|Home | About | Journals | Submit | Contact Us | Français|
Borrelia burgdorferi (Bb) adapts to its arthropod and mammalian hosts by altering its transcriptional and antigenic profiles in response to environmental signals associated with each of these milieus. In studies presented here, we provide evidence to suggest that mammalian host signals are important for modulating and maintaining both the positive and negative aspects of mammalian host adaptation mediated by the alternative sigma factor RpoS in Bb. Although considerable overlap was observed between genes induced by RpoS during growth within the mammalian host and following temperature-shift, comparative microarray analyses demonstrated unequivocally that RpoS-mediated repression requires mammalian host-specific signals. A substantial portion of the in vivo RpoS regulon was uniquely upregulated within dialysis membrane chambers, further underscoring the importance of host-derived environmental stimuli for differential gene expression in Bb. Expression profiling of genes within the RpoS regulon by quantitative reverse transcription polymerase chain reaction (qRT-PCR) revealed a level of complexity to RpoS-dependent gene regulation beyond that observed by microarray, including a broad range of expression levels and the presence of genes whose expression is only partially dependent on RpoS. Analysis of Bb-infected ticks by qRT-PCR established that expression of rpoS is induced during the nymphal blood meal but not within unfed nymphs or engorged larvae. Together, these data have led us to postulate that RpoS acts as a gatekeeper for the reciprocal regulation of genes involved in the establishment of infection within the mammalian host and the maintenance of spirochetes within the arthropod vector.
In order to be maintained within its enzootic cycle, Borrelia burgdorferi (Bb), the causative agent of Lyme disease, must be able to grow within two distinctly different environments, the arthropod vector (Ixodes ticks) and a mammalian host, typically small rodents. Over the past decade, convincing evidence has emerged to that Bb accomplishes this feat by altering its transcriptional and polypeptide profiles in response to the environmental signals associated with each of these milieus, a dynamic process collectively referred to as host adaptation (Akins et al., 1998; Anguita et al., 2003; Pal and Fikrig, 2003; Stevenson et al., 2006). The transcriptome changes associated with growth in vivo have been of particular interest as it is presumed that the polypeptides encoded by these differentially-expressed loci are required to fulfil virulence-related functions or the physiological adjustment to growth within the mammalian host. Although the majority of differential-expression studies in Bb have focused on genes whose expression is increased during infection, the downregulation of loci involved in the tick phase of the enzootic cycle is believed to be an equally important aspect of the spirochete's pathogenic strategy (Brooks et al., 2003; Pal et al., 2004; Yang et al., 2004; Caimano et al., 2005; Strother et al., 2007).
Utilization of alternative sigma factors to redirect transcription in response to environmental stimuli has emerged as an important mechanism for stress adaptation in many bacteria, including pathogens (Molofsky and Swanson, 2004; Kazmierczak et al., 2005; Rodrigue et al., 2006). One of the best-studied examples of a global stress response is the RpoS regulon in Escherichia coli, which co-ordinates the expression of genes involved in adaptation to nutrient limitation, acid shock, reactive oxygen species and UV irradiation (Hengge-Aronis, 2002). The annotated genomic sequence of Bb type strain B31-MI revealed three sigma factors, the housekeeping sigma factor σ70 (RpoD, BB0712), and two alternative sigma factors, RpoS (BB0771) and RpoN (BB0450) (Fraser et al., 1997). In a series of seminal studies, Norgard and co-workers (Hubner et al., 2001; Yang et al., 2003; Smith et al., 2007) demonstrated that RpoN acts in concert with the response regulator protein Rrp2, to induce the expression of RpoS, which in turn is required for the transcription of ospC and dbpBA. In E. coli, RpoS levels are regulated by a complex network of transcriptional, post-transcriptional and translational mechanisms as cells enter stationary phase (Hengge-Aronis, 2002). In Bb, RpoS appears to be controlled primarily at the transcriptional level and can be induced in exponentially growing cultures by temperature-shift (Caimano et al., 2004); efficient translation of rpoS mRNA during early logarithmic growth also requires the small RNA, DsrA (Lybecker and Samuels, 2007). While no study, thus far, has examined expression of rpoS in Bb-infected ticks or mice, several RpoS-dependent loci (i.e. ospC, dbpBA, ospF) are known to be induced during the nymphal blood meal and continue to be expressed during infection (Schwan and Piesman, 2000; Gilmore et al., 2001; Hefty et al., 2001; McDowell et al., 2001; Ohnishi et al., 2001; Hodzic et al., 2002; Liang et al., 2002; Tokarz et al., 2004). Previously, we demonstrated that RpoS also is required for the downregulation of at least two tick phase-associated lipoproteins, OspA and Lp6.6, in response to mammalian host signals (Caimano et al., 2005). The loss of RpoS does not significantly impair the ability of Bb to survive diverse environmental stressors, including growth within dialysis membrane chambers (DMCs), but it does render spirochetes avirulent in mice (Caimano et al., 2004). Together, these findings led us to postulate that, rather than serving as a global stress response regulator, RpoS in Bb orchestrates the expression of a subset of genes associated with the mammalian host-adaptation process (Caimano et al., 2004; 2005).
Microarrays have proven to be a valuable tool for delineating the RpoS regulons of other microorganisms (Lacour and Landini, 2004; Patten et al., 2004; Schuster et al., 2004; Yildiz et al., 2004; Weber et al., 2005; Nunez et al., 2006; Rahman et al., 2006). While microarray technology has been used to examine differential gene expression in Bb cultivated under different environmental conditions (Liang et al., 2002; Revel et al., 2002; Brooks et al., 2003; Ojaimi et al., 2003; Tokarz et al., 2004; Fisher et al., 2005; Hyde et al., 2006), only one study (Fisher et al., 2005) has examined the contribution of alternative sigma factors to gene expression in Bb on a genome-wide scale, and then, only under standard in vitro growth conditions. The extent to which the manipulation of in vitro growth conditions can be used to reproduce the full spectrum of mammalian host signals that trigger differential gene expression in Bb during tick transmission remains uncertain (Stevenson et al., 2006). Studies by us (Akins et al., 1995; 1998; Lahdenne et al., 1997; Hefty et al., 2001; Parveen et al., 2003; Caimano et al., 2004; 2005) and others (Revel et al., 2002; Brooks et al., 2003; Tokarz et al., 2004) indicate that the RpoS regulon could differ substantially during growth within the mammalian host and collectively underscore the importance of examining the RpoS regulon under conditions that approximate the environmental stimuli encountered by Bb during natural infection. Indeed, comparative microarray analyses of RpoS-dependent gene expression within spiro-chetes grown in vitro following temperature-shift and within DMCs performed as part of this study identified a large group of genes that are uniquely up- and downregulated by RpoS in response to mammalian host signals. Expression profiling of a subset of RpoS-dependent genes by quantitative reverse transcription polymerase chain reaction (qRT-PCR) revealed a level of complexity to gene regulation by this alternative sigma factor beyond that observed by microarray, most notably, the identification of genes whose expression is only partially dependent on RpoS, particularly in vivo. Based on our analysis of rpoS expression in Bb-infected Ixodes scapularis ticks, we have formulated a temporal framework for RpoS function and RpoS-dependent gene regulation during the spirochete's enzootic cycle.
