at the same time activating synthesis of the short transcript. How BosR does this is unclear. To address this, spirochaetes were engineered to express either BosR or the naturally occurring variant BosRR39K. Mice became infected by the spirochaetes expressing BosR but not by the spirochaetes expressing BosRR39K. Furthermore, the spirochaetes expressing BosR activated rpoS transcription during growth in culture whereas the spirochaetes expressing BosRR39K did not. Thus, BosR's activation of rpoS transcription somehow involves Arg39. This arginine is highly conserved in other FUR proteins and therefore other FUR proteins may also require this arginine to function.

Introduction

Lyme disease is caused by the spirochaete Borrelia burgdorferi (Benach et al., 1983; Burgdorfer et al., 1982; Steere et al., 1983). Infection occurs in humans when the spirochaete is transmitted by an infected tick during tick feeding. Normally, however, the spirochaete completes its life cycle in the wild, passing back and forth between an Ixodid tick vector and various vertebrate hosts (e.g. mice, deer). This requires that the spirochaete survive and multiply in two very different environments. To do this, B. burgdorferi employs a separate set of genes for each of these environments. Determining how the spirochaete turns these genes on and off is currently an area of intense research (Radolf et al., 2012; Samuels, 2011).

In B. burgdorferi, many of the genes required for mammalian infection are under the control of the alternative sigma factor RpoS (Radolf et al., 2012). Transcription of rpoS in B. burgdorferi is complex. Under certain growth conditions (i.e. higher temperatures, low cell density), the spirochaete produces a long rpoS transcript from an RpoD (σ⁷⁰)-dependent promoter; however, under other growth conditions (i.e. acid pH, higher temperatures, high cell density), the spirochaete produces a short rpoS transcript from an RpoN (σ^N)-dependent promoter (Hübner et al., 2001; Lybecker & Samuels, 2007; Lybecker et al., 2010; Samuels, 2011). Translation of the long transcript (but not the short transcript) also requires the assistance of a small RNA (Samuels, 2011).

Production of the short rpoS transcript is tightly regulated, requiring both the alternative sigma factor RpoN (Hübner et al., 2001) and the Fur homologue BosR (Hyde et al., 2009; Ouyang et al., 2009). To initiate transcription, RpoN must first be activated by the response regulator Rrp2 (Yang et al., 2003) which, in turn, must first be phosphorylated (Xu et al., 2010). The binding site for RpoN within the rpoS promoter is known (Smith et al., 2007; Studholme & Buck, 2000). However, even though Rrp2 has a predicted DNA-binding domain, it appears that Rrp2 acts on RpoN without the benefit of binding DNA (Blevins et al., 2009; Burtnick et al., 2007; Yang et al., 2003). In vitro, BosR binds multiple sites within the rpoS promoter (Ouyang et al., 2011, 2014, 2015). Presumably, in vivo, BosR must bind one or more of these sites to activate rpoS transcription.

Originally, BosR was annotated as Fur (ferric uptake regulator) because it showed homology to the FUR family of DNA-binding proteins (Fraser et al., 1997). Boylan et al. (2003) renamed it BosR (Borrelia oxidative stress regulator) and put forward the idea that BosR's function in B. burgdorferi was to regulate the oxidative stress response. Genes which may be regulated by BosR in this fashion include napA (renamed bicA), sodA and cdr (Boylan et al., 2003, 2006).

Although BosR may be involved in the oxidative stress response (Hyde et al., 2009, 2010), it appears its main function is to act as a global regulator. To prepare the spirochaete for mammalian infection, BosR upregulates transcription of rpoS (Hyde et al., 2009, 2010; Ouyang et al., 2009, 2011) and also downregulates transcription of various tick-phase genes (e.g. ospA and ospD), which need only be expressed in the tick and which if allowed expression in the mammal could lead to immune clearance (Shi et al., 2014; Wang et al., 2013). When BosR activates rpoS, it binds the direct repeat (DR) sequence TAAATTAAAT (Ouyang et al., 2011); however, when BosR represses ospA and ospD, it binds a variant of the Per box sequence TTATAAT-ATTATAA (Wang et al., 2013). How BosR can bind two very different sequences and generate two very different outcomes is currently unclear.

B. burgdorferi strain B31 was isolated in 1981 from a tick collected on Shelter Island (Burgdorfer et al., 1982). Genome sequencing of strain B31 identified a fur homologue (bb0647) and determined that residue 39 of the predicted protein was an arginine (Fraser et al., 1997). Since then, clonal isolates of strain B31 have been found that contain a single nucleotide change in bosR, making residue 39 of BosR a lysine (unpublished data; Seshu et al., 2004). This residue is highly conserved as arginine in virtually all FUR proteins. With BosR, substitution of this arginine with lysine affects the protein's ability to bind DNA in vitro (Seshu et al., 2004). Not known is what effect this substitution has on the protein's ability to function in vivo during infection (Hyde et al., 2006; Seshu et al., 2004).

Here, strains of B. burgdorferi were constructed that expressed either BosR or BosRR39K. Mice became infected by the strains expressing BosR but not by the strains expressing BosRR39K. This was, at least in part, because the spirochaetes expressing BosR activated rpoS transcription while the spirochaetes expressing BosRR39K did not. The R39K mutation affected the ability of BosRR39K to bind the rpoS promoter in vitro. It is therefore possible that BosRR39K failed to activate rpoS transcription because it could not bind the rpoS promoter in vivo. This study shows that BosR needs Arg39 to upregulate rpoS and in so doing support mammalian infection.

