Physiological and molecular characterization of genetic competence in Streptococcus sanguinis

SUMMARY

Streptococcus sanguinis is a major component of the oral flora and an important cause of infective endocarditis. Although S. sanguinis is naturally competent, genome sequencing has suggested significant differences in the S. sanguinis competence system relative to those of other streptococci. An S. sanguinis mutant possessing an in-frame deletion in the comC gene, which encodes competence-stimulating peptide (CSP), was created. Addition of synthetic CSP induced competence in this strain. Gene expression in this strain was monitored by microarray analysis at multiple time points from 2.5 to 30 min after CSP addition, and verified by quantitative RT-PCR. Over 200 genes were identified whose expression was altered at least two-fold in at least one time point, with the majority upregulated. The “late” response was typical of that seen in previous studies. However, comparison of the “early” response in S. sanguinis with that of other oral streptococci revealed unexpected differences with regard to the number of genes induced, the nature of these genes, and their putative upstream regulatory sequences. S. sanguinis possesses a comparatively limited early response, which may define a minimal streptococcal competence regulatory circuit.

INTRODUCTION

Streptococcus sanguinis is one of the earliest colonizers of teeth and is an abundant component of dental plaque (Carlsson, 1965; Aas et al., 2005). S. sanguinis may be beneficial in the oral environment by interfering with colonization of teeth by Streptococcus mutans and other pathogens (Hillman et al., 1985; Stingu et al., 2008). When introduced into the bloodstream, however, S. sanguinis and other oral streptococci can prove harmful as a cause of infective endocarditis (Di Filippo et al., 2006).

S. sanguinis possesses the ability to take up and incorporate free extracellular DNA from the environment (Gaustad and Håvarstein, 1997), a trait known as genetic competence. This property is shared by certain other bacteria, including several oral streptococcal species (Martin et al., 2006). Genetic competence is best understood in Streptococcus pneumoniae, in which it was first discovered (Johnsborg and Håvarstein, 2009). Seven “early” com genes are essential for induction and regulation of the S. pneumoniae competence system (Johnsborg and Håvarstein, 2009). The comC gene encodes a 41-aa precursor peptide that is cleaved and exported by the products of the comAB operon, ComA and ComB (Hui et al., 1995). The mature 17-aa ComC peptide, termed “competence-stimulating peptide” (CSP), is sensed by the ComD sensor kinase, which presumably responds by phosphorylating the ComE response regulator. The latter two proteins are encoded along with ComC in the comCDE operon (Pestova et al., 1996). ComE activates expression of the comAB and comCDE operons (Ween et al., 1999; Johnsborg and Håvarstein, 2009). These genes thus form a “quorum sensing” system, which results in amplification of the component gene products when the concentration of cells (and, thus, extracellular CSP) reaches a critical density. ComE also activates the comX gene, which is present in two identical copies. ComX serves as an alternative sigma factor, directing the expression of “late” genes required for uptake and incorporation of DNA (Lee and Morrison, 1999). The comW gene product has been shown to stabilize ComX and activate it by an additional, undefined mechanism (Sung and Morrison, 2005). Studies in Streptococcus gordonii have shown that the genetic and physiological requirements for competence in this species are similar to those in S. pneumoniae (Håvarstein et al., 1996; Lunsford, 1998).

Microarray analyses in S. pneumoniae (Peterson et al., 2000; Dagkessamanskaia et al., 2004; Peterson et al., 2004) have extended earlier findings of transient, sequential gene expression upon competence induction (Alloing et al., 1998; Rimini et al., 2000). A similar analysis has been performed in S. gordonii (Vickerman et al., 2007). In both species, addition of synthetic CSP to the culture medium produced two temporal waves of transcriptional induction. Genes categorized as “early” were upregulated within 5 minutes of exposure to CSP, and included comAB, comCDE, and two identical copies of genes encoding the competence-specific sigma factor (ComX in S. pneumoniae and ComR in S. gordonii) (Peterson et al., 2004; Vickerman et al., 2007). Genes categorized as “late” showed maximal expression 10 to 15 minutes post-CSP exposure and most were dependent on the competence sigma factor for induction (Alloing et al., 1998; Peterson et al., 2004; Vickerman et al., 2007). This category includes genes encoding proteins required for uptake and recombination of DNA. Transcription of both early and late genes returned to near basal levels in both species by 30 to 40 min. The kinetics of genetic competence are consistent with this expression pattern, with transformation frequencies peaking 10 to 20 min after CSP addition and decreasing dramatically by 40 min in both species (Håvarstein et al., 1995b; Peterson et al., 2004; Vickerman et al., 2007).

S. mutans has also been the subject of a recent microarray analysis examining CSP-dependent gene expression (Perry et al., 2009). Genetic competence in this species has been shown to differ from that in S. pneumoniae and S. gordonii in several respects (Martin et al., 2006), including delayed induction of competence gene expression (Ahn et al., 2006) and transformation (Kreth et al., 2005) in response to CSP. The results were notable for the identification of a large number of upregulated bacteriocin genes (Perry et al., 2009), in keeping with the earlier suggestion that the comC and comDE genes of S. mutans function primarily in induction of bacteriocin production rather than genetic competence (van der Ploeg, 2005; Martin et al., 2006).

Because S. sanguinis is closely related to the fellow mitis group members S. pneumoniae and S. gordonii, it was expected to possess a similar competence system (Martin et al., 2006). Yet, when its genome was sequenced (Xu et al., 2007), it was discovered that two of the early genes indispensable for competence induction in S. pneumoniae (Peterson et al., 2004) and S. gordonii (Lunsford and London, 1996), comA and comB, were seemingly absent from SK36. The genome sequence also confirmed that ComC of SK36, like that of S. sanguinis NCTC 7863 (Håvarstein et al., 1997), lacks the Gly-Gly motif common to all other streptococcal ComC sequences examined to date, which serves as the recognition sequence for ComA-mediated cleavage and concomitant CSP secretion (Håvarstein et al., 1995a; Ishii et al., 2006). Fundamental components of the competence regulatory system in S. sanguinis therefore appear to differ from those in other mitis group species, as well as other transformable streptococci such as S. mutans, suggesting the need to examine competence more closely in S. sanguinis. We report here the initial characterization of genetic competence in S. sanguinis at the physiological and molecular levels. We have performed a comparison of the transcriptional response of S. sanguinis to CSP to that of S. pneumoniae, S. gordonii, and S. mutans, allowing for the identification of common and unique features. S. sanguinis exhibited a remarkably limited early gene response, characterized by fewer competence regulatory genes and non-competence genes than found in other streptococci.