In order to define the experimental conditions for our microarray analysis of the RpoS regulon, we began by examining the extent to which the manipulation of in vitro conditions influences the expression of proteins known to be under the control of this alternative sigma factor. Bb strain 297 wild-type clone c162 was temperature-shifted from 23°C to 37°C into BSK-H medium at standard or reduced pH (7.5 versus 6.8 respectively) and grown to late-logarithmic phase for comparison with lysates of this same isolate cultivated within DMCs (Fig. 1A). In agreement with previous studies (Schwan et al., 1995; Carroll et al., 1999; 2000; Yang et al., 2000; Hubner et al., 2001; Revel et al., 2002; Caimano et al., 2004; 2005), expression of OspC and DbpA was induced by temperature-shift and enhanced further by a combination of increased temperature and reduced pH. Cultivation of Bb c162 following temperature-shift at reduced pH, on the other hand, did not noticeably diminish expression of either OspA or Lp6.6 as has been previously reported (Yang et al., 2000; Revel et al., 2002). Because spirochetes within DMCs replicate under conditions in which physiologic pH is strictly maintained by homeostatic mechanisms (Caimano et al., 2004), the results shown in Fig. 1A demonstrate unequivocally that a decrease in pH is not a prerequisite for the downregulation of OspA and Lp6.6. Temperature-shift, with or without reduced pH, also failed to reproduce changes in the low molecular mass region of the spiro-chete's polypeptide profile (~20 kDa) that are characteristic of DMC-cultivated organisms (Akins et al., 1998; Caimano et al., 2005). Gilbert et al. (2007) recently developed a Bb strain 297 isolate, RpoSi, in which expression of rpoS is tightly controlled by a synthetic flgB promoter modified to include the lac operator. Using this isolate, we examined the effect of RpoS expression on the polypeptide profiles of Bb subjected to otherwise identical in vitro growth conditions. As shown in Fig. 1A, induction of rpoS resulted in the increased expression of OspC and DbpA without any noticeable downregulation of OspA and Lp6.6.
Schwan and Piesman (2000) reported that expression of OspC diminishes when spirochetes are serially passaged at 37°C. This finding, reproduced in the present study (Fig. 1B), indicates that spirochetes become desensitized to the ability of a thermal stimulus to induce RpoS-dependent changes in gene expression. We next used continuous cultivation of Bb within DMCs to model the prolonged exposure to mammalian host-derived signals that would occur during natural infection. In contrast to their in vitro passaged counterparts, Bb serially passaged within DMCs continuously expressed high levels of OspC as well as diminished levels of OspA (Fig. 1B); the host-adapted phenotype was maintained for nine serial passages extending over approximately a 4-month time period (data not shown).
Microarray methodology has been used to delineate the RpoS regulon in Bb strain B31-MI under standard in vitro growth conditions (34°C in BSK-H) (Fisher et al., 2005). Based on the studies described above, we postulated that the analysis of in vitro cultivated Bb would yield only a partial representation of the RpoS regulon as it exists within the mammalian host. For this reason, we performed microarray analyses of wild-type and RpoS-deficient Bb cultivated within DMCs. In order to gain a better understanding of how the composition of the RpoS regulon differs in response to mammalian host-specific signals, we extended these studies to include parallel microarray analyses of strain 297 wild-type and rpoS mutant Bb following temperature-shift in vitro.
For the gene expression analyses described below, we employed a 70-mer oligonucleotide microarray that contains gene targets for 1741 putative open reading frames (ORFs) (Terekhova et al., 2006) based on the genomic sequence of Bb strain B31-MI (Fraser et al., 1997). We first confirmed the suitability of these microarrays for use with our extensively characterized strain 297 wild-type and rpoS mutant isolates (Caimano et al., 2004; 2005; Eggers et al., 2004; 2006) by performing comparative hybridizations using genomic DNAs from both B31-MI and 297 wild-type isolates. Consistent with our earlier report demonstrating high genetic similarity between B31-MI and 297 (Terekhova et al., 2006), only 101 of the borrelial gene targets present on the microarrays gave significantly lower hybridization signals for 297 genomic DNA compared with B31-MI (mean log2 ratio ≤ −1.5) (Table S1). The single non-hybridizing chromosomal gene, bb0845.1, is annotated as a non-functional pseudogene. Two-thirds (66/100) of the B31-MI plasmid-borne genes that gave lower hybridization signals for 297 are found on plasmids known to be either truncated (i.e. lp36) or absent (i.e. lp38 and lp56) in the latter strain (Eggers et al., 2002; Iyer et al., 2003; Terekhova et al., 2006). The absence of a hybridization signal for ospC was expected given the low sequence identity (73%) between the bbb19/ospC 70-mer oligo and the corresponding region of the strain 297 ospC orthologue. For the same reason, we were unable to detect any of the strain B31 ospF alleles (bbm38, bbo39, bbr42 and bbs41); however, the RpoS dependence of the ospF paralogues in vitro and within DMCs already has been established (Caimano et al., 2004; Eggers et al., 2006).
Using the statistical criteria outlined in Experimental procedures, a total of 137 genes were identified as being differentially regulated by RpoS during cultivation of Bb within DMCs. One hundred and three genes were expressed at significantly lower levels in the rpoS mutant, implying that RpoS contributes, either directly or indirectly, to their transcription (Table 1), while 34 genes were downregulated in the wild type, implying that RpoS is required for their repression (Table 2). Consistent with previous microarray studies (Revel et al., 2002; Brooks et al., 2003; Ojaimi et al., 2003; Tokarz et al., 2004), the majority of differentially expressed genes were localized to plasmid replicons (81% of upregulated genes and 73% of RpoS-repressed genes) (Fig. 2).
The established expression profiles of known RpoS-dependent genes such as ospC, dbpA and ospF suggest that many genes within the upregulated portion of the DMC RpoS regulon are induced during the tick-to-mammal transition and/or early infection. Indeed, the upregulation of OspC in response to the blood meal within a Bb-infected tick is critical for the establishment of infection (Ohnishi et al., 2001; Grimm et al., 2004; Stewart et al., 2006; Tilly et al., 2006). While the overwhelming majority (78/103) of genes upregulated by RpoS within DMCs encode hypothetical proteins and conserved hypothetical proteins, a survey of the functionally annotated genes upregulated by RpoS revealed several additional genes whose products may contribute to establishing infection or perform virulence-related functions. Key among these are four genes related to chemotaxis: bb0565 encodes the purine-binding chemotaxis protein CheW-2, while bb0578, bb0680 and bb0681 encode the methyl-accepting chemotaxis proteins Mcp1, Mcp4 and Mcp5 respectively. Although annotated as hypothetical proteins, bba36 and bbi42 have been shown to encode surface-exposed lipoproteins (Brooks et al., 2006), and therefore may function at the pathogen–host interface.
