Identification of Group A Streptococcus Antigenic Determinants Upregulated In Vivo

Abstract

Group A Streptococcus (GAS) causes a range of diseases in humans, from mild noninvasive infections to severe invasive infections. The molecular basis for the varying severity of disease remains unclear. We identified genes expressed during invasive disease using in vivo-induced antigen technology (IVIAT), applied for the first time in a gram-positive organism. Convalescent-phase sera from patients with invasive disease were pooled, adsorbed against antigens derived from in vitro-grown GAS, and used to screen a GAS genomic expression library. A murine model of invasive GAS disease was included as an additional source of sera for screening. Sequencing DNA inserts from clones reactive with both human and mouse sera indicated 16 open reading frames with homology to genes involved in metabolic activity to genes of unknown function. Of these, seven genes were assessed for their differential expression by quantitative real-time PCR both in vivo, utilizing a murine model of invasive GAS disease, and in vitro at different time points of growth. Three gene products—a putative penicillin-binding protein 1A, a putative lipoprotein, and a conserved hypothetical protein homologous to a putative translation initiation inhibitor in Vibrio vulnificus—were upregulated in vivo, suggesting that these genes play a role during invasive disease.

Group A Streptococcus (GAS) causes a spectrum of diseases in humans ranging in severity from mild infections such as pharyngitis, to diffuse invasive infections such as cellulitis, to life-threatening severe invasive diseases such as necrotizing fasciitis (NF) and streptococcal toxic shock syndrome (STSS) (7). Complications of GAS infections can also lead to rheumatic fever and glomerulonephritis, which are immunologically mediated diseases (7). Despite the fact that a large repertoire of known and suspect virulence factors that contribute to the pathogenicity of GAS have been identified and considerable progress has been made in understanding the roles that these factors play in streptococcal infections, the molecular basis for the varying severity of GAS infections remains unclear. Furthermore, there has been a resurgence in the incidences and severity of invasive GAS infections since the mid-1980s in many developed countries (9, 21, 42); whether this is a result of the emergence of virulent clones, the exposure of GAS to an immunologically naive population, or a combination of both is still unclear.

During disease, GAS must adapt to a range of environments (e.g., blood, nasopharyngeal mucosa, skin, etc.), and survival in any one niche would likely require the expression of a distinct subset of virulence factors. Hence, genes that are upregulated in vivo may play an important role in disease. Conventional genetic and biochemical approaches used to study GAS virulence determinants cannot mimic the complex and dynamic environmental stimuli that occur at the site of infection. Recently, however, several technologies have been developed to identify genes that are expressed during an infection, including in vivo antigen technology (IVIAT), in vivo expression technology (IVET), differential fluorescence induction (DFI), and signature-tagged mutagenesis (STM) (15, 19, 40). A wide array of in vivo-induced (IVI) genes involved in adhesion, invasion, nutrient acquisition, regulation, and structural integrity have been identified by using these technologies and mutants deficient in some of these genes indicate that they have a role in virulence (reviewed in references 15, 19, and 40).

The promising data generated by these technologies prompted us to investigate the pathogenesis of invasive GAS infections by using IVIAT (16, 17), a technique that relies on antibodies produced during a natural infection. The IVIAT scheme consists of three steps: (i) serum selection and adsorption, (ii) construction of a GAS genomic expression library, and (iii) screening of the GAS genomic library with the selected, pooled, adsorbed sera. The advantages of IVIAT are that antigenic determinants induced in vivo will be identified and that, unlike other in vivo technologies, it does not limit the investigator to a single strain of bacteria for the identification of IVI genes. This is important because no conclusive evidence has been found demonstrating that a particular serotype or clonotype of GAS is responsible for invasive disease (23, 25, 33, 45). In the present study, convalescent-phase sera from patients with invasive GAS disease and GAS-immunized mice sera were utilized as sources of antibodies produced during an invasive infection with GAS. The use of sera from multiple patients allows for the identification of a wide array of immunogens by taking advantage of the variability in host immune responses to IVI antigens. Screening of the GAS genomic library resulted in the identification of 16 putative IVI genes. Of these, seven genes were selected for analysis by real-time PCR (11) to confirm their in vivo upregulation. Three genes were shown to be upregulated in vivo; a putative penicillin-binding protein 1A gene (pbp1A), a putative lipoprotein gene (atmB), and a conserved hypothetical protein gene (tdcF).