METHODS

Bacterial strains, plasmids, and growth conditions

A list of strains and plasmids used in this study is shown in Table 1. S. sanguinis strain SK36 was selected for this study based on the availability of its genome sequence (Xu et al., 2007) and its demonstrated genetic competence (Paik et al., 2005). SK36 and derivatives were routinely grown in Brain Heart Infusion (BHI) broth (Bacto, Sparks, MD.) at 37°C under reduced oxygen (7% H₂, 7% CO₂, 80% N₂, 6% O₂). For transformation, strains were grown aerobically overnight at 37°C in Todd Hewitt (TH) broth (Bacto) supplemented with 2.5% (v/v) horse serum (Invitrogen, Grand Island, NY) (TH-HS). For examination of pH effects on growth, SK36 was grown overnight in TH-HS (pH 7.6) at 37°C then diluted 1:100 in TH-HS or TH-HS acidified with HCl in a 96-well plate and incubated at 37°C in a FLUOstar plate reader (BMG Labtechnologies, Durham, NC) where the OD₄₅₀ was recorded every 10 minutes immediately following agitation. Wells containing media alone were used for background subtraction.

TABLE 1.

Bacterial strains, plasmids, and primers used in this study.

Strain, plasmid, or primer	Description	Reference
Strains
S. sanguinis
SK36	Human dental plaque isolate	(Xu et al., 2007)
JFP41	ΔcomC; derived from SK36	This study
Plasmid
pJFP16	Cmr and Knr; containing nrdD1::cat	(Turner et al., 2009)
comC Mutagenesis Primers
ComC-A	CGAACAAGAATACCTGATT	This study
ComC-B	TCCAATTATTTCAAGTGACATTAACTATCTCCTATCTTTTTA	This study
ComC-C	TCACTTGAAATAATTGGATTA	This study
ComC-D	TGGCACTCTCAATAGCA	This study
RT-PCR Primers
0195-FWD	CCTGCAGCACCGAAATTTAT	This study
0195-RVS	CTGAGAATATGGGCCAGCAT	This study
comD-FWD	CGAGAAAGAAGAGCGTTTGC	This study
comD-RVS	AATTTGCATCAGCCAAGACC	This study
comE-FWD	TCAAGTCAGGATGGAAACGAC	This study
comE-RVS	TTCCTCCCCCTTAATGTCAA	This study
comX-FWD_3	CGCCAGCCTTATGAAGAAGT	This study
comX-RVS_3	TCGCTTCTCTGAAGGCAACT	This study
comEA-FWD	CTGAGAGCAGGAGCGAGAGT	This study
comEA-RVS	GCCTCTGTCGCCGTATTTAG	This study
comYD-FWD	TACCAGGAGACGCAGAGGTT	This study
comYD-RVS	ACTGTCCTGTCCTCCGTCTG	This study
gapA-FWD	GCTTTCCGTCGTATCCAAAA	This study
gapA-RVS	ACACCGTCAGTAGCCCAGTC	This study
tuf-FWD	GCTTCAACTGACGGACCAAT	This study
tuf-RVS	CGTCACCTGGGAAGTCGTAT	This study

Transformation

Transformation was performed as previously described (Paik et al., 2005) for mutant generation, or using a modified protocol for competence assessment. Unless otherwise indicated, overnight cultures of each strain grown as described above were diluted 1:200 into pre-warmed TH-HS, and incubated at 37°C until the optical density at 660 nm (OD₆₆₀) reached ~ 0.07. Aliquots of 330 μl were then transferred into pre-warmed 0.7-ml microfuge tubes containing 70 ng of S. sanguinis CSP (Håvarstein et al., 1997) and 10 ng of pJFP16, a suicide plasmid containing the chloramphenicol (Cm) acetyl transferase gene flanked on either side by ~1 kb of the SK36 nrdD locus (Table 1). Preliminary studies revealed that 10 ng of this plasmid was saturating for transformation. Cells were incubated at 37°C for 20 min before addition of DNase I (10.5 ng/μl) and 10 min of further incubation to eliminate extracellular DNA. Cells were then serially diluted in PBS, transferred to 12.5-ml of Tryptic Soy Broth (Bacto) supplemented with 1% (w/v) sterile molten low-melting agarose (Promega, Madison, WI), and dispensed into Petri dishes for solidification. Plates were then incubated for 2 h at 37°C to allow expression of antibiotic resistance. A top layer of 12.5 ml of TSA ± Cm (10 μg/ml) was then added, plates were again allowed to solidify, and then transferred to 37°C. Transformation frequency was determined after 48 h of incubation and calculated as CFU/mlfrom plates with Cm divided by CFU/ml from plates with no antibiotic.

Construction of a comC deletion mutant

An overlap extension PCR technique (Ho et al., 1989) was used to fuse the start codon of the comC gene to the second codon of comD, thus deleting the comC gene and intervening sequence (Dagkessamanskaia et al., 2004). The upstream fragment was generated with primers ComC-A and ComC-B, and the downstream fragment with primers ComC-C and ComC-D. The two fragments were fused in a final reaction containing ComC-A and ComC-D. All primers used for this study are listed in Table 1. The purified 2.9-kb fusion PCR product was introduced into SK36 by transformation, as described above. Mutants were identified via PCR screening using primers flanking the deletion. One mutant, designated JFP41, was selected for further study. DNA sequence analysis confirmed that it possessed the desired deletion and had no other mutations in the flanking sequences.

Microarray analysis

JFP41 was cultured in TH-HS, as for competence assessment. An aliquot was harvested for RNA isolation, and the remainder of the culture was treated with S. sanguinis CSP (Gaustad and Håvarstein, 1997). Additional samples were harvested at timed intervals. Detailed methods for RNA isolation, cDNA synthesis and labeling, microarray hybridization, and data analysis are provided in the Microarray Methods file in the supplemental material. Microarray data have been submitted to the NCBI Gene Expression Omnibus (GEO) repository under accession number GSE21619.