In E. coli, many of the genes controlled by RpoS encode proteins with annotated functions related to metabolism, nutrient transport and/or stress adaptation (Schellhorn et al., 1998; Patten et al., 2004; Weber et al., 2005; Rahman et al., 2006). In contrast, only a handful of borrelial genes (bb0116, bb0447, bb0548, bb0728, bb0812 and bba34) with predicted roles in spirochete metabolism and/or physiology were upregulated by RpoS within DMCs. bb0116/malX encodes a putative maltose/glucose-specific EIICB transporter component of the carbohydrate phosphotransferase system (PTS). bb0447, annotated as a Na+/H+ antiporter, also contains a putative fructose/mannitol-specific IIA PTS subunit domain involved in sugar transport and/or phosphorylation. Two genes, bb0812 and bb0728, upregulated by RpoS within DMCs encode enzymes involved either in the synthesis or utilization of coenzyme A. bb0812 encodes a putative DNA/pantothenate flavoprotein involved in the synthesis of coenzyme A (Kupke et al., 2000), while bb0728 encodes a NAD-dependent coenzyme A disulphide reductase (Cdr) thought to be involved in maintaining intracellular redox potential, as well as adaptation to oxidative stress (Boylan et al., 2006). bb0728 is predicted by TIGR to be part of a bicistronic operon with bb0729, encoding a glutamate transporter (GltP); we confirmed the co-transcription of these genes by RT-PCR using primers (Table S3) that amplify across the bb0729-bb0728 junction (data not shown). Examination of the raw microarray data for bb0729 revealed that this gene also was expressed at greater levels in the wild type compared with the rpoS mutant (4.57-fold, P = 0.0057) but fell just below the threshold required for statistical significance and was excluded from the DMC data set (see Experimental procedures). In addition to providing Bb with an essential nutrient, GltP also may provide a means of maintaining the cell's turgor in response to high osmolality by increasing the intracellular concentration of glutamate (Csonka, 1989). bb0548/polA encodes DNA polymerase I, the enzyme responsible for removing the RNA primer during DNA synthesis (Kornberg and Baker, 1992); the increased expression of this gene also was observed following the addition of human blood to the growth medium in vitro (Tokarz et al., 2004). Interestingly, only one metabolic gene upregulated by RpoS within DMCs is plasmid-borne. bba34, located on lp54, encodes OppA5, one of five periplasmic oligopeptide (Opp)-binding proteins in Bb (Bono et al., 1998; Wang et al., 2002; Medrano et al., 2007). Medrano et al. (2007) recently confirmed the RpoS dependence of oppA5 in vitro.
The majority of the 83 plasmid-borne genes upregulated by RpoS within DMCs localized to lp54, lp28-2 and members of the cp32 plasmid family (Fig. 2). Twenty of the 83 are distributed along the length of lp54. Although lp54 is known to serve as a depot for differentially expressed genes (Revel et al., 2002; Brooks et al., 2003; Ojaimi et al., 2003; Tokarz et al., 2004; Fisher et al., 2005), this report is the first to recognize the RpoS dependence of many of these. In addition to bba24/dbPA, bba25/dbpB and bba34/oppA5, RpoS controls the expression of bba04 and bba05, encoding the S2 and S1 lipoprotein serodiagnostic antigens respectively (Feng et al., 1996). A cluster of six (bba64, bba65, bba66, bba71, bba72 and bba73) genes, all exhibiting > 150-fold induction by RpoS (Table 1), is located at the right end of lp54 (Fig. 3). Recently, these genes have garnered considerable interest due to their differential expression in response to temperature, pH, and in vivo signals, and their potential role in virulence (Anguita et al., 2000; Carroll et al., 2000; Liang et al., 2002; Brooks et al., 2003; Ojaimi et al., 2003; Tokarz et al., 2004; Clifton et al., 2006; Gilmore et al., 2007). With the exception of bba71, each is predicted to encode a lipoprotein. Once considered to be members of a single paralogous gene family (pgf54), all but two (bba71 and bba72) are now designated by TIGR as being non-paralogous. While their similar expression profiles and close proximity on lp54 might suggest that these genes are co-transcribed, Northern blot analyses identified individual mRNA bands corresponding to bba64, bba65, bba66 and bba73 (Carroll et al., 2000); the RpoS dependence of one of these, BBA66, recently has been confirmed at the protein level (Clifton et al., 2006). lp28-2, a plasmid not previously recognized as a repository for differentially expressed genes, harbours 17 of the 83 plasmid-borne genes upregulated by RpoS within DMCs (Fig. 2), all of which encode hypothetical proteins (Table 1). An examination of the intergenic spacing between each of the RpoS-dependent lp28-2-borne genes suggests that at least some of these may be co-transcribed (e.g. bbg13-12, bbg21-bbg16 and bbg24-23). Twenty-three genes upregulated by RpoS are distributed among the members of the cp32 family of circular plasmids (Fig. 2). The majority of these cp32-borne genes encode proteins of unknown function, with half belonging to just two paralogous families, gbb fam_PF06381 and gbb fam_PF02989, located within the putative late phage operon of the cp32-prophage ϕBB-1 (Eggers et al., 2000; Zhang and Marconi, 2005).
Thirty-four genes were identified as being downregulated by RpoS within DMCs. Based on the established expression patterns of OspA and Lp6.6 (Schwan et al., 1995; Lahdenne et al., 1997; Akins et al., 1998; Schwan and Piesman, 2000; Ohnishi et al., 2001), we hypothesize that many of these genes are induced following acquisition by naïve larvae and then expressed constitutively within flat nymphs until the next blood meal. As with the RpoS upregulated genes, the majority (21/34) of the genes subject to RpoS-mediated repression encode hypothetical proteins or conserved hypothetical proteins. Ten genes (bb0365, bba62, bba69, bbf20, bbg01, bbi29, bbi36, bbi38, bbi39, bbk01 and bbk19) downregulated by RpoS within DMCs, in addition to ospA and ospB, are known or predicted to encode lipoproteins. Four of the genes downregulated by RpoS within DMCs are involved in glycerol utilization. The first three, bb0240, bb0241 and bb0243, are part of an operon (Ojaimi et al., 2003) encoding a putative glycerol uptake facilitator (bb0240/glpF), a glycerol kinase (bb0241/glpK) and an anaerobic glycerol 3-phosphate dehydrogenase (bb0243/glpA) respectively; bb0242, which was also downregulated by RpoS, encodes a protein of unknown function. The co-ordinated downregulation of these four genes is consistent with a previous study noting the increased expression of bb0240-b0243 at 23°C (Ojaimi et al., 2003). The fourth gene related to glycerol utilization, bb0368, encodes a putative NAD(P)-dependent glycerol 3-phosphate dehydrogenase. bb0582, encoding a putative D-alanyl, D-alanine carboxypeptidase, may be involved in recycling of peptidoglycan and/or maintaining cell shape (Young, 2003). bbk17, the only plasmid-borne metabolic gene downregulated by RpoS, encodes an adenine deaminase (AdeC) (Jewett et al., 2007) that may be involved in either salvaging adenine for purine nucleotide biosynthesis or providing sources of nitrogen and pentose sugars (Endo et al., 1983; Nygaard et al., 1996; Tozzi et al., 2006).
As with the upregulated portion of the RpoS regulon, a high percentage (71%) of the plasmid-borne genes downregulated by RpoS localized to a small number of replicons (Fig. 2). More than half (13/34) are located on lp36, all but two of which encode proteins of unknown function (Table 2). An examination of the transcriptional orientation and intergenic spacing between the lp36-borne genes suggests at least some may be co-transcribed (e.g. bbk23-21 and bbk33-35); for these, downregulation by RpoS would likely be mediated via repression at the upstream promoter for the first gene in each putative operon. lp54, in addition to harbouring ospA, ospB and lp6.6, contains two other downregulated genes, bba74 and bba69 (Fig. 3). The protein product of bba74, annotated as an outer membrane-associated porin (Oms28) (Skare et al., 1996; 1997; Fraser et al., 1997), has recently been shown to lack the β-barrel structure typical of outer membrane porins and localized to the periplasm (Mulay, 2007; Mulay et al., 2007). bba69 is one of five putative lipoprotein genes downregulated by RpoS belonging to the gbb pep_35 paralogous protein family. Three of the remaining downregulated gbb pep_35 genes (bbi36, bbi38 and bbi39) are located on lp28-4, while the location of the fifth, bbj41, in strain 297 is uncertain as this isolate lacks lp38 (Iyer et al., 2003; Terekhova et al., 2006). Although originally annotated as belonging to paralogous family 54 (pfg54) (Fraser et al., 1997; Casjens et al., 2000), all five of these genes have been placed by TIGR into pep_35 based on phylogenetic analyses (http://cmr.tigr.org/tigrscripts/CMR/ParalogDescription.cgi?align_id=147991&align_name=gbb%20fam_b_burgdorferi_b31.pep_35&ori_db=gbb). In contrast to the large number of RpoS-upregulated genes on lp28-2, only one gene on this replicon, bbg01, was downregulated by RpoS within DMCs.