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

To obtain in vitro-induced antigens, GAS strains (Table 1) were cultured overnight at 37°C in Todd-Hewitt broth (Difco Laboratories, Detriot, MI) under either aerobic or microaerobic (5.0% CO₂) conditions, and whole cells, cell extracts, and spent media were prepared as described below. GAS strains for inoculation of mice were prepared as previously described (2). To obtain mRNA from in vitro cultures, MGAS166 (32) was cultured aerobically at 37°C and periodically sampled at various time points during its growth. The Escherichia coli strains utilized for the construction of the GAS genomic library were grown in Luria-Bertani broth (Difco Laboratories) at 37°C under aerobic conditions.

TABLE 1.

Eight invasive clinical GAS strains used for the construction of the GAS genomic library and for deriving the in vitro antigens used to adsorb sera from patients with invasive disease, GAS-immunized mice, and healthy individuals

M/T type of GAS strain	Source of GAS isolate	Case definition
M1/T1	Blood	Invasive
M1/T1	Throat	Invasive
M3/T3	Blood	STSS
M4/T4	Blood	Invasive
M6/T nontypeable	Blood	STSS
M11/T11	Blood	Invasive
M12/T12	Blood	Invasive
M28/T11/28	Blood	Invasive

In vitro antigen preparation.

Equal volumes of each strain (Table 1) of GAS cultures grown to late log phase were pooled and centrifuged, the spent media were removed, and whole cells were resuspended in 1× phosphate-buffered saline (PBS). Cell extracts were prepared from whole cells that were concentrated 10-fold and processed with an FP120 Fastprep Machine (Bio 101, Mississauga, Ontario, Canada) at a setting of 6.0 for 30 s and placed on ice for 30 min to allow the beads to settle, and cell extracts were removed by aspiration. Denatured cell extracts were obtained by placing cell extracts in a boiling water bath for 10 min. The pooled, cell-free supernatant was freeze dried by using a Benchtop 3.3/Vacu-Freeze Dryer (VirTris Company, Gardiner, NY) and resuspended in 1× PBS. All antigen preparations were stored at −70°C for up to 1 month until ready for use.

Human sera.

Convalescent human sera, collected between 2 to 3 weeks after diagnosis, were selected from 14 of 21 patients with invasive GAS infections, such as NF and STSS (Table 2). Note that these serum samples were not from the same patients as those from whom the eight invasive GAS isolates were collected (Table 1). Hence, the strain and serum samples were not paired. Control human sera were obtained from human subjects with no previous history of invasive GAS infection.

TABLE 2.

List of sera from patients with invasive GAS disease used for screening the GAS genomic library

Serum no.	Diagnosis	Source of GAS isolate	M type of GAS isolate
1	NF and STSS	Abscess	M1
2	Cellulitis	Abscess	M nontypeable
3	Cellulitis	Blood and abscess	M1
4	Cellulitis and arthritis	Blood	M12
5	NF and STSS	Blood and abscess	M75
6	NF	Tissue	N/A
7	Cellulitis	Blood and abscess	M28
8	Arthritis	Throat or aspirate	M3
9	NF and STSS	Abscess	M1
10	Necrotizing myositis and STSS	Blood	N/A
11	Peritonitis	Blood	N/A
12	Pneumonia	Blood	M1
13	NF	Abscess	M4
14	Cellulitis	Abscess	M1

Murine subcutaneous model. (i) Mouse sera.

Immunocompetent 4-week-old, female, crl:SKH1 (hrhr) hairless mice (Charles River Laboratories, Wilmington, MA) were utilized for the invasive subcutaneous infection model of GAS (2). This mouse model was used for generating anti-GAS mouse antibodies for screening the GAS genomic library. Two mice were infected with each of the eight invasive GAS strains (Table 1) to give a total of 16 infected mice. The immunization protocol included an initial inoculation of 10⁶ CFU, followed by a primary boost (10⁶ CFU for those mice that developed lesions and 10⁸ CFU for those mice that did not develop a lesion) after 2 weeks and a secondary boost (10³ CFU) after an additional 2 weeks. Sera from 10 noninfected mice were used as controls against sera from GAS-immunized mice. Serum was obtained by cardiac puncture and stored at −70°C. All experimental procedures were in accordance with the principles of the Animal Care Committee of Mount Sinai Hospital, Toronto, Ontario, Canada.

(ii) In vivo gene expression.

Mice were inoculated subcutaneously with MGAS166 for analysis of gene expression in vivo. Necrotic lesions that developed at the site of inoculation were excised from each of two mice sacrificed at 24 and 48 h postinoculation, snap-frozen in liquid nitrogen, and stored in −80°C.

(iii) Attenuation of virulence.