Methods

Bacterial strains and culture conditions

Table S1 (available in the online Supplementary Material) lists the B. burgdorferi strains used in this study. Strain B31-A3 was kindly provided by Patricia Rosa and strain B31-F was kindly provided by Justin Radolf. Spirochaetes were routinely cultured at 33 °C in BSK-II (Barbour, 1984) or BSK-H medium (Sigma) supplemented with 6 % rabbit serum. To upregulate rpoS, spirochaetes were cultured at 37 °C in BSK-II or BSK-H medium supplemented with 6 % rabbit serum and adjusted to pH 6.8. Spirochaetes were enumerated by dark-field microscopy.

bosR disruption mutants and complemented strains

bosR disruption mutants and complemented strains were constructed as detailed in the Supplementary Material.

Mouse infections

C3H/HeN mice (Taconic) were infected by intradermal injection at 8 weeks of age. Ears, heart and bladder were collected 21 days post-infection, and the tissues were cultivated at 33 °C in BSK-H medium containing 6 % rabbit serum with or without 50 μg rifampicin ml^− 1. Outgrowth of spirochaetes in these cultures was assessed by dark-field microscopy at weekly intervals for up to 1 month at which time the infections were deemed negative if no spirochaetes were detected. All animal procedures were carried out according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at Stony Brook University.

Quantitative real-time PCR (qRT-PCR)

qRT-PCR was carried out on a model 7500 machine (Applied Biosystems) using the relative quantification method (ΔΔC_t) with flaB as endogenous control. RNA was extracted with Tri-Reagent LS (Molecular Research Center) and further purified on an RNeasy spin column (Qiagen). cDNA was synthesized by using random hexamers with the Superscript III first strand synthesis system (Invitrogen). qRT-PCR primers (Table S2) were designed with ABI Primer Express software. ABI Power SYBR Green PCR master mix was used with 200 nM primers.

Gel shift assays

Gel shift assays were carried out as previously described (Katona et al., 2004). Briefly, recombinant BosR (or BosRR39K) was mixed with target DNA in 20 mM Bistris/borate (pH 7.5) buffer containing 1.5 mM DTT, 1.3 mM MgCl₂, 53 mM KCl, 3.3 % glycerol and 133 μg BSA ml^− 1, incubated for 20 min at room temperature, and separated on a 6 % nondenaturing gel. The DNA was stained with SYBR Green I (Molecular Probes) and the gel was scanned in blue fluorescence mode on a Storm 860 imager (Amersham-Pharmacia Biotech).

Target DNA

Target DNA was generated by PCR from cloned sequences. For P_rpoS, a 1.6 kb region containing the rpoS gene plus upstream sequence was PCR-amplified from strain B31-A3 genomic DNA with primers BB0771-SphI F and BB0771-XhoI R and inserted into the SphI–XhoI site of pST-Blue1 (Novagen), giving pLK43/1. The P_rpoS target DNA was then amplified by PCR from pLK43/1 using primers BB0771p F1 and BB0771p R1. For BbDR, equal molar amounts of the synthesized single strands DRF2 and DRR2 were mixed, heated for 5 min at 95 °C, and slowly cooled to room temperature. Next, the duplex oligonucleotide was inserted into the EcoRV site of pST-Blue1, giving pLK45/16. The BbDR target DNA was then amplified by PCR from pLK45/16 using primers T7 promoter and U-19mer. Table S2 lists all primer sequences.

Recombinant BosR and BosRR39K

Recombinant proteins were prepared as previously described (Katona et al., 2004). Briefly, the His-tagged proteins were overexpressed in Escherichia coli BL21(DE3) from pLK1/5 or pLK42/2, purified by zinc chelate chromatography, and then treated with thrombin to free the N-terminal His tags. pLK42/2 was constructed by PCR-amplifying the bosRR39K gene from pLK38/2 using primers FLAF and FRR (Table S2) and inserting it into the NdeI–XhoI site of pET28a (Novagen). pLK1/5 was from an earlier study (Katona et al., 2004).

Tricine SDS-PAGE and Western blotting

Tricine SDS-PAGE was carried out as previously described (Katona et al., 2004). Western blots of recombinant proteins were probed with mouse anti-His-1 monoclonal antibody (Sigma) to detect the N-terminal His tags. Western blots of Borrelia lysates were probed with mouse anti-FlaB CB1 monoclonal antibody (Coleman & Benach, 1989), rat anti-OspC (kindly provided by Justin Radolf) or rat anti-BosR. Rat anti-BosR was raised against purified recombinant BosR that retained the N-terminal His tag. All blots were developed with IRDye-conjugated secondary antibodies (Rockland or LI-COR) and scanned on an Odyssey infrared imager (LI-COR). Odyssey 2-colour protein markers were used.

Statistics

Fisher's exact test was used to measure statistical significance in the mouse studies and Student's t-test was used to measure statistical significance in the qRT-PCR studies.

Results

bosR disruption mutants are generated in strain B31-A3

In previous studies, knockouts of bosR were generated in strains B31-MI (Ouyang et al., 2009), ML23 (Hyde et al., 2009) and 297 (Ouyang et al., 2011). Strains B31-MI and 297 are infectious in mice (Ouyang et al., 2009, 2011); however, strain ML23 lacks the lp25 linear plasmid and therefore is not infectious in mice (Hyde et al., 2009). Here, strain B31-A3 was chosen for study. Strain B31-A3 is a clonal isolate of strain B31-MI that retains infectivity (Elias et al., 2002).