Quantitative RT-PCR

For analysis of SSA_0195 expression, 12.5-ml aliquots of JFP41 cultures were incubated for 5 min with 25 ml RNAprotect (Qiagen, Valencia, CA), then centrifuged for 20 min at 8,000 × g. Cells were lysed according to the manufacturer’s protocol for gram-positive bacteria, except that a Fastprep beadbeater with lysing matrix B (MP Biomedicals, Solon, OH) was used in place of disruption with acid-washed beads. RNA was isolated via the RNeasy Mini Kit (Qiagen), with the optional DNase I treatment. Purified RNA was subjected to a second DNase treatment (Ambion, Austin, TX), quantitated, and analyzed for integrity as described in the Microarray Methods file in the supplementary material. First-strand cDNA synthesis was performed with SuperScript III Reverse transcriptase (200 units/μl; Invitrogen). For analysis of com gene orthologs, aliquots of RNA samples previously isolated for microarray analysis were subjected to an additional DNase treatment (Ambion) and purified as per the manufacturer’s protocol. cDNA templates were created via Ready-To-Go^™ You-Prime First-Strand Beads (Amersham Biosciences, Pittsburgh, PA) as described in the manufacturer’s protocol. For both analyses, qRT-PCR reactions were carried out in a 7500 Fast RT-PCR system (Applied Biosystems, Carlsbad, CA) using RT² Real-Ti^™ SYBR Green/Rox PCR master mix (Qiagen) and primers listed in Table 1. Control reactions performed in the absence of reverse transcriptase confirmed the absence of contaminating genomic DNA in all samples. Normalization was performed with the tuf or gapA genes. A standard curve was generated for each gene from DNA templates of known concentration.

Bioinformatic and phylogenetic analysis of putative upstream regulatory sites

For identification of potential ComE binding sites, conserved sequences (Ween et al., 1999) upstream from the comC gene in S. pneumoniae, Streptococcus mitis, Streptococcus oralis, Streptococcus cristatus, S. gordonii, and S. sanguinis were entered into the program MEME for motif identification (http://meme.nbcr.net/) (Bailey and Elkan, 1994). The 31-bp position-specific scoring matrix (PSSM) produced by the program from the input sequences was used to search a database of nucleotide sequences extending 1000 bp upstream and 200 bp downstream from the start site of all annotated genes in S. sanguinis SK36 using the program MAST (Bailey and Gribskov, 1998). Default settings were employed except that reverse complement strands were treated as separate sequences. The same PSSM was used for a MAST search of S. pneumoniae (strain TIGR4), S. gordonii (strain CH1), and S. mutans (strain UA159) databases. The process was repeated after adding the S. mutans SMU.1914c-proximal sequence (van der Ploeg, 2005) to the six comC sequences for PSSM construction. Sites from each species identified in either search were used for further rounds of PSSM creation and database searching until no additional significant matches from that species emerged. Potential combox sites (Campbell et al., 1998) upstream from S. sanguinis late genes were initially sought by manual inspection. A 200-bp region upstream from each gene identified in this way was then used for PSSM creation and iterative searching, as described for ComE. The search was discontinued when no additional significant matches were obtained. Phylogenetic analysis of putative ComE-binding sites was performed with the MEGA4 program (Tamura et al., 2007). Phylogenetic trees were created using the Neighbor-Joining method with evolutionary distances computed using the Jukes-Cantor method.

RESULTS

Competence induction in S. sanguinis

Given the unique features of S. sanguinis competence regulatory genes determined by DNA sequencing, we were interested in further characterizing genetic competence in this species. A notable feature of S. pneumoniae is a rapid cell-density dependent increase and decay of competence (Håvarstein et al., 1995b). Competence was assessed in S. sanguinis SK36 grown to different densities as described in METHODS by addition of a saturating amount of pJFP16, a plasmid that confers Cm resistance by integration into the chromosomal nrdD locus (Turner et al., 2009). As shown in Fig. 1A, S. sanguinis likewise demonstrated rapid induction of competence, in this case, when the cell density approached 10⁶ CFU ml⁻¹ (corresponding to an OD₆₆₀ of 0.068–0.074). As the cell density surpassed 10⁷ CFU ml⁻¹, competence decreased dramatically (Fig. 1A). Density dependence of transformation has also been demonstrated for S. mutans (Perry and Kuramitsu, 1981) and S. gordonii (Pakula and Walczak, 1963).

FIG. 1.

Characterization of competence development in SK36 and a comC mutant, JFP41. (A) The effect of culture density on transformation in SK36. Cells were cultured and assessed for transformation frequency as described in METHODS. Error bars represent the SD derived from one experiment that was representative of six. Mean and SD values from separate experiments could not be combined because it was not possible to obtain identical cell densities in separate experiments. (B) The effect of CSP concentration on transformation frequency in JFP41. Exogenous CSP was added to cultures at the concentration indicated prior to transformation and plating. (C) Transformation of JFP41 after various durations of exposure to 220 ng/ml CSP. CSP was added at time 0, transforming DNA was added at the time indicated, and incubation was continued for 10 min before addition of DNase I and plating. Error bars in B and C represent SDs from the combined data from three separate experiments.

A primary goal of this study was to perform a transcriptomic analysis of gene regulation in response to competence induction. Previous studies performed with S. gordonii and S. pneumoniae have employed growth in acidic conditions to inhibit natural competence, allowing for the controlled induction of competence by addition of exogenous CSP (Peterson et al., 2000; Peterson et al., 2004; Vickerman et al., 2007). The acidic conditions did not affect the growth of either species (Peterson et al., 2000; Vickerman et al., 2007). To determine whether this strategy could be used with S. sanguinis, SK36 was cultured in normal (pH 7.6) and acidic (pH 6.9) TH-HS prior to transformation with pJFP16. The transformation frequency decreased from approximately 12% at pH 7.6 to below the level of detection (3.4 × 10⁻⁵ %) at pH 6.9, suggesting that natural competence was effectively inhibited. It was observed, however, that cultures grown in the pH 6.9 medium had significantly lower cell densities than those grown in neutral conditions (data not shown). When an intermediate condition of pH 7.3 was used, growth was only moderately affected; however, competence was also partially restored (data not shown).