Insertionally inactivated mutants were tested for attenuation of virulence in the invasive subcutaneous infection model of GAS (2). Ten mice each were inoculated with 10⁶ CFU of wild-type NZ131 and GAS 5448, as well as their corresponding atmB and pbp1A insertional mutants. An additional 10 mice inoculated only with cytodex acted as controls. The mice were monitored for weight gain/loss, lesion formation, size, and rate of healing for a period of 10 days. At 3 and 10 days postinoculation 3 and 7 mice, respectively, were sacrificed from each of the 10 mice inoculated with the wild-type parental strains and their corresponding mutants. The necrotic lesions in each of these mice were excised, macerated with a disposable tissue grinder (Kendall), resuspended in 1.0 ml of 1× PBS, and plated on Columbia blood agar (CBA) plates to determine the CFU of bacteria/ml. The insertionally inactivated mutants were also plated on CBA plates containing 2.5 μg of chloramphenicol/ml to ensure that the plasmid was not lost during growth in vivo and that relative counts on CBA alone were the same.

Indirect ELISA.

An indirect enzyme-linked immunosorbent assay (ELISA) was used for screening the human and mice sera by using in vitro-derived GAS antigens (refer to in vitro antigen preparation). Immulon IIHB plates (Dynex Technologies, Chantilly, VA) were coated overnight at 4°C with each antigen (whole cells, cell extracts, and spent media), which was diluted in freshly prepared carbonate bicarbonate buffer consisting of 20 mM sodium carbonate (Fisher Scientific, Nepean, Canada) and 50 mM sodium bicarbonate (BDH Chemicals, Toronto, Ontario, Canada). The assay procedure described previously was followed (6). The antibody titer was defined as the highest serial dilution of serum at which the optical density at 490 nm (OD₄₉₀) was two standard deviations above the mean OD₄₉₀ of the negative controls (without primary antibody or without antigen). Antibody titers were converted to logarithmic values [log₂ (x), where x equals the reciprocal of the serum dilution] for calculation of geometric means.

Serum adsorption.

Equal volumes of selected invasive patient sera and GAS-immunized mice sera were pooled in a species-specific manner and successively adsorbed with in vitro-derived GAS antigens. In addition, sera from 14 healthy individuals were also pooled and successively adsorbed with in vitro-derived GAS antigens. The successive adsorption steps consisted of five times with whole cells, one time with cell extracts, one time with denatured cell extracts, and one time with spent media. Adsorptions were carried out by incubating the pooled sera overnight at 4°C with antigen-saturated nitrocellulose membranes (Millipore, Bedford, MA). After each successive adsorption, the pooled sera were removed, and the membrane was washed with 500 μl of 1× PBS, which was then added to the pooled sera. To check the efficacy of each adsorption step, a 10-μl aliquot of the serum was removed after each adsorption and an indirect ELISA was performed.

Construction of an inducible expression GAS genomic DNA library.

Chromosomal DNA from eight GAS strains (Table 1) was extracted by using a cetyltrimethylammonium bromide (CTAB) protocol (47). The library was constructed by partial Sau3AI digestion of the genomic DNA that was ligated into pET30-abc vectors (Novagen, Inc., Madison, WI) and electroporated into E. coli DH10B nonexpression cells (Invitrogen, Toronto, Ontario, Canada) as described previously (24).

Genomic library screening.

An aliquot of the plasmid DNA library in E. coli DH10B nonexpression hosts was extracted by using the QIAprep Spin Miniprep kit (QIAGEN, Inc., Toronto, Ontario, Canada) and transformed into chemically competent E. coli BL21(DE3) expression host (Novagen). The library was screened by colony Western blot analysis with pooled adsorbed or unadsorbed human and mouse sera in a species specific-manner as described previously (24) utilizing a Western blot detection kit (Bio-Rad Laboratories, Hercules, CA). Clones expressing the seven genes selected for further analysis by real-time PCR were also screened with pooled, adsorbed sera from healthy individuals to determine their reactivity to this serum set.

DNA sequencing.

DNA sequencing was done with an ABI Prism 377 automatic DNA sequencer by the double-strand dideoxy chain termination method at the Hospital for Sick Children Sequencing Facility, Toronto, Ontario, Canada. Sequences were analyzed by using the BLAST algorithm of the National Center for Biotechnology Information.

RNA isolation.