Two sets of bosR mutants were generated (Fig. S1a). A kanamycin resistance (kan) cassette was introduced by allelic exchange in the forward orientation to yield the L-series mutants and in the reverse orientation to yield the K-series mutants. The kan cassette had a flgB promoter to drive constitutive expression of the kanamycin resistance gene and a T7 terminator to avoid overexpression of downstream genes. A total of 58 transformants were identified: 28 with the kan cassette in the forward orientation and 30 with the kan cassette in the reverse orientation. Clones L9 and K18 were selected for further study. Both clones contained the full complement of linear and circular plasmids originally identified by Elias et al. (2002) in strain B31-A3 (data not shown). PCR products of a size consistent with the introduction of a kan cassette were obtained for both clones (Fig. S1b). Sequence analysis of these PCR products confirmed that bosR was disrupted and that no unwanted mutations were incorporated.

The complemented strains are engineered to express either BosR or BosRR39K

Repeated attempts were made to complement the bosR mutants in trans by introducing a pKFSS1 shuttle vector harbouring bosR; however, no transformants were identified. Li et al. (2007) succeeded in complementing their dps mutant by inserting dps with its promoter into the intergenic region between bb0445 and bb0446. This approach offered several advantages: the location on the main chromosome ensured that the gene was present as a single copy and because the gene was located downstream of both bb0445 and bb0446, transcriptional read-through from the inserted sequences was less of a concern. Here, their approach was adopted, but with some modifications. Li et al. (2007) inserted dps downstream of bb0445 and followed it with a streptomycin resistance (str) cassette. This arrangement, however, made the inserted gene susceptible to transcriptional read through from promoters upstream of bb0445. Here, bosR along with its promoter was inserted downstream of the str cassette and a T7 terminator was then added to the cassette to prevent read through from the cassette's flgB promoter.

Three complemented strains were constructed for each bosR mutant strain: a mock complement that contained only the str cassette, a bosR complement that contained the str cassette plus a WT bosR gene, and a bosRR39K complement that contained the str cassette plus an R39K-mutated bosRR39K gene (Fig. S2a). Multiple transformants were identified. A single clone was selected from each set and assayed by PCR to determine that it contained the desired insert (Fig. S2b). All of the selected clones were then screened for plasmid content and found to have the same profile as strain B31-A3 (data not shown). DNA sequence analysis confirmed the identity of the inserts and established that no unwanted mutations were incorporated. Clones KN1 (mock complement), KN107 (bosR complement) and KN203 (bosRR39K complement) were from bosR mutant K18. Clones LN1 (mock complement), LN101 (bosR complement) and LN202 (bosRR39K complement) were from bosR mutant L9.

bosR is expressed at near WT levels in both of the bosR-complemented strains

Ouyang et al. (2009) reported that the region bb0648-bb0647-bb0646 functioned as an operon with transcription initiating from a promoter upstream of bb0648. Because genes within an operon can often be transcribed from their own promoters as well, qRT-PCR analyses were carried out on the bosR-complemented strains to determine if bosR was being transcribed. For these studies, the spirochaetes were cultured at 33 °C and the RNA was isolated during early exponential phase, mid-exponential phase and stationary phase. As expected, the bosR-mutant strains (L9 and K18) produced no bosR transcript (Table S3). However, the bosR-complemented strains (LN101 and KN107) each produced nearly the same level of bosR transcript as the WT strain during all three stages of growth (Table S3). Therefore, the sequence located immediately upstream of bosR does appear to function as a promoter to allow transcription independent of the operon.

qRT-PCR analyses were also carried out on the gene located immediately downstream of bosR. Compared with the WT strain, the two bosR-mutant strains each showed a lower level of expression of bb0646 during early exponential phase and mid-exponential phase, and a much lower level of expression during stationary phase (Table S3). This decreased expression of bb0646 was not due to the loss of bosR expression, because both of the bosR-complemented strains also showed decreased expression of bb0646 (Table S3). Interestingly, the orientation of the kan cassette did not seem to matter: bosR mutant L9 (forward orientation) and bosR mutant K18 (reverse orientation) showed the same decreased expression of bb0646 (Table S3). Thus, if the T7 terminator embedded within the kan cassette was responsible for terminating transcription of bb0646 from an upstream promoter, then it did so from either orientation.

From these data, it was clear that the bosR-mutant strains showed two defects: loss of bosR expression and reduced expression of bb0646. The reduced expression of bb0646 due to disruption of bosR is a cis effect since complementation of bosR did not restore bb0646 expression. Neither of these defects, however, affected the ability of the spirochaetes to grow in culture (Fig. S3 and data not shown).

The complemented strains express comparable levels of bosR and bosRR39K transcript

Seshu et al. (2004) reported that strain CHP100 encodes a variant of the bosR gene. This variant, bosRR39K, is also present in strain B31-F (unpublished observation). Strains CHP100 and B31-F were both derived from high-passage strain B31: CHP100 in the Skare laboratory (Hyde et al., 2006; Seshu et al., 2004) and B31-F in the Radolf laboratory (Eggers et al., 2002). Other mutations in the bosR gene have not been identified.

Seshu et al. (2004) further reported that BosRR39K and BosR did not function the same in vitro or in vivo. However, because their studies were carried out with non-infectious isolates, it remained unclear if BosRR39K, like BosR, could support infection in mice. Here, spirochaetes were engineered to express bosR or bosRR39K in the infectious background of strain B31-A3. According to the qRT-PCR analyses, both of the bosRR39K-complemented strains (LN202 and KN203) produced transcript at a level comparable to that of the bosR-complemented strains (Table S3). Also, both of the bosRR39K-complemented strains expressed decreased levels of bb0646 (Table S3). Thus, the bosR- and bosRR39K-complemented strains seemed to differ only in their expression of bosR versus bosRR39K.