We recently proposed that repression of ospA and lp6.6 by RpoS is mediated by the binding of a repressor protein to the polyT tracts located just upstream of the −35 sequences for these σ70-dependent promoters (Sohaskey et al., 1999; Caimano et al., 2005). Using the ospA and lp6.6 promoters as guides, we examined the upstream regions of the other downregulated genes (Table 2) for T-rich regions to determine whether these genes are regulated by a common mechanism. Results from these analyses yielded five additional genes with polyT tracts in close proximity to their putative promoters, including bb0240, the first gene in the glp operon, and the adenine deaminase gene bbk17 (Table 3). A T-rich region also was found flanking bba69 but is noticeably further upstream than those of the other in vivo RpoS-repressed genes (Table 3). Analysis of the promoter regions for genes lacking polyT tracts revealed that two of the five downregulated gbb pep_35 paralogous genes, bbi36 and bbi38, contain perfect 18 bp inverted repeats immediately upstream of their putative −35 sequences (Table 3). Of note, the bbi36/bbi38 inverted repeats are dissimilar to the repeat motif recently identified upstream of ospC in strain B31 (Xu et al., 2007).
A comparison of the in vitro- and DMC-derived microarray data sets revealed that 44 of the genes upregulated by RpoS during DMC cultivation also were upregulated following temperature-shift in vitro and are therefore designated as `core' RpoS-dependent genes. While a small number of core genes encode products with known or predicted functions (i.e. DbpA, DbpB, OppA5, Mcp4 and Mcp5), the majority encode hypothetical and conserved hypothetical proteins of unknown function, almost half of which are predicted to be lipoproteins (Table 1). The upregulated portion of the DMC data set contains 59 genes that were not present within the in vitro data set (Table 1). Although the vast majority (48/59) of these encode hypothetical proteins or conserved hypothetical proteins (Table 1), this group includes a small number of genes encoding functions related to chemotaxis and spirochete metabolism (Table 1). Interestingly, all 17 of the lp28-2-borne genes upregulated by RpoS within DMCs were present exclusively within this data set.
The in vitro data set contains 66 genes which were not present within the DMC data set (Table S2); all but two of these, bbk32 and bbq03, exhibited low levels of induction. An examination of the raw data from the DMC-derived microarrays revealed that 21 of the 66 genes present exclusively within the in vitro data set also exhibited a low level of RpoS-dependent upregulation within DMCs (> 2.5-fold, P ≤ 0.01) but their mean log2 ratios fell below the threshold (≥ 4.59-fold) used to screen the DMC microarray data (see Experimental procedures). This finding suggests that some of the genes present exclusively within the in vitro data set also are upregulated by RpoS within DMCs. Indeed, qRT-PCR analyses, presented below, confirmed that this was the case for at least one gene, bb0670 (Fig. 5B). qRT-PCR analyses revealed that bbk32 was expressed at substantially lower levels in the DMC-cultivated wild-type Bb (0.005 copies per 100 flaB) used for the microarray hybridizations compared with their in vitro counterparts (20.79 copies per 100 flaB). These data, however, seemed to be at odds with the demonstrated expression of BBK32 during infection (Lahdenne et al., 2006; Li et al., 2006) and its potential role in virulence (Seshu et al., 2006), prompting us to examine RNAs from additional DMC-cultivated c162 (wild type) and c174 (rpoS mutant). Results from these subsequent analyses yielded substantially higher bbk32 transcript levels (11.13 copies per 100 flaB) in the wild-type isolate and demonstrated a level of RpoS dependence (46.5-fold, P < 0.01) comparable to that observed at 37°C in vitro (data not shown). One of the most striking observations to emerge from these comparative microarray analyses was that no genes were downregulated by RpoS in wild-type Bb following temperature-shift in vitro, providing further evidence that mammalian host-derived signals are essential for RpoS-mediated repression.
Unlike previous microarray studies of the Bb, which used qRT-PCR to measure relative differences in transcripts (i.e., ΔΔCT) (Brooks et al., 2003; Ojaimi et al., 2003; Tokarz et al., 2004; Fisher et al., 2005; Hyde et al., 2006), we opted to quantify transcript copy numbers for selected genes identified by the microarrays using the primer pairs described in Table S3. At the outset, we measured transcript levels for rpoS and two prototypical RpoS-dependent core genes, ospC and dbpA (Hubner et al., 2001), in Bb cultivated in vitro following temperature-shift (37°C) and within DMCs. As expected, expression of ospC and dbpA was absolutely dependent on the presence of RpoS (Fig. 4). Interestingly, we observed only a modest, although significant (P < 0.05), difference between the levels of ospC in wild-type Bb grown at 37°C compared with DMCs despite the presence of markedly greater rpoS transcript levels within DMC-cultivated Bb. Transcript levels for dbpA, on the other hand, were noticeably lower in wild-type Bb during DMC cultivation compared with 37°C (Fig. 4). Using the comparative microarray data as a guide, we selected 21 additional genes, 15 upregulated and six downregulated, for analyses by qRT-PCR, with the majority of these encoding polypeptides with either predicted metabolic functions or potential roles in virulence. Of the five (bb0680, bb0844, bba07, bba34 and bbi42) core genes examined, all but one (bb0680) was found to be absolutely dependent on RpoS, although their transcript copy numbers were markedly lower than those of ospC and dbpA (Fig. 4).
Seven genes were selected from those identified by microarray as being upregulated by RpoS exclusively within DMCs (Table 1). Consistent with their `exclusively in vivo' designation, five of the genes within this group, bb0548, bbg13, bbg17, bbg19 and bbg24, displayed significant RpoS dependence and greater absolute transcript levels within DMC-cultivated Bb compared with their 37°C counterparts (Fig. 5A). The expression profile for the sixth gene, bb0728, encoding Cdr, differed in that its expression was upregulated by RpoS at 37°C as well as within DMCs; upregulation of bb0729, co-transcribed with bb0728, also was RpoS-dependent under both growth conditions. We next selected three genes from those identified by microarray as being upregulated by RpoS exclusively in vitro (Table S2). Although qRT-PCR analyses of all three genes within Bb grown at 37°C were in agreement with the in vitro-derived microarray data, each displayed a distinct expression profile within DMCs. Expression of bb0588 was induced 3.77-fold by RpoS at 37°C but only 1.33-fold within DMCs (Fig. 5B); the lower level of induction for bb0588 observed within DMCs is primarily due to increased RpoS-independent expression. bb0021 was expressed at 2.84-fold higher levels in wild-type compared with rpoS mutant Bb in vitro, but was not RpoS-dependent in vivo (Fig. 5B). Expression of the third gene, bb0670, encoding CheW-3, was significantly upregulated by RpoS both in vitro and within DMCs (Fig. 5B); however, the 3.18-fold induction observed in the DMC-derived microarrays (Table 3), and confirmed here by qRT-PCR, was below the threshold required for inclusion in the DMC data set.