Given these results, we sought to create a comC deletion mutant in which competence could be induced by the addition of exogenous CSP. The comC gene in S. sanguinis is presumably co-transcribed with the adjacent comD and comE genes, as in S. pneumoniae and S. gordonii. To minimize effects on expression of these downstream genes, which are required for competence induction in other streptococci, a PCR overlap extension technique (Ho et al., 1989) was used to fuse the start codon of comC to the second codon of comD, deleting comC, as well as the 29-bp comCD intergenic region. Test transformations were performed as described above. In the absence of added CSP, transformation of JFP41 could not be detected. When synthetic CSP was added, competence was restored to near wild-type levels (data not shown). The results demonstrated two critical phenotypes expected for the mutant: first, that natural competence was drastically reduced or eliminated, as would be expected for a mutant incapable of CSP production; and second, that competence was restored upon the addition of exogenous CSP, suggesting that the comD and comE genes were subject to normal CSP-dependent induction. An identical strategy was used previously for microarray analysis of CSP-dependent gene expression in S. pneumoniae (Dagkessamanskaia et al., 2004).

The effect of varying concentrations of CSP on JFP41 transformation frequency was also examined. As shown in Fig. 1B, competence increased with increasing CSP concentration until saturation, peaking at approximately 200–240 ng/ml. The concentration of 220 ng/ml was chosen for the remaining studies.

Finally, the timing of competence induction following addition of exogenous CSP to JFP41 cultures was examined. JFP41 cells at an OD₆₆₀ of approximately 0.07 were exposed to CSP, and at varying intervals, transforming DNA was added and the cultures examined for transformation frequency as above, except that incubation was continued for only 10 min prior to addition of DNase. The results are shown in Fig. 1C. JFP41 competence peaked after approximately 10 minutes of exposure to CSP and returned to near baseline levels by 40 minutes. These results are similar to those observed in S. pneumoniae (Håvarstein et al., 1995b; Rimini et al., 2000; Peterson et al., 2004) and S. gordonii (Vickerman et al., 2007), and differ from S. mutans, in which a 2-hr delay was noted between addition of CSP and development of competence (Kreth et al., 2005).

Microarray analysis of transcriptional response to CSP

Conditions were chosen for microarray analysis based on the results shown in Fig. 1. JFP41 cells were grown to an OD₆₆₀ of 0.068–0.074 (3.2–4.2 × 10⁶ CFU/ml), and a culture aliquot was removed for RNA isolation as described in the Microarray Methods included in the supplementary material. CSP was then added to 220 ng/ml, and additional culture aliquots harvested for RNA isolation 2.5 to 30 min later. The RNA isolation protocol employed hot acid phenol to quickly halt transcription, as with previous microarray analyses of the CSP response in S. pneumoniae (Peterson et al., 2000; Peterson et al., 2004) and S. gordonii (Vickerman et al., 2007).

Data were collected from microarrays and filtered based on signal quality and significance. (See the Microarray Methods in the supplementary material.) Of the 2394 genes represented on the array, 205 were identified whose expression was statistically verified and altered at least two-fold for at least one time point relative to the untreated sample from the same culture; 122 genes were upregulated, and 83 downregulated. Normalized values for the S. sanguinis genes, expressed as the log₍₂₎ of the ratio of the values for CSP-treated samples relative to untreated samples, are shown in Table S1 in the supplementary material. Fig. 2 displays these data in a colorimetric graph in which genes are clustered based on similarity of expression patterns.

FIG. 2.

Colorimetric map of S. sanguinis genes whose expression was statistically validated and regulated at least two-fold by CSP. Each column represents RNA sampled from one time point, and each row the expression of a single gene relative to the “0” time point. Red squares indicate upregulation and green squares indicate downregulation, with the intensity of the color proportional to the degree of change. Black squares indicate no significant difference, and gray squares indicate incomplete data. Genes were clustered based on similarity of expression patterns using the Cluster 3.0 program (Eisen et al., 1998) and visualized using TreeView (Eisen et al., 1998). Data were derived from three separate experiments. Enlargements indicate clusters that include genes with the highest levels of regulation. Labels in red indicate orthologs of genes that have been shown to be required for transformation in S. pneumoniae (Peterson et al., 2004).

Early-induced genes

The induced genes were categorized as “early” or “late,” with the early category composed of four genes that demonstrated greater than five-fold upregulation at 2.5 min, peak expression within 10 min, and reduced expression at later time points: the known competence regulatory genes, comD, comE, comX; and a gene encoding a conserved uncharacterized protein, SSA_1889 (Fig. 2). This group corresponds kinetically to the “early” competence gene category identified previously in S. gordonii (Vickerman et al., 2007) and a previous S. pneumoniae analysis (Peterson et al., 2004). (Additional microarray analyses in S. pneumoniae and S. mutans did not apply this classification.)(Dagkessamanskaia et al., 2004; Perry et al., 2009) Table 2 lists the early genes from S. sanguinis, S. pneumoniae, and S. gordonii, along with any orthologs present in the other species, which were identified as mutual best hits by BLASTP, which also had alignment e values <10⁻⁵. The most striking feature of this comparison is the lesser number of genes assigned to this category in S. sanguinis—five genes, in comparison to 28 for S. pneumoniae (Peterson et al., 2004) and 35 for S. gordonii (Vickerman et al., 2007). SK36 lacks recognizable genes encoding the CSP-transport proteins ComA and ComB, as well as a duplicate copy of the competence sigma factor gene (comX in S. pneumoniae or comR in S. gordonii), all of which occur in the other two species. Also, the S. sanguinis comC gene is not detectable in our study due to its deletion from the test strain. The greater contribution to this discrepancy, however, lies in the large number of early genes with no defined role in competence induction in the other two species: 20 in S. pneumoniae and 27 in S. gordonii. As shown in Table 2, these genes are remarkable in that their upregulation is almost entirely species-specific. SP_1918 in S. pneumoniae and SGO_1750 of S. gordonii encode orthologous ABC transport proteins, both of which belong to the early response. Every other gene is upregulated in only one of the three species.

TABLE 2.