Mice become infected by spirochaetes expressing BosR but not BosRR39K

Others have reported that bosR-knockouts of B. burgdorferi are not infectious in mice (Hyde et al., 2009; Ouyang et al., 2009, 2011). Here, the WT strain B31-A3 was infectious at a dose of 20 000 spirochaetes per mouse, whereas neither of the bosR-mutant strains (L9 and K18) was infectious at this dose or at a higher dose of 100 000 spirochaetes per mouse (Table 1 and data not shown).

Table 1. Infectivity of B. burgdorferi clones in mice.

Mice were infected by intradermal injection of 2 × 10⁴ spirochaetes.

Strain	Description	No. of cultures positive/total no. of cultures			No. of mice infected/total no. of mice*
		Skin	Bladder	Heart
A3	WT	17/18	17/18	17/18	17/18
L9	bosR mutant of A3	0/8	0/8	0/8	0/8
K18	bosR mutant of A3	0/8	0/8	0/7	0/8
LN1	Mock complement of L9	0/10	0/10	0/10	0/10
KN1	Mock complement of K18	0/10	0/10	0/10	0/10
LN101	bosR complement of L9	4/10	4/10	4/10	4/10
KN107	bosR complement of K18	6/10	6/10	6/10	6/10
LN202	bosRR39K complement of L9	0/10	0/10	0/10	0/10
KN203	bosRR39K complement of K18	0/10	0/10	0/10	0/10

^*P < 0.001 by Fisher's exact test for the following comparisons: (i) bosR mutants L9 and K18 vs WT A3, (ii) bosR complements LN101 and KN107 vs mock complements LN1 and KN1, and (iii) bosRR39K complements LN202 and KN203 vs bosR complements LN101 and KN107.

Ouyang et al. (2009) also reported that their bosR-complemented strain regained its infectious phenotype. Here, both of the bosR-complemented strains (LN101 and KN107) were infectious; however, the frequency of infection was about half that of the WT strain (Table 1). In part, this may have been because the bosR-complemented strains expressed slightly less bosR transcript than the WT strain (Table S3). Also, the defect in bb0646 expression may have been a contributing factor (Table S3). Shaw et al. (2012) found that while a bb0646 knockout was infectious in mice, its ability to infect was attenuated. bb0646 encodes an enzyme that has lipase activity (unpublished observation; Shaw et al., 2012). It is thus possible that the spirochaete uses this enzyme to help promote infection.

Regardless of the reason for the reduced infectivity of the bosR-complemented strains, the results still show that expression of bosR from the bb0445–446 intergenic site allowed the spirochaete to regain infectivity. In contrast, Table 1 shows that neither of the bosRR39K-complemented strains (LN202 or KN203) was infectious at the dose of 20 000 spirochaetes per mouse. Thus, while the bosR-complemented strains infected 10 out of 20 mice, the bosRR39K-complemented strains infected 0 out of 20 mice (Table 1). According to Fisher's exact test, there is a highly significant difference between these two frequencies of infection (P < 0.001).

Taken together, the data show that under this one set of infection conditions, the spirochaetes expressing BosR were capable of infecting mice whereas the spirochaetes expressing BosRR39K were not.

OspC is detected in spirochaetes expressing BosR but not BosRR39K

Others have suggested that the reason why their bosR-knockout strains were not infectious in mice was, at least in part, because they failed to produce sufficient OspC (Hyde et al., 2009; Ouyang et al., 2009). To test this, WT strain B31-A3, bosR mutant L9, mock complement LN1, bosR complement LN101 and bosRR39K complement LN202 were each cultured under conditions previously shown to upregulate expression of OspC in vitro (Yang et al., 2000). Under these conditions (37 °C, pH 6.8), OspC was expressed at high levels in both the WT strain and bosR complement, but not in the bosR mutant, mock complement or bosRR39K complement (Fig. 1). The level of BosR, or BosRR39K, present in the samples was also determined. The bosR mutant and mock complement showed no protein; however, the WT strain, bosR complement and bosRR39K complement each showed about the same level of protein (Fig. 1). Thus, the bosRR39K complement did not fail to express OspC only because the BosRR39K was not expressed or not stable. A second experiment gave similar results (data not shown).

Fig. 1.

Expression of OspC, BosR and BosRR39K in bosR-mutant and complemented strains. Spirochaetes were cultured for 6 days (late-exponential phase) at 37 °C in BSK-H medium (pH 6.8) containing 6 % rabbit serum. Whole-cell lysates were prepared and subjected to Tricine SDS-PAGE. Each lane received the lysate from ∼6 × 10⁷ spirochaetes. Western blots were probed with mouse monoclonal antibody CB1 to detect FlaB, rat anti-OspC antiserum to detect OspC, and rat anti-BosR antiserum to detect BosR or BosRR39K. FlaB served as the loading control. Protein size standards (in kDa) appear to the left.

Transcripts of ospC and rpoS are detected by qRT-PCR in spirochaetes expressing BosR

OspC expression in B. burgdorferi is regulated at the level of transcription by the alternative sigma factor RpoS (Samuels, 2011). Both Ouyang et al. (2009) and Hyde et al. (2009) reported that their bosR-knockout strains produced less RpoS than their WT strains during growth in culture. Ouyang et al. (2009) also reported that their bosR-knockout strain produced less rpoS transcript. Here, the WT strain, bosR mutant L9 and bosR complement LN101 were each cultured as before and the RNA was isolated for qRT-PCR during late-exponential phase. The results in Table 2 show that the level of ospC transcript was greatly decreased in the bosR-mutant strain compared with the WT strain (125-fold), thus confirming that BosR was needed to see upregulation of ospC.

Table 2. Relative expression levels of rpoS and ospC as determined by qRT-PCR.