Quantitative RT-PCR analyses of six genes (bb0240, bb0242, bba15, bba62, bbk17 and bba74) selected from those subject to RpoS-mediated repression within DMCs (Table 2) confirmed that each was expressed at significantly lower levels (P ≤ 0.05) in wild-type compared with rpoS mutant Bb within DMCs (Fig. 6). Interestingly, the in vitro expression profiles for these genes differed notably. Rather than being repressed, all six were expressed at high levels in wild-type Bb grown in vitro. Moreover, four of the six genes examined (bb0242, bba62, bba74 and bbk17) were expressed at significantly lower levels in the rpoS mutant compared with wild-type Bb at 37°C (Fig. 6). The transcript differences observed between wild type and the rpoS mutant for bba62, however, did not result in an appreciable difference at the protein level (Fig. 1 and Caimano et al., 2005). As a whole, these findings strike a cautionary note against relying on the use of in vitro conditions alone to predict a given gene's expression pattern during natural infection.
Previous studies in Bb have demonstrated that the sequence of the `extended' −10 region is critical for promoter recognition by RpoS (Eggers et al., 2004; 2006; Yang et al., 2005). In an attempt to develop a consensus sequence for RpoS binding in Bb, we used MEME (http://meme.sdsc.edu) (Bailey et al., 2006) to search for conserved motifs within the upstream promoter regions of two subsets of RpoS-dependent genes. The first consisted of the 400 bp regions found upstream of the translational start codons for the 44 core genes upregulated by RpoS both in vitro and within DMCs (Table 1). The second consisted of the upstream regions for 10 genes whose absolute RpoS dependence has been verified experimentally as part of this study (Fig. 4) or elsewhere (Hubner et al., 2001; Caimano et al., 2004; Clifton et al., 2006; Eggers et al., 2006), including four (ospC, dbpBA, erpK and bba66) with known transcriptional start sites (Marconi et al., 1993; Hagman et al., 1998; Babb et al., 2004; Clifton et al., 2006). In contrast to E. coli, for which MEME successfully identified an RpoS consensus sequence (Weber et al., 2005), this program was unable to identify any conserved motifs in the −10 regions for either of the two RpoS-dependent Bb subsets, most likely because of the high A + T content (70%) of the borrelial genome (Fraser et al., 1997). Visual inspection of upstream regions for the 10 absolutely RpoS-dependent borrelial genes, however, enabled us to identify TG(G/A)(G/A) ATA(T/A)ATT as a putative RpoS consensus extended −10 region sequence in Bb (Table 4). This consensus sequence is somewhat divergent from that of E. coli [TGN0–2CTA(T/C)(A/G)CT] (Lacour and Landini, 2004; Weber et al., 2005), most notably at the nucleotide corresponding to position −15. In E. coli, a C in this position strongly favours RpoS, while a G favours σ70 (Becker and Hengge-Aronis, 2001). The borrelial RpoS consensus sequence contains an A at this position. Interestingly, mutagenesis studies of the ospC promoter suggest that this position may be involved in promoter recognition by RpoS but does not appear to serve as a `discriminator' nucleotide for sigma factor selectivity (Eggers et al., 2004; 2006; Yang et al., 2005). We next examined the upstream regions for 10 genes that were found by qRT-PCR to be recognized by both RpoS and at least one of Bb's other sigma factors (RpoN or σ70) (Fig. 5). We were unable to find putative promoter elements for five (bb0670, bbg13, bbg17, bbg19 and bbg24) of the 10 genes examined, including four that exhibited enhanced RpoS-dependent expression within DMCs (Table 1). The putative extended −10 elements identified in the upstream regions for the remaining five (bb0021, bb0548, bb0588, bb0680 and bb0729–0728), however, were highly similar to the consensus sequence derived from the absolutely RpoS-dependent genes. Putative −35 sequences were identified within the upstream regions for each of the absolutely RpoS-dependent genes, as well as for the five dually transcribed genes that had identifiable −10 regions (Table 4). None of the upstream regions for the dually transcribed genes contained the RpoN promoter consensus sequence described by Studholme et al. (2000) while only two (bb0021 and bb0588) were scored as positive by the SEQSCAN RpoN promoter scoring matrix used by Fisher et al. (2005). Thus, σ70 is likely responsible for the RpoS-independent expression component for the majority of dually transcribed genes.
No study, thus far, has examined the expression of rpoS within Bb-infected I. scapularis ticks. In order to gain a broader understanding of the RpoS regulon and RpoS function during the enzootic cycle, we used qRT-PCR to measure rpoS transcript levels in Bb strain 297-infected I. scapularis nymphs, prior to and during a blood meal on a naïve mouse, and in engorged naïve larvae following a blood meal on a Bb-infected mouse. As shown in Fig. 7, the flaB-normalized transcript levels for rpoS were dramatically increased in Bb within engorged nymphal ticks as compared with the levels observed in either flat nymphs or fed larvae; the marked disparity between the rpoS transcript levels in the two `fed' states was particularly noteworthy. The normalized rpoS transcript levels observed in engorged Bb-infected nymphs (2.66 copies/100 flaB) were comparable to those observed in wild-type Bb cultivated within DMCs (3.02 copies per 100 flaB) (Figs 4 and and77).
In E. coli, the alternative sigma factor RpoS acts as the master regulator of the general stress response as cells enter stationary phase growth and/or encounter adverse environmental conditions (Hengge-Aronis, 1996). In recent years, there has been a growing body of evidence that the RpoS regulon in Bb evolved to facilitate adaptation of the spirochete to environmental conditions quite unlike those encountered by free-living bacteria. Indeed, one of the most striking general observations made in the present study is that the Bb RpoS regulon contains a paucity of genes with annotated physiological or metabolic functions. Rather than functioning as a master regulator of a general stress response, as in E. coli, the borrelial RpoS appears to serve as an activator of genes acting at the tick–mammalian host interface during transmission and at least into early infection (Caimano et al., 2004). Recently, we demonstrated that RpoS also is required for the downregulation of the tick phase-specific lipoproteins OspA and Lp6.6 in response to mammalian host-derived signals (Caimano et al., 2005). Thus, RpoS appears to act as a `gatekeeper' for the co-ordinated reciprocal regulation of borrelial loci involved in the mammalian and arthropod host phases of the enzootic cycle.
The qRT-PCR analyses of a representative subset of genes upregulated by RpoS, presented here, brought to light a degree of regulatory complexity that was not discernible from the relative expression levels provided by microarrays. Most notably, qRT-PCR enabled us to identify two distinct classes of RpoS-dependent genes in Bb. The first consisted of core genes whose expression was absolutely dependent on the presence of RpoS both in vitro and within DMCs. Genes within the second class, in contrast, were expressed at appreciable levels in the absence of RpoS, suggesting that these genes are dually transcribed by both σ70 and RpoS. A comparison of the normalized transcript copy numbers for all RpoS upregulated genes examined by qRT-PCR revealed that their expression levels varied remarkably. The consistently lower transcript levels for the dually transcribed compared to the core genes suggest that the `hybrid' promoter elements for the dually transcribed genes are recognized less efficiently by RNA polymerase compared with those promoters which are recognized by RpoS exclusively. In E. coli, promoter strength and sigma factor selectivity are determined, in large part, by a combination of cis-acting sequence elements, including individual nucleotides both within and outside of the consensus −10 and −35 sites, the spacing between these two elements, DNA supercoiling and the presence of upstream motifs such as UP elements (Kusano et al., 1996; Gourse et al., 2000; Alverson et al., 2003; Bordes et al., 2003; Typas and Hengge, 2005; 2006; Yang et al., 2005; Typas et al., 2007). A comparison of the promoter regions for the core and dually transcribed borrelial genes enabled us to identify a putative RpoS consensus extended −10 motif binding site. The close similarity between the binding sites for RpoS and σ70, combined with the presence of consensus −35 sequences upstream of both the absolutely dependent and dually transcribed genes, however, raises the enigmatic question of why some borrelial RpoS-dependent promoters are recognized by σ70 while others are not. Numerous studies in E. coli have demonstrated that trans-acting factors, such as IHF, CRP, Fis, Lrp and H-NS, also play a crucial role in modulating promoter recognition either by interfering with or by enhancing the binding of one sigma factor over another (Typas et al., 2007). The Bb genome contains at least two DNA-binding proteins, Hbb (BB0232) and Gac (Knight et al., 2000), either of which could contribute to promoter recognition and/or RpoS-mediated transcription.