Early-induced genes and orthologs in S. pneumoniae, S. sanguinis, and S. gordonii

S. pneumoniae		S. sanguinis		S. gordonii
Locusa	Gene	Locus	Gene	Locus	Gene
SP_0014,	comX1,	SSA_0016	comX	SGO_1707,	comR1,
SP_2006	comX2			SGO_2130	comR2
SP_0018	comW	---		---
SP_0019	purA	SSA_2185	purA	SGO_1989	purA
SP_0042	comA	---		SGO_2097	comA
SP_0043	comB	---		SGO_2096	comB
SP_0429		---		---
SP_0430		---		---
SP_0530	blpA	---		---
SP_0545	blpY	---		---
SP_0546	blpZ	---		---
SP_0547		---		---
SP_0635		---		---
SP_1110		SSA_0936	mreA	SGO_1050	ribF
SP_1547		SSA_1395		SGO_1307
SP_1548		---		---
SP_1549	def-2	SSA_0458		SGO_1754
SP_1716		---		---
SP_1717		---		---
SP_1918		SSA_0462		SGO_1750
SP_1942		SSA_1909		SGO_0535
SP_1943		SSA_1910		SGO_0534
SP_1944		SSA_1911		SGO_0533
SP_1945		---		---
SP_2156		SSA_1965		SGO_1730
SP_2235	comE	SSA_2378	comE	SGO_2145	comE
SP_2236	comD	SSA_2379	comD	SGO_2146	comD
SP_2237	comC2	---		---
---		SSA_1889		---
---		SSA_0195†		---
---		---		SGO_0097
SP_0158		SSA_2263	nrdI	SGO_0098
---		---		SGO_0115	sthB
---		---		SGO_0376
---		---		SGO_0377
---		SSA_0621		SGO_0647
SP_0785		SSA_1532	acrA	SGO_0750
SP_0786		SSA_1531	salX	SGO_0751
SP_0787		SSA_1530		SGO_0752
SP_1381		SSA_1026		SGO_0798
SP_1380		SSA_1027		SGO_0799
SP_1162		SSA_1175	acoC	SGO_1131	sucB
SP_1263	topA	SSA_1184*	topA*	SGO_1197	topA
SP_0868		SSA_1955		SGO_1721	sufD
SP_2000		SSA_1972		SGO_1731
SP_2001		SSA_1973		SGO_1732
SP_2002		SSA_1974		SGO_1733
SP_0338		SSA_2096	clpL	SGO_1856
SP_1722		SSA_0456	scrA	SGO_1857
---		SSA_0860		SGO_2013
---		---		SGO_2086
---		---		SGO_2088
SP_0054	purK	SSA_0040	purK	SGO_2089	purK
SP_0051	purD	SSA_0037	purD	SGO_2092	purD
---		---		SGO_2093
---		SSA_0036*		SGO_2094
---		---		SGO_2095	comX
---		---		SGO_2147	comC

^aGenes listed on the same line are orthologs as determined by bidirectional BLASTP comparison in all combinations. Bold text indicates classification as an early gene in this study, that of Peterson et al (Peterson et al., 2004), or Vickerman et al (Vickerman et al., 2007).

^†gene identified as early by qRT-PCR

^*genes that were classified as belonging to the late competence response

We wondered whether this difference might be explained by stricter criteria used in our study for categorization of genes as early versus late. Table 2 indicates that this is not the case. Only two genes, SSA_1184, encoding DNA topoisomerase I, and SSA_0036, encoding a putative cell wall amidase, were identified as late in S. sanguinis while their orthologs in one of the other two species (SGO_1197 and SGO_2094, respectively) were identified as early. Moreover, the two S. gordonii genes are preceded by combox sequences characteristic of late gene expression, as described below, and were expressed more highly at 15 and 40 min than at 5 min—another characteristic of late genes (Vickerman et al., 2007). We also noted that a previous microarray analysis performed in S. pneumoniae employed a mixture of cDNA derived from all time points in place of a “0” time point reference to increase the reliability of identifying genes with low basal expression levels (Peterson et al., 2004). We reexamined our data and noted a single gene that would have been classified as early had it not been filtered out due in part to low expression at time 0 (see Microarray Methods in the supplemental material). This gene, SSA_0195, encoding a hypothetical protein, exhibited kinetics consistent with early gene expression, appearing upregulated more than 15-fold at 2.5 min and more than four-fold at all later time points, suggesting that it might be an early gene (data not shown). Expression of this gene was therefore examined by quantitative RT-PCR, using gapA for normalization. SSA_0195 expression at time 0 was barely detectable. In comparison, mean normalized log₂ values for expression of SSA_0195 at 2.5, 5, and 15 min post-CSP addition were 32.6, 36.1, and 30.6, respectively, indicating strong and rapid induction. Inclusion of SSA_0195 along with the missing comC gene suggests a total of six early genes: comCDE, comX, SSA_1889, and SSA_0195.

Late-induced genes

Apart from the early genes indicated above, all other upregulated genes (Table S1 in the supplemental material) were classified as “late”. All were expressed at levels lower than the early genes at 2.5 min. Those genes whose expression was induced most strongly and persistently are indicated in Fig. 2. Most are orthologs of genes previously associated with competence, including 9 of the 14 late genes shown to be essential for transformation in S. pneumoniae (Peterson et al., 2004). The remainder are orthologs of CSP-induced genes previously identified in S. pneumoniae, S. gordonii, or S. mutans, with the exception of SSA_0639 and SSA_0640, which encode hypothetical proteins and are found only in S. sanguinis.

In S. pneumoniae, the late response was followed by a “delayed” response, which was characterized by a continued increase in expression beyond that of the early and late genes (Peterson et al., 2004). The delayed class contained several genes encoding chaperones and other stress-response proteins. A comparable response was not evident in S. gordonii (Vickerman et al., 2007). In S. sanguinis, four orthologs of S. pneumoniae delayed-response genes, including hrcA, grpE, dnaJ, and groES were significantly upregulated. Their expression did not display delayed kinetics, however, peaking 10 to 15 min after CSP addition.