In Experiment I, spirochaetes were inoculated at an initial density of 250 spirochaetes ml^− 1 and cultured at 37 °C for 6 days (late-exponential phase) in BSK-II medium (pH 6.8) containing 6 % rabbit serum. In Experiment II, spirochaetes were inoculated at an initial density of 143 spirochaetes ml^− 1 and cultured at 37 °C for 7 days (early stationary phase) in BSK-II medium (pH 6.8) containing 6 % rabbit serum.

Gene	Relative expression level*
	WT	bosR mutant	bosR complement	bosRR39K complement	Mock complement
	A3	L9	LN101	LN202	LN1
Experiment I
bb0771 (rpoS)	1.000	0.490 ± 0.005	1.328 ± 0.007	nd†	nd
bb0771L (rpoS-L)	1.000	2.103 ± 0.147	1.130 ± 0.113	nd	nd
bbb19 (ospC)	1.000	0.008 ± 0.001	1.208 ± 0.016	nd	nd
bb0647 (bosR)	1.000	0.000 ± 0.000	1.277 ± 0.043	nd	nd
bb0646	1.000	0.132 ± 0.001	0.123 ± 0.008	nd	nd
Experiment II
bb0771 (rpoS)	nd	nd	1.000	0.079 ± 0.008	0.132 ± 0.004
bb0771L (rpoS-L)	nd	nd	1.000	0.789 ± 0.079	1.519 ± 0.017
bbb19 (ospC)	nd	nd	1.000	0.002 ± 0.000	0.003 ± 0.000
bb0647 (bosR)	nd	nd	1.000	0.744 ± 0.018	0.000 ± 0.000
bb0646	nd	nd	1.000	0.870 ± 0.008	1.569 ± 0.155

^*Mean ± sem of two determinations, each with four replicates. Statistical significance was evaluated by Student's t-test for the following comparisons. Total rpoS transcript: L9 vs A3 (P = 0.01), L9 vs LN101 (P < 0.01), LN101 vs LN1 (P = 0.02), LN202 vs LN1 (P = 0.07) and LN101 vs LN202 (P < 0.01). Long rpoS transcript: L9 vs A3 (P = 0.04), L9 vs LN101 (P = 0.02), LN101 vs LN1 (P = 0.01), LN202 vs LN1 (P = 0.08) and LN101 vs LN202 (P = 0.23).

^†nd, Not done.

As noted earlier, Borrelia rpoS can be transcribed from an RpoD-dependent promoter to yield a long transcript or from an RpoN-dependent promoter to yield a short transcript (Lybecker & Samuels, 2007; Smith et al., 2007). Here, two sets of qRT-PCR primers were designed to allow detection of both transcripts (Figs. 2 and S4). The first set of primers (Fig. 2, primers 1 and 2) recognized sites near the 3′ end of both long and short transcripts and therefore yielded data on the level of total rpoS transcript (i.e. the sum of long and short transcripts). The second set of primers (Fig. 2, primers 3 and 4) recognized sites near the 5′ end of the long transcript (but not the short transcript) and therefore yielded data on the level of long rpoS transcript only. qRT-PCR analysis with primers 1 and 2 showed that the level of total rpoS transcript was decreased twofold in the bosR-mutant strain compared with the WT strain (Table 2, Experiment I – rpoS). However, qRT-PCR analysis with primers 3 and 4 showed that the level of long rpoS transcript was increased 2.1 fold in the bosR-mutant strain compared with the WT strain (Table 2, Experiment I – rpoS-L). Thus, BosR appears to have repressed long rpoS transcription while at the same time activated total rpoS transcription (Fig. 3a, b). These data do not indicate if BosR acted directly or indirectly; however, two additional studies gave similar results (Tables S4 and S5). Taken together, the combined data (Fig. 4a, b) showed that the increased expression of total rpoS transcript by the bosR mutant was highly statistically significant (P < 0.01) while the decreased expression of the long rpoS transcript by the bosR mutant was not quite statistically significant (P = 0.11).

Fig. 2.

Diagram illustrating the long and short transcripts of B. burgdorferi rpoS. Small arrows indicate the sites recognized by the qRT-PCR primers in the cDNA synthesized from the rpoS transcripts. Primers 1 and 2 (BB0771 RTF1 and BB0771 RTR1) allow amplification of both long and short transcripts while primers 3 and 4 (BB0771 RTF3 and BB0771 RTR3) allow amplification of the long transcript only.

Fig. 3.

Regulation of rpoS transcription by BosR versus BosRR39K. The plots show the relative expression levels for total rpoS transcript (a, c and e) and long rpoS transcript (b, d and f) as determined by qRT-PCR. (a, b) Experiment I: WT strain B31-A3 was compared with bosR mutant L9 and bosR complement LN101. (c, d) Experiment II: bosR complement LN101 was compared with bosRR39K complement LN202 and mock complement LN1. (e, f) Experiment III: bosR complement LN101 (black bars) was compared with bosRR39K complement LN202 (white bars) using spirochaetes harvested during mid-exponential phase (day 6), late-exponential phase (day 7) and stationary phase (day 8). Error bars show the sem (n = 2).

Fig. 4.

BosR both activates total rpoS transcription and represses long rpoS transcription. The plots show the relative expression levels for total rpoS transcript (a, c and e) and long rpoS transcript (b, d and f) as determined by qRT-PCR. (a, b) WT strain B31-A3 was compared with bosR mutant L9 using data taken from Tables 2, S4 and S5 (n = 3). (c, d) bosR mutant L9 and mock complement LN1 (both of which are bosR disruption mutants) were compared with bosR complement LN101 using data taken from Tables 2 and S6 (n = 3). (e, f) bosR complement LN101 was compared with bosRR39K complement LN202 using data taken from Tables 2, S6 and S7 (n = 5). Statistical significance was evaluated by Student's t-test. P values ≤ 0.05 were interpreted to indicate statistical significance. Error bars show the sem.