Almost two-thirds of the 103 genes significantly upregulated by RpoS within DMCs were absent from the in vitro-derived data set. qRT-PCR analyses of a subset of these `in vivo only' genes revealed that each belonged to the dually transcribed class of RpoS-dependent genes. We postulate that the enhanced RpoS-dependent expression of these genes observed during DMC cultivation is due, in part, to increased promoter recognition by RpoS in response to mammalian host signals. The notion that RpoS-dependent promoter recognition can vary in response to specific environmental stimuli is not without precedent. Weber et al. (2005) demonstrated that the composition of the RpoS regulon in E. coli was modular and dependent on the environmental stress to which the bacteria were exposed. What is the potential biological significance of dual promoter recognition? One possibility is that RpoS is required to augment transcription of some genes within selective niches, while the σ70 ensures a sufficient level of basal expression during points within the enzootic cycle in which RpoS is not expressed (see below). The continued expression of these dually transcribed genes in the absence of RpoS also provides a reasonable explanation for the ability of the rpoS mutant to grow normally within DMCs (Caimano et al., 2004).
The identification of a large number of genes that are subject to RpoS-mediated repression within DMCs represents a significant advance in our understanding of how Bb modulates its transcriptome between the arthropod and mammalian hosts. Almost one-third of the genes downregulated by RpoS within DMCs are known or predicted to encode lipoproteins. The products of two of these, OspA and OspB, interact with ligands present on the surface of cells within the tick midgut and are required to colonize ticks following acquisition (Pal et al., 2000; 2001; 2004; Fikrig et al., 2004; Yang et al., 2004). Loss of bb0365, encoding the LA7 lipoprotein, resulted in markedly lower survival rates in ticks following the nymphal blood meal and post-molt but had no affect on virulence in mice or larval acquisition (U. Pal, pers. comm.). Bb lacking lp28-4, the plasmid location for three of the five downregulated gbb pep_35 paralogous lipoproteins, exhibited a decreased ability to infect tick midguts (Strother et al., 2005) and were unable to be tick-transmitted to naïve mice (Elias et al., 2002). While little is known about the unique physiological requirements of spirochetes within ticks, the increased expression of bb0240-0243 suggests that glycerol may be available as an alternative carbon source between the larval and nymphal blood meals. Another notable metabolic gene within this portion of the RpoS regulon is bbk17, encoding an adenine deaminase believed to be involved in purine salvage; enhanced expression of this gene within flat versus fed ticks, as would be predicted by qRT-PCR analyses of Bb-infected ticks (see below), may enable spirochetes to adjust to distinct differences in the purine precursors present within the flat tick midgut (Sonenshine et al., 2003) compared with host blood (McCann and Katholi, 1990). Unlike the other downregulated genes examined here by qRT-PCR, bbk17 was expressed at low levels within DMC-cultivated wild-type Bb, a finding consistent with recent studies by Jewett et al. (2007) demonstrating a role for this gene during infection.
Previously, we proposed that a polyT tract upstream of the ospA and lp6.6 promoters comprises part of the binding site for an as-yet-unidentified in vivo specific repressor (Margolis and Samuels, 1995; Caimano et al., 2005). The finding of a similar motif within the upstream regions of eight additional downregulated genes provides further support for this notion. The presence of an 18 bp inverted repeat motif upstream of bbi36 and bbi38, however, suggests that RpoS-mediated repression in Bb may be controlled by at least two distinct mechanisms. The absence of a conserved sequence motif within the upstream regions for the remaining downregulated genes, on the other hand, raises the possibility that the downregulation of these genes is mediated by sigma factor competition. This regulatory model, whereby alternative sigma factors compete with each other and σ70 for a limiting amount of RNA polymerase, has been proposed as a mechanism for the RpoS-mediated repression of numerous genes in E. coli (Farewell et al., 1998) and Pseudomonas (Schuster et al., 2004).
Microarray technology, in concert with the DMC model, has enabled us to define the genomic boundaries of the RpoS regulon during growth in vivo. In order to place these data within the context of the Bb enzootic cycle, it is also necessary to establish the temporal boundaries for RpoS function and RpoS-dependent gene regulation. Based on the signal transduction pathways that initiate expression of RpoS in vitro (Hubner et al., 2001; Yang et al., 2003) and studies demonstrating the regulation of RpoS-dependent genes/proteins within Bb-infected ticks (Schwan et al., 1995; Schwan and Piesman, 2000; Ohnishi et al., 2001; Hodzic et al., 2002; Tokarz et al., 2004), it has been widely presumed that the upregulation of RpoS-dependent genes and RpoS-mediated repression (i.e. the RpoS-ON state) commences during the nymphal blood meal. Here, we have provided the first definitive evidence that this is indeed the case. Because RpoS-deficient Bb are avirulent (Caimano et al., 2004), it is not possible to confirm the RpoS dependence of individual genes expressed during infection. Nevertheless, one can surmise that the strong expression of rpoS and RpoS-dependent genes observed in DMC-cultivated organisms represents a reasonable facsimile of the RpoS regulon as it exists in spirochetes at the outset of infection. Consistent with this conjecture, Liang et al. (2002) detected transcripts for 18 RpoS-dependent genes within the skin of acutely infected mice. Less clear, however, is whether the RpoS-ON state persists throughout the duration of infection. Three lines of evidence argue for this being the case. First, at least six RpoS-dependent genes (dbpA, dbpB, bb0384, bba64, bba65 and bbi42) detected in acute mouse tissues continued to be expressed up to 4 months post inoculation (Liang et al., 2002). Second, with rare exceptions (Akin et al., 1999; Liang et al., 2004), repression of OspA and OspB is well maintained throughout infection (Montgomery et al., 1996; Coleman et al., 1997; Hodzic et al., 2003; Crother et al., 2004; Lederer et al., 2005). Third, we showed here that spirochetes exhibit the classic mammalian host-adapted phenotype (i.e. OspC+/OspA−) for months within the immunoprivileged niche created by DMCs (Elias et al., 2002; Purser et al., 2003) (Fig. 1B). How, then, would Bb maintain the ON state of the regulon while contending with the development of humoral immune responses against surface-exposed RpoS-dependent gene products that are no longer required? As recently shown for OspC and BBA64, one means of resolving this dilemma would be to selectively downregulate individual RpoS-dependent genes (Gilmore et al., 2007; Xu et al., 2007). Regardless of regulon status during chronic infection, the markedly reduced levels of rpoS (Fig. 7), in concert with the re-initiation of OspA expression (de Silva et al., 1997; Schwan and Piesman, 2000; Gilmore et al., 2001), demonstrate that, by the time of larval acquisition, spirochetes have reverted to the RpoS-OFF state that persists until the next blood meal. Paradoxically, the mammalian host-derived stimuli that are such powerful inducers of rpoS during the nymphal blood meal are superseded or overridden within feeding naïve larvae. The signal transduction apparatus and pathways that enable the spirochete to distinguish between these two ostensibly identical stimuli have yet to be discovered, although recent findings by Scheckelhoff et al. (2007) suggesting a role for host-derived catecholamines in promoting induction of OspA during the mammalian host-to-tick transition have provided an important clue.