The most common function of the late genes, as assessed by their Cluster of Orthologous Group (Tatusov et al., 2001) classification, was “DNA replication, recombination, and repair.” Seventeen genes were assigned to this category; fewer than half that number were assigned to any other. Table S2 in the supplemental material presents a comparison of the late/delayed response of S. pneumoniae (Peterson et al., 2004), S. sanguinis, and S. gordonii (Vickerman et al., 2007), as well as CSP-induced genes identified in S. mutans (Perry et al., 2009). Orthologous genes upregulated in at least two of these species are shown, along with any orthologs present in the other species. Approximately half of the late genes (65 of 118) contained orthologs that were also upregulated by CSP in S. pneumoniae, S. gordonii, or S. mutans. We suspected that many of the apparent differences among the species were due to technical issues with individual arrays or induction levels falling just below the level of significance. In particular, we noted that the five SK36 orthologs of late genes identified previously as required for transformation in S. pneumoniae (Peterson et al., 2004) that were not identified as late genes in our study were deleted during the filtering process rather than because of unchanged expression. Nevertheless, we wanted to confirm that this was not indicative of a fundamental difference in transformation machinery in the two species. We took advantage of an existing collection of SK36 allelic exchange mutants in which genes were replaced by the aphA-3 gene encoding kanamycin resistance (Xu et al, manuscript in preparation) to test the effect on transformation efficiency of mutating 15 genes that were either orthologs of genes essential for transformation in S. pneumoniae, or were located in the same gene cluster. Transformation was assessed in SK36 and the mutants as above, except that the incubation with CSP and DNA was begun at a lower OD and extended to three hours for increased sensitivity, and surface plating was used. The results are presented in Table S2 in the supplemental material, along with previously published data for S. pneumoniae (Peterson et al., 2004) and S. mutans (Merritt et al., 2005). Our results were comparable to those published previously. The combined mutant transformation data from S. pneumoniae, S. sanguinis, and S. mutans suggest that the comY operon contains seven genes required for transformation. If so, this would suggest a shared core regulon of 16 late genes required for streptococcal transformation (Table S2).

The comparison also suggests gene clusters with no obvious connection to competence that are shared among a subset of species. For example, the fruRBK gene cluster encoding genes involved in fructose uptake and metabolism is conserved in all four species, and has been shown to be co-transcribed in S. mutans (Wen et al., 2001) and S. gordonii (Loo et al., 2003). Yet, the entire operon is upregulated in S. sanguinis and S. gordonii, but not the other two species. The significance of these differences is unknown.

Downregulated genes

Genes significantly downregulated by CSP were fewer in number than those upregulated (Table S1 and Fig. 2). The most common categorizations of these genes were as encoding hypothetical proteins (14 genes), ribosomal proteins (13 genes), or ABC transporters (11 genes). Genes belonging to these three categories were also prevalent among the genes downregulated by CSP in S. pneumoniae, S. gordonii, and S. mutans (Peterson et al., 2004; Vickerman et al., 2007; Perry et al., 2009). These commonalities in broad functional categories generally did not extend to individual orthologous genes, however. Among hypothetical protein genes, SSA_1941 was orthologous to SGO_1699, but there were no other orthologs in any of the three species that were downregulated. Similarly, fewer than half of the downregulated S. sanguinis ABC transporters had orthologs that were downregulated in any of the other three species. Most tellingly, there were no orthologous genes that were downregulated in all four species (data not shown). The combined results suggest that CSP-mediated gene downregulation in these species may result primarily from non-specific responses to changes in growth rate and other cellular properties rather than from concerted, gene-specific repression.

Examination of selected competence genes by qRT-PCR

To confirm the results observed from the microarrays, qRT-PCR reactions were performed with selected orthologs of known competence gene. Three early response genes, comD, comE, and comX, were examined, along with the late gene comEA and the suspected late gene, comYD. Orthologs of the latter gene are required for competence in S. pneumoniae and are CSP-induced in S. pneumoniae, S. gordonii, and S. mutans. The microarray data suggested that the tuf gene, encoding elongation factor Tu, was expressed highly and constitutively at all time points. This gene was therefore chosen as a normalization control. Another housekeeping gene frequently used for normalization, gapA, was included as an additional control. As shown in Fig. 3, the results corresponded to those obtained from the microarray, with early gene expression peaking at 5 min and the late gene comEA peaking later and declining less during the course of the study. The comYD gene was not identified as a late gene in the microarray analysis, despite upregulation ≥32-fold at 5 min and beyond, because it was deleted during the filtering process (data not shown). Fig. 3 indicates, however, that its expression pattern was almost identical to that of comEA, confirming its expected status as a late gene. This is also consistent with our finding that comYD is required for competence. Expression of the gapA gene was similar at all time points, as expected, suggesting that either the tuf or gapA genes can be used as normalization controls for these studies.

FIG. 3.

Examination of the expression of selected competence-related genes by qRT-PCR. Values are presented in units of log₂ of the ratio of each gene’s normalized expression relative to that at T₀. Samples were examined in triplicate in two or three separate experiments. Error bars indicate SEM.

Conserved upstream regulatory sequences

The simplest model for CSP-dependent upregulation of early genes entails direct binding of phosphorylated ComE to upstream regulatory sites. By aligning the sequences upstream from the comC genes of S. pneumoniae, S. mitis, S. oralis, S. cristatus, S. gordonii, and S. sanguinis, Ween et al identified a consensus motif consisting of two ~9-bp imperfect direct repeats, which was predicted to serve as a ComE binding site (Ween et al., 1999). This pattern is consistent with known binding sites for transcriptional regulators in the AlgR/AgrA/LytR family, to which ComE belongs (Nikolskaya and Galperin, 2002).

To identify possible ComE binding sites in S. sanguinis, the alignment produced by Ween et al (Ween et al., 1999) from the comC-proximal sequences of six mitis group species was used to create a position-specific scoring matrix (PSSM), which was then searched against an S. sanguinis SK36 nucleotide database, as described in METHODS. Statistically significant matches were found upstream of three genes: SSA_2394 (comC); SSA_1889; and SSA_0195 (Fig. 4A). The first two S. sanguinis genes were identified as early response genes by the microarray analysis (Fig. 2), and the third by qRT-PCR. The SSA_1889 match overlapped the gene’s start codon; however, use of an alternative ATG start codon 78 bp downstream from that annotated results in a match that is similarly spaced from a consensus -10 element (Fig. 4A) and start codon (not shown). This start codon results in a protein with an amino terminus similar or identical to that of its two closest homologs in GenBank, as determined by BLASTP analysis (data not shown), suggesting its authenticity.

FIG. 4.