Ouyang et al. (2009) previously reported that BosR activated rpoS transcription; however, their qRT-PCR analysis determined the level of total rpoS transcript and did not distinguish between the long and short transcripts. Here, the qRT-PCR analysis showed that BosR activated short rpoS transcription (Fig. 4a), but not long rpoS transcription (Fig. 4b). All three studies showed, however, that the WT strain and bosR mutant strain differed in their expression of bb0646 (Tables 2, S4 and S5). Thus, the rpoS result is valid only if it was unaffected by the varied expression of bb0646. To address this concern, additional studies were carried using the complemented strains. Experiment I showed that the bosR-complemented strain mirrored the behaviour of the WT strain, yet expressed bb0646 at a level comparable to that of the bosR mutant (Table 2, Fig. 3a, b). Thus, adopting this strategy of comparing the complemented strains could circumvent the issues with bb0646.

Experiment II employed three complemented strains: the bosR complement LN101 which expressed only BosR, the bosRR39K complement LN202 which expressed only BosRR39K, and the mock complement LN1 which expressed neither BosR nor BosRR39K and thus functioned as the bosR mutant. Each strain was cultured at 37 °C, pH 6.8 to ensure high-level expression of ospC and upregulation of rpoS. Spirochaetes were harvested during early stationary phase and the RNA was isolated for qRT-PCR analysis. The results of the qRT-PCR analysis (Table 2) showed that the bosR complement produced 333-fold more ospC transcript than the mock complement, confirming that the culture conditions did allow for high-level expression of ospC. The results also showed that, as expected, all three complemented strains expressed similar levels of bb0646 transcript. Finally, the results showed that the bosR complement produced eightfold more total rpoS transcript than the mock complement and 1.5-fold less long rpoS transcript than the mock complement (Table 2, Fig. 3c, d). Therefore, BosR appears to have both activated total rpoS transcription and repressed long rpoS transcription. A second study gave similar results (Table S6). Student's t-test was used to evaluate the data from the three studies which compared the bosR complement (LN101) with a bosR mutant (L9 or LN1). This analysis showed that the increased expression of total rpoS transcript by the bosR complement was highly statistically significant (Fig. 4c; P < 0.01) while the decreased expression of the long rpoS transcript by the bosR complement was also statistically significant (Fig. 4d; P = 0.03). These data indicate that BosR both activates total rpoS transcription and represses long rpoS transcription.

Spirochaetes expressing BosR, but not BosRR39K, activate rpoS transcription

The next question was: what effect did BosRR39K have on rpoS transcription? The results of Experiment II showed that the bosRR39K complement produced about the same level of ospC transcript as the mock complement, indicating that, unlike BosR, BosRR39K did not upregulate ospC (Table 2). These results also showed that the bosRR39K complement produced 1.7-fold less total rpoS transcript than the mock complement, and therefore while BosR activated total rpoS transcription, BosRR39K did not (Table 2, Fig. 3c). A second study gave similar results (Table S6). Taken together, the combined data from the two studies gave a P value of 0.53, according to Student's t-test. These data indicate that, unlike BosR, BosRR39K did not activate total rpoS transcription.

The results for the long rpoS transcript were less clear. Experiment II showed that the bosRR39K complement produced less long rpoS transcript than the mock complement (Table 2). However, the second study showed that the bosRR39K complement produced a similar level of long rpoS transcript as the mock complement (Table S6). In neither study was the difference between LN202 and LN1 statistically significant (Tables 2 and S6). Taken together, the combined data gave a P value of 0.40, according to Student's t-test. These data suggest that BosRR39K did not repress long rpoS transcription; however, further studies are needed to clarify this point.

Experiment III was designed to gain a better picture of how BosR regulates rpoS transcription. Strains LN101 and LN202 were grown in parallel under conditions conducive to upregulation of rpoS. Spirochaetes were harvested during mid-exponential phase (6 day cultures), late-exponential phase (7 day cultures) and stationary phase (8 day cultures) and the RNA was isolated and analysed by qRT-PCR. Table S7 shows the results of this analysis. Plots of the rpoS data also appear in Fig. 3(e, f).

As expected, strains LN101 and LN202 expressed similar levels of bb0646 transcript and similar levels of bosR or bosRR39K transcript (Table S7). Also as expected, strain LN101 expressed high levels of ospC transcript while strain LN202 expressed trace levels of ospC transcript (Table S7). Compared with strain LN202, strain LN101 expressed 4.7-fold more total rpoS transcript on day 6, 5.2-fold more on day 7 and 5.8-fold more on day 8 (Table S7). And, compared with strain LN202, strain LN101 expressed 1.1-fold less long rpoS transcript on day 6, 1.3-fold less on day 7 and 1.6-fold less on day 8 (Table S7). Thus, while strains LN101 and LN202 differed greatly in their production of total rpoS transcript, they differed less in their production of long rpoS transcript.

Strains LN101 and LN202 were also compared in Experiment II (see Tables 2 and S6). The results in Table 2 show that, compared with strain LN202, strain LN101 produced 12.7-fold more total rpoS transcript and 1.3-fold more long rpoS transcript. The results in Table S6 show that, compared with strain LN202, strain LN101 produced 4.2-fold more total rpoS transcript and 1.3-fold less long rpoS transcript. Taken together, the combined data from all three studies showed that strains LN101 and LN202 differed statistically in their expression of total rpoS transcript (Fig. 4e, P < 0.01), but not in their expression of long rpoS transcript (Fig. 4f, P = 0.23). Of course, BosR and BosRR39K may both regulate long rpoS transcription, but by different means.