Virulent wild-type Bb isolates c162 [a clonal derivative of strain 297 originally isolated from the cerebrospinal fluid of a Lyme disease patient (Steere et al., 1984)] and c174 [(c162 rpoSerm) (Caimano et al., 2004; Eggers et al., 2004)] were cultivated in BSK-H medium (Sigma-Aldrich Chemical, St Louis, MO) supplemented with 6% rabbit serum (Pel-Freeze Biologicals, Rogers, AK); c174 was maintained under selection using erythromycin (0.06 μg ml−1). The plasmid content of all isolates was monitored as described previously (Eggers et al., 2002) and passaged no more than twice prior to use in experiments. Infectivity of c162 and avirulence of c174 was confirmed in mice (Caimano et al., 2004). Spirochetes were enumerated by darkfield microscopy using a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA). For standard temperature-shift experiments, organisms were cultivated at 23°C to mid-logarithmic phase (~1 × 107 to 3 × 107 spirochetes per ml) then transferred to BSK-H medium at a starting density of 3000 spirochetes per ml and grown at 37°C until late-logarithmic phase (~7 × 107 to 1 × 108 spirochetes per ml) prior to being harvested; 23°C cultures were then allowed to continue until late-logarithmic phase. The pH of culture medium was measured at the time of harvest to ensure that the growth medium remained in the neutral range (pH ≥ 7.0). For reduced acid pH studies, cultures were temperature-shifted from 23°C to 37°C into BSK-H medium adjusted to pH 6.8 prior to inoculation. To obtain organisms in a host-adapted state, spirochetes were cultivated in DMCs (Spectra-Por, 8000 molecular weight cut-off) in BSK-H medium at a starting dilution of 3000 spirochetes per ml and implanted into the peritoneal cavities of either rats or rabbits as previously described (Akins et al., 1998; Sellati et al., 1999). For RpoS-induction experiments, Bb RpoSi, a strain 297 isolate in which expression of rpoS is controlled by an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible hybrid flac promoter (Gilbert et al., 2007), was temperature-shifted, in triplicate, in BSK-H medium in the absence or presence of IPTG (5 mM) and kanamycin (400 μ ml−1;to select for transformants). For serial passage experiments, spirochetes grown at 23°C were either temperature-shifted to 37°C or implanted within DMCs as described above, then at late-logarithmic phase, spirochetes were immediately used to inoculate fresh BSK-H medium at a starting density of 3000 spirochetes per ml and allowed to reach late-logarithmic phase before being passaged again under the same growth condition for a total of five passages. All E. coli strains were maintained in Luria-Bertani broth (LB) (1% tryptone, 0.5% yeast extract, 1% NaCl) with the appropriate antibiotic. Solid-phase selection was performed on LB agar plates (LB with 1.5% agar) supplemented with the appropriate antibiotic. Routine molecular cloning and plasmid propagation were performed with E. coli strain Top10 (Invitrogen, Carlsbad, CA).
Whole-cell lysates were prepared from spirochetes cultivated in BSK-H at 23°C, following temperature-shift to 37°C, or within DMCs. Cultures were harvested by centrifugation at 8500 g for 20 min, and the resulting pellets washed twice with phosphate-buffered saline (PBS). Equivalent amounts of cells (~1 × 107 total spirochetes) were re-suspended and boiled in reducing Laemmli sample buffer (Bio-Rad, Hercules, CA) and separated through 12.5% separating polyacrylamide mini-gels. Separated proteins were visualized by silver stain according to the method described by Morrissey (1981). For immunoblotting, proteins were transferred to nylon-supported nitrocellulose membrane (Micro Separations, Westborough, MA) and incubated with rat polyclonal antiserum directed against Lp6.6 (Lahdenne et al., 1997), DbpA (Hagman et al., 1998) or OspE (Akins et al., 1998) followed by horseradish peroxidase-conjugated goat anti-rat (Southern Biotechnology Associates, Birmingham, AL) secondary antibody. Rat polyclonal antiserum directed against FlaB (Caimano et al., 2005) was used to assess loading and electrotransfer uniformity. Membranes were developed using the SuperSignal West Pico chemiluminescence substrate (Pierce, Rockford, IL).
Total genomic DNA was isolated from 50 ml of Bb cultures (1 × 108 spirochetes per ml) using Isoquick nucleic acid extraction Kit (ORCA Research, Bothell, WA) according to the manufacturer's instructions and quantified spectrophotometrically. DNA was digested with HindIII for 2 h at 37°C to generate fragments < 2 kb. Two micrograms of digested 297 DNA (test DNA) was labelled with cy3-dUTP and 2 μg of B31-MI (reference) DNA was labelled with cy5-dUTP. Equal amounts of cy3- and cy5-labelled DNAs were hybridized as described previously (Terekhova et al., 2006). Raw spot intensities were generated using GenePix Pro v6 software (Axon instruments). Background subtraction, data normalization and mean log2 (cy3/cy5) ratio calculations were all performed using GeneTraffic software Version 3.11 (Iobion, La Jolla, CA).
Wild-type (c162) and rpoS mutant (c174) Bb were cultivated in BSK-H following temperature-shift as described above. Culture densities were assessed daily to ensure that each sample was taken during late-logarithmic phase (8 × 107 to 1 × 108 spirochetes per ml). Following harvest, the pH of spent growth medium was measured to ensure that each culture remained within a neutral range (pH 7.2–7.5). SDS-PAGE (Fig. S1) and immunoblot using sera directed against Lp6.6 and DbpA (data not shown) were used to confirm that each c162 and c174 experimental sample exhibited the expected polypeptide profile prior to being used in microarray experiments. cDNAs generated from c162- and c174-derived RNAs were analysed by qRT-PCR to ensure that transcripts for rpoS and ospC were present only in the wild-type isolate and induced by temperature-shift and growth within DMCs.