Examination of putative upstream competence regulatory sites. Genes are indicated by name and/or locus tag. Streptococcal species included are: SSA, S. sanguinis; SP, S. pneumoniae; SMU, S. mutans; SGO, S. gordonii; SCR, S. cristatus; SMI, S. mitis; and SOR, S. oralis. (A) Alignment of putative ComE-binding sites and -10 promoter elements. The motif identified by Ween et al (Ween et al., 1999) is indicated for reference; MEME, a simplified representation of the PSSM created by the MEME program. Dots indicate identity to the PSSM or conserved -10 sequences; differences are shown. (B) Phylogenetic tree of the putative ComE binding sites shown in A (without -10 elements). Bootstrap values greater than or equal to 50% from 2000 replicates are indicated next to the branches. Scale indicates the number of base substitutions per site. (C) Alignment of putative S. sanguinis combox sequences. Conservation is indicated by the sequence logo, with the overall height of each stack indicating the sequence conservation at that position, as measured in bits, and the height of each symbol reflecting the relative frequency of the corresponding nucleotide (Crooks et al., 2004). Nucleotides present in more than half of the aligned sequences are indicated in bold. Numbers indicate distances to the downstream start codon. The S. pneumoniae consensus combox sequence (Peterson et al., 2004) is shown for comparison.

A second PSSM was created from the preceding three matches to allow for the possibility of S. sanguinis-specific sequence variations. The only additional significant match occurred upstream from SSA_0016, encoding comX—the final early response gene identified in our analysis. Interestingly, however, the site was farther upstream than the other genes and in the reverse orientation (Fig. 4A). Examination of the comX upstream sequence in the positive orientation uncovered an alignment that had less similarity, but was appropriately spaced relative to a -10 sequence, as shown in Fig. 4A. Inclusion of the comX sequence in either orientation for another round of PSSM creation and searching resulted in no additional significant matches. S. pneumoniae (Peterson et al., 2004), S. gordonii (Vickerman et al., 2007), and S. mutans (van der Ploeg, 2005) have also been examined for ComE-binding sites using different methods. To better compare the four species, the same iterative PSSM strategy used above was applied to the other three species, except that the site between the S. mutans comC gene (SMU.1915) and SMU.1914c was also included in the PSSM. Significant matches from these species, along with the S. cricetus, S. mitis, and S. oralis comC sites, are indicated in Fig. 4A. The degree of sequence divergence among the sites is depicted in Fig. 4B.

Previous studies have also identified a conserved “combox” or “cinbox” site upstream from a number of late competence genes, which is thought to serve as the binding site for the competence-specific sigma factor—ComX in S. pneumoniae (Campbell et al., 1998; Peterson et al., 2004) and S. mutans (Okinaga et al., 2010) or ComR in S. gordonii (Vickerman et al., 2007). Combox sequences were readily detected preceding 19 late genes or gene clusters in S. pneumoniae TIGR4 (Peterson et al., 2004) 12 in S. gordonii (Vickerman et al., 2007) and 15 in S. mutans (Okinaga et al., 2010). Nine of these sites were observed in all three studies, and a tenth was shared in the latter two. The same ten sites were identified in S. sanguinis, as shown in Fig. 4C. An eleventh site was observed upstream of SSA_2192, an S. sanguinis late gene with no orthologs in S. pneumoniae or S. mutans. Further examination using a variety of strategies yielded no further matches to the combox in the S. sanguinis genome.

DISCUSSION

This study has characterized the transcriptional response associated with competence induction in S. sanguinis, providing an unprecedented comparison of streptococcal competence regulatory systems. The magnitude of the overall transcriptional response to CSP in S. sanguinis—122 genes upregulated, and 83 downregulated—was similar to that seen previously in other streptococci. One S. pneumoniae study identified 124 and 64 up and downregulated genes, respectively (Peterson et al., 2004), while the corresponding numbers for another were 105 and 42 (Dagkessamanskaia et al., 2004). In S. gordonii, 162 and 89 were up- and downregulated (Vickerman et al., 2007), while 202 and 75 were similarly classified in S. mutans (Perry et al., 2009). In addition, the genes identified as downregulated or upregulated late in S. sanguinis were similar to those identified in previous studies with regard to general function.

Given this result, it was surprising to see the sparseness of the early CSP-induced transcriptional response in S. sanguinis relative to S. pneumoniae and S. gordonii. There are a number of possible explanations for this. One possibility is that this difference stems from different growth conditions employed in the studies. Previous studies identifying early genes in S. pneumoniae (Peterson et al., 2004) and S. gordonii (Vickerman et al., 2007) employed growth in acidic conditions to prevent natural competence induction while we used a comC deletion mutant and growth at neutral pH. However, an identical strategy to ours was used previously in another S. pneumoniae study (Dagkessamanskaia et al., 2004), and although genes were not categorized as early or late or examined at 2.5 min, 16 of the 23 previously identified early genes (Peterson et al., 2004) were significantly upregulated at 5 min, suggesting a similar early response. Our use of CSP at 220 ng/ml is also comparable to the CSP concentrations of 100 ng/ml for S. gordonii (Vickerman et al., 2007) and 200 (Peterson et al., 2004) or 500 ng/ml (Dagkessamanskaia et al., 2004) for S. pneumoniae, suggesting that this was not a factor. Another possibility is that this discrepancy results from differences in statistical procedures employed in the different studies. This again seems unlikely. Examination of the raw data revealed no evidence of induction of any additional early genes, apart from those already indicated in Table 2. Moreover, the set of early genes identified in S. sanguinis coincides almost exactly with the intersection of early genes shared by S. pneumoniae and S. gordonii. This suggests that the S. sanguinis early response is confined almost entirely to genes required for competence induction.