Fig. 3(e, f) shows the same rpoS data as Table S7, but expressed in different form. In the plotted data, all of the results are expressed relative to the day 6 – LN202 data point (which is arbitrarily assigned a value of 1). It is clear from the plotted data that the BosR-expressing spirochaetes increased their expression of total rpoS transcript during the transition from mid-exponential phase to stationary phase (Fig. 3e), yet maintained a fairly constant level of expression of long rpoS transcript (Fig. 3f). Because the level of long rpoS transcript remained essentially constant, it is clear that the increase in total rpoS transcript was owing to short rpoS transcript. These data provide direct evidence that BosR activates short rpoS transcription while BosRR39K does not.

BosR, but not BosRR39K, binds the DR sequence in vitro

Ouyang et al. (2011) reported that recombinant BosR bound the DR sequence TAAATTAAAT within the rpoS promoter from strain 297. The rpoS promoter in strain B31 differs slightly from that in strain 297 (Fig. S5); however, it still contains all four DR sequences identified in the 297 promoter (Ouyang et al., 2011). To test binding to the B31 promoter, recombinant BosR and BosRR39K were prepared in parallel. Both proteins ran with an estimated mass of 20.5 kDa on SDS-PAGE and appeared free of major contamination (Fig. S6). The results of gel shift assays showed that although both proteins bound the B31 promoter, BosR showed a different pattern of binding than BosRR39K (Fig. 5a). Because competitor DNA was not added to the assay, both proteins may have bound the DNA non-specifically. However, at low concentration (8 nM), BosR generated three shifted bands while BosRR39K generated a single shifted band (Fig. 5a). The same results were obtained when the two proteins were assayed together (Fig. S7).

Fig. 5.

Binding of BosR and BosRR39K to the DR sequence located within the rpoS promoter. (a) Strain B31 rpoS promoter DNA (270 bp) was assayed at 2.5 nM with BosR or BosRR39K at the concentrations indicated (0–244 nM). (b) DR sequence TAAATTAAAT located immediately upstream of the RpoN site within the rpoS promoter (Fig. S5). (c) BbDR target DNA (257 bp), which contained the DR insert, was assayed at 2.5 nM with BosR or BosRR39K at the concentrations indicated (0–319 nM). (d) Vector control DNA (229 bp), which contained no insert, was assayed at 2.5 nM with BosR or BosRR39K at the concentrations indicated (0–319 nM). Arrows indicate the complexes formed between the target DNA and BosR or BosRR39K.

The putative BosR binding site located immediately upstream of the − 24/ − 12 RpoN site contains a perfect DR sequence in both strains (Fig. S5). This site (Fig. 5b) was inserted into the EcoRV cloning site of pST-Blue1 and the insert plus flanking vector sequence was PCR-amplified to generate BbDR target DNA for testing (Fig. S8). The vector sequence with no insert was also PCR amplified to provide control DNA (Fig. S8). The BbDR target DNA was 257 bp. The control DNA was 229 bp. Thus, the insert (28 bp) accounted for ∼10 % of the entire length. The results of the gel shift assay showed that BosRR39K and BosR both bound the BbDR target DNA (Fig. 5c) and both also bound the vector control DNA (Fig. 5d). However, while BosRR39K showed the same pattern of binding for both targets, BosR showed a different pattern of binding. Thus, at 10 nM, BosR generated a major shifted band with the BbDR target DNA (Fig. 5c, arrow) that was not present in the assay with the vector control (Fig. 5d). These data provide preliminary evidence that BosR bound the BbDR insert while BosRR39K did not.

Discussion

This study shows that a single nucleotide change in the bosR gene to generate bosRR39K had the effect of rendering B. burgdorferi non-infectious in mice. Others have shown that bosR is required for B. burgdorferi to activate expression of the alternative sigma factor RpoS (Hyde et al., 2009; Ouyang et al., 2009). Because RpoS controls the expression of virulence factors needed for mammalian infection (Radolf et al., 2012), failure to upregulate rpoS could account for why the spirochaetes expressing BosRR39K were non-infectious. Assays carried out to determine the level of rpoS transcript present in the spirochaetes grown under conditions conducive to rpoS expression found that while the BosR-expressing spirochaetes did transcribe rpoS, the BosRR39K-expressing spirochaetes did not. Furthermore, the BosR-expressing spirochaetes produced high levels of ospC transcript and OspC protein while the BosRR39K-expressing spirochaetes did not. Transcription of ospC is known to require RpoS (Radolf et al., 2012). In vivo, OspC is expressed on the surface of the spirochaete at the point when the spirochaete moves from the gut of the tick into the haemolymph just prior to invading a mammalian host (Grimm et al., 2004; Pal et al., 2004; Tilly et al., 2006). Therefore, although genes other than rpoS may have had a role in making the BosRR39K-expressing spirochaetes non-infectious, failure to upregulate rpoS appears to have been a major cause.

Studies published in 2009–2010 established that BosR controls expression of rpoS in B. burgdorferi (Samuels & Radolf, 2009). Ouyang et al. (2011) identified sites within the promoter of strain 297 rpoS that bound BosR in vitro. They observed that these sites contained near matches to the DR sequence TAAATTAAAT and suggested that BosR activated rpoS transcription by binding one or more of these sites in vivo. Later, Ouyang et al. (2014, 2015) redefined this putative BosR binding site as a palindromic 6-1-6 inverted repeat consisting of the sequence ATTTAA-TTAAAT. They termed this sequence the BosR box. If BosR does bind this BosR box in vivo, it is still unclear how this binding activates rpoS transcription.