Duplicate cultures of MGAS166 grown in vitro were harvested at OD₆₀₀s of 0.1, 0.2, 0.4, 0.6, 0.8, and 1.0. OD₆₀₀s of 0.1 to 0.2, 0.2 to 0.7, and 0.8 to 1.0 correspond to lag phase, log phase, and stationary phase, respectively. The bacterial pellet was snap-frozen in liquid nitrogen and stored at −80°C. Frozen mouse tissue biopsies were macerated with a disposable tissue grinder (Kendall, Mansfield, MA). To isolate RNA, the bacterial pellet or ground tissue was resuspended in TRIzol reagent (Invitrogen), and cells were lysed by using an FP120 Fastprep machine (Bio 101) at a setting of 6.0 for 30 s. RNA was then treated with RQ1 DNase (Promega, Madison, WI), and its integrity was verified by agarose gel electrophoresis. Bacterial RNA from the infected mouse tissue was enriched by using a MICROBEnrich kit (Ambion, Inc., Austin, TX) in accordance with the manufacturer's instructions.

Real-time PCR analysis.

DNase-treated RNA samples were reverse transcribed by using a first-strand cDNA synthesis kit (MBI Fermentas, Inc., Hanover, MD) in accordance with the recommendations of the supplier. Controls for cDNA synthesis and DNase treatment included a no-RNA template sample and one without reverse transcriptase. The real-time PCR assays were performed by using a Cepheid SmartCycler system (Sunnyvale, CA) and a Quantitect SYBR-Green PCR kit (QIAGEN). Each 25-μl reaction included 2 μl of cDNA (400 to 800 ng), 250 nmol of each primer (Table 3), and 2× SYBR-Green mix. The reactions were cycled in the Smart Cycler with the following parameters: 95°C for 15 min for the hot-start, followed by 40 cycles of 94°C for 30 s, annealing at optimal temperature (Table 3) for 30 s, and primer extension at 72°C for 30 s. The primers were designed by using homologous regions of the genomic DNA sequences of M1 (SF370), M3 (MGAS315), and M18 (MGAS8232) that are available on the NCBI database.

TABLE 3.

List of target genes, normalizing gene (gyrA), primers, and optimal annealing temperatures used for the real-time PCR analysis

Gene	Optimal annealing temp (°C)	Primer
		Orientationa	Sequence
atmB	65	F	5′-TCTCTGGAACCAGTCAAGATGG-3′
		R	5′-AGAGGGCACGGCTTTCATTG-3′
coaA	60	F	5′-CCTCACACAAGAGGAACTTAAAAG-3′
		R	5′-GCAATTTTGTAAACCTGGATCAG-3′
pbp1A	60	F	5′-CTGGGGTTCTACCATGAAGC-3′
		R	5′-TCCAGGCCAATAATAGACTGAG-3′
SPy1674	60	F	5′-CAGGTAAAACCACGACCATCAGC-3′
		R	5′-TCAGGAGAATCTGGCACATAAGC-3′
SPy1733	60	F	5′-TCCAAGTGGGAAGGAAACAG-3′
		R	5′-TTGGGTCCAGATAACGTGG-3′
SPy1784	60	F	5′-CTTTCTAAATTAATGGTGATGCTGC-3′
		R	5′-TGTTCTAGATCTGCCTGTTGAGA-3′
tdcF	60	F	5′-TATCTGCCATTAAAGCGCTC-3′
		R	5′-GGGGCTCACATTAGATGCAG-3′
gyrA	65	F	5′-AGCGAGACAGATGTCATTGCTCAG-3′
		R	5′-CCAGTCAAACGACGCAAACG-3′

^aF, forward; R, reverse.

Real-time PCR analysis of in vitro- and in vivo-derived MGAS166 RNA was performed in triplicate with both the test gene and the gyrase A (gyrA) primers from two independently derived RNA samples. Standard curves for each gene, generated with serially diluted known concentrations of MGAS166 genomic DNA plotted against the C_T values, were used to determine cDNA levels for gyrA and each test gene. gyrA was chosen as an internal standard for normalizing target gene expression because its expression was not affected under a variety of experimental conditions (14, 41). Target gene expression was normalized with gyrA expression in each sample by dividing the target gene cDNA concentration by the gyrA cDNA concentration. This ratio of target gene to gyrA is referred to as the normalized gene expression. The normalized target gene expression for each of the duplicate in vitro and in vivo samples was averaged, and the standard deviations were obtained. The data were analyzed in this manner not only to measure expression levels under the conditions tested but also to ensure that there was little variability within each sample and between duplicates.

Construction of insertionally inactivated mutants.