It was also interesting that our bioinformatic analysis detected a putative ComE-binding site upstream from every early gene operon identified by expression analysis, and at no other sites in the genome. Peterson et al (Peterson et al., 2004) identified 13 gene clusters in S. pneumoniae as belonging to the early response by microarray analysis. A ComE motif was noted upstream from 9 of these, suggesting both direct and indirect control of early gene expression by ComE. In contrast, Vickerman et al (Vickerman et al., 2007) failed to detect a ComE motif upstream from any of the 18 transcriptional units identified by microarray analysis as belonging to the early response in S. gordonii. ComE binding sites were not examined in the S. mutans microarray analysis (Perry et al., 2009), although such sites have been noted previously upstream from five bacteriocin genes (van der Ploeg, 2005). When we applied the same bioinformatic strategy to the other three species, four significant matches were detected in S. pneumoniae; three coincided with ComE sites noted previously (Peterson et al., 2004), while the fourth occurred upstream from SP_0429, which was also classified as an early gene by microarray analysis (Peterson et al., 2004). We detected all five bacteriocin genes noted previously in S. mutans (van der Ploeg, 2005), one site upstream from the bacteriocin gene SMU.1896c, and a non-bacteriocin gene upregulated 1.74-fold by CSP, SMU.751 (Perry et al., 2009). SMU.1896c was not identified as CSP regulated by microarray analysis (Perry et al., 2009), and a transcriptional reporter analysis has indicated that SMU.1896c expression is not affected by ComE or CSP (van der Ploeg, 2005). Although a -10 element was not part of the PSSM used for these analyses, the alignment in Fig. 4A indicates that a highly conserved -10 element is located 28–30 bp upstream from every gene in all four species except SMU.1896c. The absence of a consensus -10 element for SMU.1896c suggests an explanation for its lack of induction by ComE and CSP, as well as supporting the specificity of the PSSM.

Only two potential ComE-binding sites were identified by the PSSM in S. gordonii (Fig. 4A), both of which preceded genes identified as belonging to the early response (Vickerman et al., 2007). The phylogenetic analysis in Fig. 4B indicates that these sites are more closely related to the other mitis group sequences than are most of the S. mutans sites. It is thus remarkable that so few sites were identified upstream from S. gordonii early genes. This suggests there is either more similarity in ComE binding sites among species than among sites within this single species, or there are very few early genes regulated directly by ComE in S. gordonii. Our findings with S. sanguinis suggest more divergence in mechanisms of early gene regulation within the mitis group than previously envisioned.

It has recently been reported that comX expression and transformation are upregulated by a stand-alone transcriptional regulator, termed ComR, and a small peptide, named ComS in Streptococcus thermophilus and Streptococcus salivarius (Fontaine et al., 2010) and by an orthologous system found in S. mutans and the pyogenic and bovis group streptococci (Mashburn-Warren et al., 2010). Related consensus sequences identified upstream from comX were predicted to serve as binding sites for ComR in the two orthologous systems (Mashburn-Warren et al., 2010). Neither of the consensus sequences were identified by the authors upstream of comX in S. sanguinis or other mitis group species examined (Mashburn-Warren et al., 2010). We confirmed these findings for S. sanguinis using an identical analysis, as well as a PSSM analysis similar to that described earlier for detection of ComE and ComX binding sites. It thus seems unlikely that comX expression is controlled by ComRS orthologs in S. sanguinis or the other mitis group species examined to date.

One goal of this study was to identify the CSP transporter(s) in SK36. In S. pneumoniae, early induction of the comAB genes encoding the CSP transporter is important for competence development, since CSP transport appears to be rate-limiting for transformation (Martin et al., 2000). Although ComAB orthologs are clearly not encoded in the S. sanguinis genome (Xu et al., 2007), we reasoned there might exist early-induced genes encoding ComAB analogs that would serve to export S. sanguinis CSP. SSA_0195 and SSA_1889 are obvious candidates. SSA_0195 is annotated as a hypothetical protein with weak homology to a putative lipoprotein in Streptococcus sp. M143 (accession no ZP_01835889.1). It also possesses weak similarity to the comM gene of S. pneumoniae (spr1762 or SP_1945), which has been shown to be an early CSP-induced gene that provides immunity to pneumococcal lysins (Håvarstein et al., 2006). SSA_1889 is annotated as a conserved uncharacterized protein (Xu et al., 2007) whose closest match in GenBank is another conserved hypothetical protein in S. oralis (ZP_06611623.1). It had no mutually best BLASTP matches with any other sequences in GenBank. Both SSA_0195 and SSA_1889 have weak homology to CAAX amino terminal protease family proteins (Pei and Grishin, 2001), although the two proteins share no significant similarity to one another. To determine whether either gene played a role in CSP processing and transport, these genes were deleted individually and together by allelic exchange mutagenesis. None of the three mutants exhibited a greater than two-fold reduction in transformation efficiency in our standard assay, in the presence or absence of exogenous CSP (data not shown). It thus appears that neither gene encodes a CSP transporter nor do they have any other essential role in transformation. Other potential roles of these genes are under investigation.

The preceding results suggest that S. sanguinis CSP may be processed and exported by a com-independent transporter with a dual function. A logical candidate would be a bacteriocin transporter. A large number of bacteriocin-like peptides are similar to non-S. sanguinis ComC sequences in possessing a Gly-Gly motif, and are processed and transported similarly to CSP by ComAB homologs (Håvarstein et al., 1995a). The genes originally identified as encoding the S. mutans CSP transporter were later shown to serve as a bacteriocin transporter (Hale et al., 2005), and the ComAB proteins of S. gordonii have been shown to process and secrete both CSP and bacteriocin-like peptides (Heng et al., 2007). Moreover, bacteriocin production is closely tied to competence induction in S. pneumoniae (Peterson et al., 2004; Martin et al., 2006), S. gordonii (Lunsford and London, 1996; Vickerman et al., 2007), and S. mutans (Hale et al., 2005; van der Ploeg, 2005; Perry et al., 2009). However, S. sanguinis differs from these other three species in lacking not only a Gly-Gly-containing ComC protein, but also any recognizable bacteriocin genes. Early reports of bacteriocin production by S. sanguinis employed isolates that are currently classified as S. gordonii (Schlegel and Slade, 1974) or are of uncertain taxonomy (Fujimura and Nakamura, 1979), and a database mining program (de Jong et al., 2006) failed to identify any classical bacteriocin candidates in the SK36 genome (data not shown). If S. sanguinis does encode bacteriocins, it is possible that the unidentified ComC transporter is also responsible for their transport. It is also possible that S. sanguinis CSP undergoes sec-dependent secretion, although no recognizable signal sequence is present (data not shown). If CSP is secreted by a housekeeping transporter, the comCDE and comX genes S. sanguinis may represent the minimal dedicated system required for CSP regulation of competence gene expression. Studies are currently underway to identify the CSP transporter in S. sanguinis.