Total RNA was isolated from organisms cultivated either in vitro following temperature-shift or within DMCs using TRizol reagent (Invitrogen) according to manufacturer's instructions. Contaminating genomic DNA was removed using Turbo DNAfree (Ambion, Austin, TX). The absence of contaminating DNA was confirmed by testing each treated RNA in amplification reactions containing primers specific for flaB as previously described (Caimano et al., 2004). For microarray experiments, 12 μg each of DNase-treated RNA from three independently implanted DMCs and five independent temperature-shifted (37°C) cultures of wild type (c162) and rpoS mutant (c174) was converted to cDNA using the PowerScript fluorescent labelling kit (Clontech, Mountain View, CA) in reactions containing random hexamer (Promega, Madison, WI) and random 18-mer (IDT, Skokie, IL). The resulting cDNA was then split into two equal aliquots for labelling with cy3 or cy5 (Amersham, Piscataway, NJ) according to the manufacturer's instructions; unincorporated dye was removed by ethanol precipitation. Precipitated cy3- and cy5-labelled probes generated from wild-type Bb RNAs were then mixed with their corresponding cy5- and cy-3-labelled rpoS mutant counterparts (i.e. dye-swap) and the combined probes were further purified using QiaQuick PCR purification columns (Qiagen, Valencia, CA), dried in a speed vac for 45–60 min and re-suspended in 25 μl of nuclease-free water. Re-suspended cy3/cy5 probes were mixed with 18 μl of DNase, RNase-free formamide (Acros Organics, Belgium) and 17 μl of 4× Hybridization buffer (Amersham Biosciences, Piscataway, NJ) in a final volume of 60 μl. The mixture was denatured by heating at 95°C for 5 min, cooled on ice for 1 min and applied to previously described Bb glass slide microarrays (Terekhova et al., 2006). Prior to use, slides were pre-hybridized in a solution of 30% formamide, 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.1% sodium dodecyl sulphate (SDS) and 0.1% bovine serum albumin (Sigma) for 45–60 min at 42°C. Following prehybridization, slides were washed twice in distilled water, dipped in isopropanol and dried immediately by centrifugation at 1400 g for 3 min. After application of the labelled probes, slides were covered with lifter slips (Fisher Scientific) and placed in humidified hybridization chamber. The chambers were placed in water bath and incubated overnight at 42°C. After overnight hybridization slides were washed twice with 2× SSC/0.1% SDS for 10 min at 42°C, twice with 0.1× SSC/0.1% SDS for 10 min at room temperature and four times with 0.1× SSC for 1 min at room temperature. Washed slides were immediately dried by centrifugation at 2700 r.p.m. for 3 min.
The hybridized microarray slides were scanned on a GenePix 4000B microarray scanner (Axon Instruments, Union City, CA) at 10 μm resolution. The resulting dual-colour images were recorded in 16-bit multi-image TIFF files which were analysed by GenePix Pro software (Axon Instruments). GenePix data were then input to Microsoft Excel for background correction, normalization and filtering. Statistical analyses were performed in Microsoft Excel using data derived from the GAL files. Briefly, pixel intensity values were corrected for background by subtracting the mean value for `blank' spots (for individual slides) from the spot intensities for each ORF. The spot intensities were then normalized by first summing the intensity values for all ORFs for each slide, setting wild type as `reference', and then dividing the sum of intensities for each slide by the sum of intensities for the reference slide; the resulting normalization factor was used to multiply all background-corrected spot intensities. The normalized values were employed for two statistical tests. First, the mean intensity values were calculated for each ORF in both wild type and mutant; the ratios (wild type/mutant) of the mean intensities were determined and log2-transformed. The mean and standard deviation (SD) of the log2 values for all ORFs was calculated. Those individual ORFs with a log2 ratio exceeding ± 2 SD of the mean log2 ratio for all ORFs were considered to have significant differences in expression between wild type and mutant. For DMC-derived microarrays, only those genes that exhibited wild type/mutant ratios ≥ 4.50 or ≤ −2.41 were considered to be expressed at significantly different levels in the wild type. For in vitro based microarrays, only those genes that exhibited a wild type/mutant ratio ≥ 2.53 or ≤ −1.60 were considered to be expressed at significantly different levels in the wild type. In addition, a two-tailed t-test was performed for each ORF to determine if pixel intensity values for the two conditions were different. ORFs with P < 0.01 were considered to represent significantly different expression levels. Except where indicated, all ORFs included in the DMC and in vitro microarray data sets passed both statistical tests.
All nucleotide sequencing was performed by the UCHC Molecular Core Facility using an Applied Biosystems model 373A-automated DNA sequencer and PRISM™ ready reaction DyeDeoxy™ Terminator cycle sequencing kits according to manufacturer's instructions (Applied Biosystems, Foster City, CA). Unless otherwise indicated, primers used in qRTPCR studies were designed using MacVector version 8.1 software (Accelrys Bioinformatics, San Diego, CA). Routine and comparative sequence analyses were performed using MacVector (version 9.5.1, MacVector, Cary, NC). MEME (http://meme.sdsc.edu) was used to search for conserved sequence motifs within the 400 bp regions upstream of the translation start sites for genes upregulated by RpoS. Lipo-protein predictions were based on SPLIP (Setubal et al., 2006), LipoP (Juncker et al., 2003), TMHMM (http://www.cbs. dtu.dk/services/TMHMM-2.0/) and PSORT (Gardy et al., 2005).
Wild-type and mutant DNase-treated RNAs (4 μg) from in vitro and DMC-cultivated spirochetes used in the microarray studies described above were converted to cDNA using SuperScript First-Strand Synthesis for RT-PCR (Invitrogen) in the presence and absence of reverse transcriptase (RT) according to the manufacturer's instructions. Individual cDNA were assessed first using primers specific for flaB and ospC (Table S3) and then pooled and assayed by real-time PCR (in quadruplicate) using iQ SYBR Green Supermix (Bio-Rad). Reaction conditions for each gene-specific primer pair (Table S3) were optimized for temperature and Mg2+ concentration. Control reactions were performed using pooled `No RT' cDNA samples. For quantification, amplicons corresponding to each gene of interest were cloned into the pCR2.1-TOPO cloning vector (Invitrogen), then purified recombinant plasmid DNAs for each amplicon were diluted (107–102 copies μl−1) to generate a standard curve. Transcript copy numbers for each gene of interest were calculated using the iCycler post-run analysis software based on internal standard curves then normalized against copies of flaB present in the same cDNA. To determine the statistical significance of observed differences, normalized copy number values were compared within Prism v5.00 (GraphPad Software, San Diego, CA) using an unpaired t-test with two-tailed P-values and a 95% confidence interval.
Pathogen-free I. scapularis larvae were obtained from a colony maintained at Oklahoma State University. To obtain Bb-infected nymphs, 3- to 5-week-old C3H/HeJ mice were syringe-inoculated intradermally with a dose of 104 spirochetes; infection was confirmed by serology and/or ear punch biopsy at 2 weeks. At 3–5 weeks post inoculation, naïve larvae (200–300 per mouse) were placed on anaesthetized Bb-infected mice, allowed to feed to repletion and then collected over water as previously described (Hagman et al., 2000). Engorged larvae were then allowed to molt into flat nymphs. To obtain fed nymphs, flat Bb-infected nymphs (10–12 per mouse) were placed in polypropylene capsules secured to the shaved backs of naïve C3H/HeJ mice as previously described (Hagman et al., 2000), allowed to feed until near or fully engorged (~48–60 h post attachment) and then removed using forceps. Individual fed nymphs were immediately placed on dry ice and then stored in TRizol (Invitrogen) until all ticks were collected. To obtain fed Bb-infected larvae, naïve pathogen-free larvae were placed on infected C3H/HeJ mice, allowed to feed to repletion, returned to a humidified chamber and then harvested at 6–8 days post repletion. To isolate RNA from infected ticks, ~200 fed larvae and flat nymphs, or 75 fed nymphs were crushed in 1 ml of TRizol using a glass homogenizer then centrifuged at 4000 g for 1 min to remove tick cuticle material. DNA-free RNA and cDNA were generated as described above. Due to an extraneous (non-specific) flaB PCR amplicon observed in the SYBR green melting curve assays with flaB primers, obtained only with tick-derived Bb cDNAs, the previously described flaB primer/FAM-490-probe-based assay (Fisher et al., 2005) was used to normalize rpoS expression in Bb-infected ticks.
This work was supported by the NIH Grants AI29735 (J.D.R. and M.J.C.) and AI45801 (I.S.). E.A.M and M.A.G. were supported by NIH Grant AI051486 to D. Scott Samuels. The authors would like to thank Darya Terekhova (New York Medical College) for sharing comparative genomic data and Scott Samuels (The University of Montana) for helpful discussions and critical reading of the manuscript.