Targeted mutagenesis of atmB and pbp1A was generated utilizing the PCR ligation mutagenesis technique (12, 26). Chromosomal regions upstream and downstream of the target genes were amplified and ligated to the erythromycin resistance gene ermB (12, 26) via FseI and AscI restriction endonuclease sites, respectively, incorporated onto the 5′ end of appropriate primers (Table 4). The P1/P2 and P3/P4 primer pairs for each gene amplified the upstream and downstream regions of the target genes, respectively. The upstream fragments and downstream fragments were ligated to the ermB cassette via the AscI and FseI restriction sites and cloned into the temperature-sensitive shuttle vector pVE6007Δ, specifying chloramphenicol (cat) (12). The resultant plasmids were transformed into two GAS serotypes: M49 serotype strain NZ131 and M1 serotype strain GAS5448. Chloramphenicol- and erythromycin-resistant transformants were identified at the permissive temperature for plasmid replication (30°C). Single-crossover Campbell-type genomic insertions were selected by incubation at a nonpermissive temperature (37°C) while maintaining chloramphenicol and erythromycin selection. Genomic DNA was examined in each of the mutant strains by PCR and Southern hybridization to confirm the single-crossover of the plasmids into the appropriate region of the genomic DNA.

TABLE 4.

Primers used for the insertional inactivation of GAS pbp1A and atmB genes

Primer	Nucleotide sequencea
atmB-P1	5′-AAC AAC ATT TCC TTT ACC AAA GAG-3′
atmB-P2	5′-GGC GCG CCG CTA AAG CGA TTC CGA CTA TTC-3′
atmB-P3	5′-GGC CGG CCC CCT TCA AAT GGT ATA GAT GTA TCC G-3′
atmB-P4	5′-GCG CCA AAA TTT GTT TAG TG-3′
pbp1A-P1	5′-GCG TTT CAA ACA AGA ATA TAG TAG C-3′
pbp1A-P2	5′-GGC GCG CCC GGC TAA GAA TTG CAC TTA ATA C-3′
pbp1A-P3	5′-GGC CGG CCC GCT TCT AGC CAA ACG ACT TCA G-3′
pbp1A-P4	5′-GTA AAG AAT TCT TAA CAG CTG ACG-3′
erm-F	5′-GGC GCG CCC CGG GCC CAA AAT TTG TTT GAT-3′
erm-R	5′-GGC CGG CCA GTC GGC AGC GAC TCA TAG AAT-3′

^aEngineered restriction endonuclease recognition sequences (in boldface) are as follows: AscI, GG/CGCGCC; FseI, GGCCGG/CC.

RESULTS

Sera selection and adsorption.

Sera were selected based on a titer of at least 1/10,000 against GAS-derived antigens as determined by an indirect ELISA. Of 21 invasive patient sera, 14 were selected based this criterion (Table 2) and were compared to sera from 14 individuals without a previous history of invasive GAS infections. The observed serum titers revealed that there was no distinct cutoff between the reactivity of the sera from patients with invasive disease and those from healthy individuals, particularly when GAS whole cells and cell extracts were used as antigens for the indirect ELISA (Fig. 1). However, a distinct cutoff was obtained in the reactivity of sera from 16 GAS-immunized mice versus sera from 10 nonimmunized mice such that sera from GAS-immunized mice were clearly more reactive than those from nonimmunized mice (Fig. 2). The sera from the 16 GAS-immunized mice were pooled, as were the 14 selected sera from patients with invasive disease. Pooled sera were adsorbed against in vitro-derived GAS antigens in order to create a serum set comprised of antibodies to proteins that are upregulated in vivo. A decrease in the reactivity of the sera was observed after each sequential adsorption of the pooled sera to the in vitro-derived GAS antigens (data not shown).

FIG. 1.

Comparison of the reactivity of sera from patients with invasive disease and healthy individuals as observed by indirect ELISA utilizing GAS whole cells (A), cell extracts (B) and spent media (C) as the antigen. N, number of individuals or patients with the given serum titer.

FIG. 2.

Comparison of the reactivity of sera from GAS-immunized and nonimmunized mice as observed by indirect ELISA utilizing GAS whole cells (A), cell extracts (B), and spent media (C) as the antigen. N, number of mice with the given serum titer.

GAS genomic library construction and screening.

The GAS genomic library, prepared from eight strains causing invasive disease (Table 1), in E. coli DH10B consisted of approximately 60,000 clones. Reactive clones were identified by comparing duplicate colony blots screened with adsorbed sera and unadsorbed sera, with the latter sera acting as a positive control (Fig. 3). The reactivity of the clones was confirmed by a secondary screen in which these clones were compared to clones bearing the vectors alone without any inserts present. Sequence analysis of the clones reactive to both human and mouse sera indicated partial or complete sequences of 16 putative IVI genes with known and unknown functions (Table 5). Clones expressing partial sequences of the seven genes, selected for further analysis by real-time PCR (see real-time PCR analysis of gene expression), were also tested for their reactivity to pooled, adsorbed sera from healthy individuals. Of the seven antigens, only pbp1A was clearly not reactive to the sera from healthy individuals. Although atmB and tdcF were reactive to the sera from healthy individuals, the intensity of this reaction, as seen using the Western blotting detection kit, was lesser than that observed with the pooled adsorbed serum sets from invasive patients and immunized mice.