The preliminary results of the gel shift assays suggested that only BosR bound the BbDR insert. However, because the BbDR insert contained both a perfect DR sequence (tTTTAAATTAAAT) and an imperfect BosR box (tTTTAAATTAAAT), it is not possible to distinguish which of these sequences BosR recognized.

The protein families database (Pfam) lists BosR as belonging to the FUR family of DNA-binding proteins (Finn et al., 2014). Pfam alignments of FUR proteins (PF01475) show that BosR's Arg39 aligns with a highly conserved arginine present in virtually all FUR proteins. X-ray crystallography studies show that FUR proteins consist of two domains: an N-terminal DNA-binding domain and a C-terminal dimerization domain (An et al., 2009; Butcher et al., 2012; Dian et al., 2011; Gilston et al., 2014; Jacquamet et al., 2009; Lin et al., 2014; Lucarelli et al., 2007; Makthal et al., 2013; Pohl et al., 2003; Sheikh & Taylor, 2009; Shin et al., 2011; Traoré et al., 2006, 2009). The DNA-binding domain contains a winged helix–turn–helix (wHTH) motif that provides the protein with the means to bind DNA (Pohl et al., 2003). The wHTH motif contains a three-helix bundle: the conserved arginine that aligns with BosR's Arg39 is located near the N-terminal end of α-helix 1 (Pohl et al., 2003). Typically, residues within the recognition helix (α-helix 3) determine the binding specificity of wHTH proteins and therefore Arg39 would not be expected to be involved in defining the binding specificity of BosR (Aravind et al., 2005; Huffman & Brennan, 2002). However, others have suggested that positively charged residues located within the N-terminal region of FUR proteins are likely to form electrostatic interactions with their target DNAs (Butcher et al., 2012; Lucarelli et al., 2007; Pecqueur et al., 2006). Recently, Streptococcus pyogenes PerR was found to require Arg21, Arg26 and Arg31, as well as other positively charged residues, to bind DNA in vitro (Lin et al., 2014). Of interest here, Arg31 is the conserved arginine in PerR that aligns with BosR's Arg39.

Models of BosR and BosRR39K (Fig. S9) were built using the S. pyogenes PerR as template (Lin et al., 2014). In these models, Arg39 and Lys39 were each located at the base of a pocket in a position adjacent to α-helix 3 (Fig. S9a). The electrostatic surface potential was determined for both proteins and showed that the surfaces appeared virtually identical (Fig. S9b).

In 2014, Gilston and coworkers published the crystal structure of E. coli Zur bound to DNA (Gilston et al., 2014). Included in their data was the observation that Zur's Arg28 was located sufficiently close to the DNA to allow an electrostatic interaction to form between the arginine side chain and phosphate backbone. Of interest here, Arg28 is the conserved arginine in Zur that aligns with BosR's Arg39.

Because arginine and lysine are both positively charged at neutral pH, both are capable of forming electrostatic interactions with the phosphate backbone of DNA. However, arginine differs from lysine in that it more readily inserts itself into narrowed minor grooves and can therefore be used for protein–DNA recognition (Rohs et al., 2009). Narrow minor grooves exist in A-tract DNA. The DR sequence has two A-tracts: TAAATTAAAT and TAAATTAAAT. The BosR box also has two A-tracts: ATTTAA-TTAAAT and ATTTAA-TTAAAT. If Arg39 were to insert itself into one or more of these A-tracts and thereby help BosR to bind this DNA, then substitution of lysine for this arginine could potentially cause a loss in binding affinity and/or specificity. Further studies are needed to determine if BosR uses Arg39 to interact with the A-tracts in the DR/BosR box DNA.

The results of these studies show that BosR's regulation of rpoS transcription is more complex than originally imagined. BosR repressed long rpoS transcription while at the same time activating short rpoS transcription. It is unclear if these two activities are linked; however, because these results may provide insight into how BosR controls rpoS transcription, I offer the following model for consideration. When BosR is present at low concentration (or when BosR is unable to bind the DR/BosR box), RNA polymerase initiates transcription from an upstream RpoD-dependent promoter and proceeds unhindered along the DNA to synthesize the long rpoS transcript (Fig. 6, top). However, when BosR is present at high concentration (or when BosR is able to bind the DR/BosR box), RNA polymerase still initiates transcription at the upstream promoter but is blocked by bound BosR and therefore fails to progress past the DR/BosR box site. RpoN is now free to bind at the − 24/ − 12 site. RNA polymerase initiates transcription at the RpoN-dependent promoter and proceeds to synthesize the short rpoS transcript (Fig. 6, bottom). In this model, the spirochaete employs transcriptional interference (Palmer et al., 2009, 2011; Shearwin et al., 2005) to control (i.e. close down) synthesis of the short rpoS transcript. The role of BosR is to act as a roadblock (King et al., 2003) and thereby reverse this control.

Fig. 6.

Model of BosR regulation of rpoS transcription. RpoD-dependent transcription from an upstream promoter produces a long transcript in the absence of BosR (top). RpoN-dependent transcription from a downstream promoter produces a short transcript in the presence of BosR (bottom). BosR is pictured binding to all four DR sequences within the rpoS promoter; however, binding to all four sequences may not be required to allow transcription to proceed from the downstream promoter. The extent of long versus short transcription is indicated by the line thickness.

Supplementary Data

Supplementary Data

Abbreviations:

kan

kanamycin resistance

qRT-PCR

quantitative real-time PCR

str

streptomycin resistance