FIG. 3.

Example of duplicate colony blots screened with unadsorbed (A) and adsorbed (B) human sera.

TABLE 5.

Putative IVI genes identified by IVIAT

Gene designation	Gene name	Descriptiona	Homologue	% Identity with homologue
SPy0032	purD	Phosphoribosylamine-glycine ligase	purD in S. agalactiae	95
SPy0319	atmB	Putative lipoprotein	atmB in S. mutans	73
SPy0630	agaW	Putative PTS-dependent N-acetylgalactosamine-IIC component	PTS system, IIC component in S. pneumoniae	83
SPy0631	agaV	Putative PTS-dependent N-acetylgalactosamine-IIB component	PTS system, IIB component in S. agalactiae	70
SPy0777	rexA	Putative ATP-dependent exonuclease, subunit A	Putative rexA in S. mutans	60
SPy1198		Putative repressor protein	Putative transcription regulator in S. mutans	58
SPy1233	coaA	Putative pantothenate kinase	coaA in S. agalactiae	71
SPy1355		Hypothetical protein	Hypothetical protein in S. mutans	65
SPy1356		Putative acetyltransferase	Acetyltransferase in S. agalactiae	53
SPy1649	pbp1A	Putative penicillin-binding protein 1A	pbp1A in S. agalactiae	71
SPy1674		Putative ABC transporter	ABC transporter in S. agalactiae	74
SPy1733		Putative transcription regulator	Putative transcription regulator in S. mutans	69
SPy1784		Putative ABC transporter	nisF in L. lactis	50
SPy1961	polC	DNA polymerase III (alpha subunit)	DNA polymerase II (alpha subunit) in S. agalactiae	84
SPy2060	tdcF	Hypothetical protein	Putative translation initiation inhibitor in V. vulnificus	60
SPy2063		Hypothetical protein	Ribosomal large subunit pseudouridine synthase in S. agalactiae	61

^aPTS, phosphotransferase system.

Real-time PCR analysis of gene expression.

Of the 16 putative IVI genes, 7 were selected for further analysis of gene expression by real-time PCR utilizing the MGAS166 strain. We selected these genes based on their putative function, such as those involved in cell morphology, metabolism, and acquisition of nutrients, as well as those of unknown function, in order to maximize on the variety of genes chosen for further analysis. The MGAS166 strain was selected for performing real-time PCR analysis because it is one of the M1T1 isolates used for the construction of the genomic library and for the adsorption of the pooled sera sets. In vitro expression of the seven putative IVI genes analyzed in the present study are shown in Fig. 4. Increased levels of atmB gene expression were observed at the early lag phase, late log phase, and stationary phase of growth, with the highest level of expression occurring at late log phase. The in vitro expression of four genes, coaA, pbp1A, SPy1784 encoding a putative ATP-binding cassette (ABC) transporter, and tdcF, reached maximal detectable expression at early log phase with lower constitutive levels observed at later stages of the growth. On the other hand, SPy1674 encoding a putative ABC transporter, and SPy1733 encoding a hypothetical protein, were expressed at higher levels during the late-log and stationary phases of growth. All of the putative IVI genes were expressed in vivo under the conditions tested (Fig. 5). With the exception of atmB and SPy1784, which were only expressed at 24 h postinoculation, the remaining five putative IVI genes were expressed at 24 and 48 h postinoculation in vivo. Since the aim of the present study was to identify genes that are upregulated in vivo, we determined the in vivo gene expression relative to the highest level of expression in vitro (Fig. 6). Of the seven genes analyzed, three genes—atmB, pbp1A, and tdcF—were upregulated in vivo relative to their highest level of expression in vitro (Fig. 6).

FIG. 4.

Average in vitro gene expression from MGAS166 grown to OD₆₀₀s of 0.1 to 1.0, spanning early lag to stationary phase, as determined by real-time PCR analysis. The standard deviations are presented from duplicate experiments.

FIG. 5.

Average in vivo gene expression from MGAS166 infecting soft tissue of mice sacrificed at 24 and 48 h postinoculation as determined by real-time PCR analysis. The standard deviations are presented from two mice each, sacrificed at 24 and 48 h.

FIG. 6.

In vivo gene expression at 24 h (A) and 48 h (B) relative to the highest level of expression in vitro by real-time PCR analysis.

Construction and in vivo analysis of insertionally inactivated mutants.

Insertionally inactivated atmB and pbp1A mutants in NZ131 and GAS5448 were confirmed by PCR and Southern utilizing the respective atmB-P1/P4, and erm-F/R primers (data not shown). In the murine model of invasive disease, all of the mice infected with both the wild-type parental strains and their mutants showed a significant weight loss at days 1 and 2 postinoculation relative to the cytodex controls. Although there was no significant weight loss between the mice inoculated with the wild-type and mutant strains, there was an observable trend of increased weight loss in the mice inoculated with the wild-type strains relative to the mice inoculated with their corresponding mutant strains (data not shown). A significant difference (P < 0.05) was observed in the rates of lesion formation and/or healing when we compared mice inoculated with the wild-type parent strains and mice inoculated with their corresponding mutants (Fig. 7). All statistical analyses were performed utilizing the Wilcoxon signed-rank test. Finally, the CFU/ml of all of the mutants relative to their corresponding parental strains was a log-fold lower at both days 3 and 10 postinoculation (data not shown).

FIG. 7.

Lesion size observed in mice at days 1, 2, 3, and 10 postinoculation with wild-type NZ131 and its mutants (A) and wild-type GAS5448 and its mutants (B). Bars represent median weight gain ± the range. ✽, Significantly different (P < 0.05) from lesion size in the wild-type parental strain on the same day as calculated by the Wilcoxon signed-rank test.

DISCUSSION

Our objective, based on the premise that genes expressed in vivo are more likely to be important in the pathogenic process, was to identify antigenic determinants that were upregulated during invasive GAS infections. This was accomplished by using IVIAT, which identifies genes that are expressed by a pathogen during an actual infection process. IVIAT has been previously applied to other gram-negative organisms, mycobacteria, and yeast (3, 4, 8, 18, 24), but this is the first report of the use of this technology in a gram-positive organism. The basis of IVIAT is that pooled sera from patients with disease adsorbed against in vitro-derived GAS antigens can be used to identify genes upregulated in vivo. Although the resurgence of invasive GAS disease has been associated with M1 and M3 serotypes, other M types have also been shown to cause invasive disease (23, 25, 33, 45). Since GAS isolates causing invasive disease are not clonal, the GAS genomic library was constructed by using eight strains of GAS, all of which were isolated from patients with invasive disease during the late 1990s in Ontario, Canada (Table 1). This strategy of IVIAT allows the investigator to analyze a pool of strains that represent various clinically important M serotypes, hence decreasing the probability of missing potentially important virulence factors during the screening process. In addition, sera from multiple patients with invasive disease (Table 2) were chosen for screening the genomic library to broaden the spectrum of antibodies reactive with GAS epitopes.

Interestingly, in our study, no distinct cutoff was observed between the reactivity of sera from patients with invasive disease and those from healthy individuals when using GAS whole cells and cell extracts as antigens for the indirect ELISA (Fig. 1). However, when GAS spent media was used as an antigen for the indirect ELISA, sera from patients with invasive disease indicated a higher reactivity compared to the sera from healthy individuals (Fig. 1). The occurrence of sera reactive to GAS antigens in healthy individuals is not surprising since nonsevere infections caused by GAS such as pharyngitis are common among all age groups, with highest infection rates occurring in the ages of 5 to 15 years (10). The seven clones bearing the partial sequences of the genes analyzed by real-time PCR (Fig. 4, 5, and 6) were evaluated for their reactivity to pooled, adsorbed sera from healthy individuals. The data indicated that, with the exception of the clone carrying the pbp1A insert, all of the clones were reactive to this serum set. There are two possible explanations for this observation. First, there might be antibodies present in the pooled, adsorbed sera from healthy individuals that are cross-reacting with atmB, coaA, SPy1674, SPy1733, SPy1784, and tdcF. Second, these genes may not be specific to invasive disease.

Screening of the GAS genomic library with sera from mice immunized with GAS through subcutaneous invasive infection resulted in the identification of clones that were also reactive to sera from patients with invasive GAS disease. No clones were identified that reacted solely to sera from patients with invasive disease or GAS-immunized mice. Thus, indicating in both humans and mice the importance of these determinants in invasive disease.

The use of sera to identify immunoreactive antigens expressed during infection has been used in two previous studies on streptococci (29, 49). Lei et al. (29) were able to identify secreted culture supernatant proteins that were reactive with both sera from mice infected subcutaneously with M1 and M3 serotype strains and sera from humans with various diseases of GAS. Similarly, in Streptococcus pneumoniae, Zysk et al. (49) used convalescent-phase patient sera to screen a genomic library of S. pneumoniae for immunoreactive antigens and were able to identify 23 immunogenic pneumococcal proteins, 6 of known and 17 of unknown function. Although both of the aforementioned studies resulted in the identification of antigenic determinants, unlike IVIAT they did not select for genes upregulated in vivo.

Sequence analysis of the reactive clones identified in the present study indicated 16 putative IVI genes, varying from ABC transporters, metabolic enzymes, phage-encoded genes, and genes of unknown function (Table 5). Real-time PCR demonstrated upregulation of three of these genes in vivo, pbp1A, tdcF, and atmB, suggesting that they might be playing a significant role during invasive GAS disease.

Although, penicillin-binding proteins (PBPs) are generally involved in peptidoglycan synthesis and are essential for cell morphology and division (13), a novel function for ponA, which encodes for PBP1A in Streptococcus agalactiae and is a homologue of pbp1A in GAS has been recently identified (22). A ponA-deficient mutant of S. agalactiae, constructed by signature-tagged mutagenesis, was found to be susceptible to phagocytic killing and showed reduced virulence in an animal sepsis model of infection (22). Since, pbp1A in GAS has significant homology with ponA in S. agalactiae and is upregulated in vivo, it could function similarly in the evasion of immune clearance. In fact, our data on the in vivo analysis of insertionally inactivated mutants of pbp1A supports this hypothesis. The pbp1A mutants in two different M serotype backgrounds were cleared at a faster rate than the parent wild-type strains. Furthermore, the rate of lesion formation and healing was significantly reduced in the mutants (Fig. 7). The precise role of pbp1A in invasion will be determined after the construction of in-frame deletion-replacement mutants.

The homologue of tdcF in GAS has not been as well studied as the pbp1A homologue. In fact, almost nothing is known about the homologue of this gene. TdcF was identified as a conserved hypothetical protein in the GAS M1 genome database and shares 60% identity with a putative V. vulnificus translation initiation inhibitor. Although the V. vulnificus gene has not been characterized, a homolog of this protein in E. coli is part of a seven-gene operon that is induced anaerobically (38). In our study, real-time PCR analysis of gene expression shows that tdcF is upregulated in vivo during subcutaneous infection in mice where oxygen is not freely available compared to the in vitro conditions investigated in the present study (refer to Fig. 4 and 5). Thus, if tdcF in GAS is similarly induced during anaerobic conditions, it could function in the O₂-limited environment of the deeper layers of the epidermis (35) encountered by GAS during invasive infections, such as necrotizing fasciitis.

Little is known about the atmB homologue of GAS in Streptococcus mutans. AtmB is a member of the lipoprotein family of proteins which have been shown to perform diverse functions, ranging from substrate-binding proteins in ABC transporter systems to being involved in antibiotic resistance, and cell signaling (43). Identification of lipoproteins that are upregulated in vivo in other pathogenic organisms have been shown to be likely important for pathogenesis. For example, in a previous study of V. vulnificus using IVIAT, a putative lipoprotein was also found to be induced in vivo when convalescent-phase sera from patients who survived V. vulnificus septicemia were used to screen a genomic library of this organism (24). Furthermore, in S. pneumoniae, psaA a metal-binding lipoprotein component of an ABC transporter, which was upregulated in vivo relative to in vitro growth (34), was shown to act as a protective immunogen in mice (44) and psaA-deficient mutants were attenuated in virulence (1). Our preliminary analysis of atmB by insertionally inactivating this gene has shown that it is involved in virulence since the mutants were attenuated in virulence in two different M serotype backgrounds.

Although the function of atmB in GAS is not known, it has been shown to be upregulated in a serotype M1 GAS strain during phagocytic interaction with human polymorphonuclear leukocytes in vitro, thus implicating a possible role for atmB in evasion or defense against the immune system (46). The clearance of the atmB mutants relative to their corresponding parental wild-type strains in the present study does indeed suggest that atmB is playing a role in the evasion or defense against the immune system. We have also shown that atmB is upregulated in vivo, with expression occurring at 24 h postinoculation in the murine model but not at 48 h (Fig. 5), suggesting that atmB is probably important earlier in the infection when GAS must be able to regulate gene expression in response to the host's innate immune response. This evidence points to another advantage of IVIAT unlike other in vivo-induced technologies, since it allows for the identification of genes that are induced throughout the infection process rather than those that are induced at a particular time in the infection process as is observed with IVET (15). Interestingly, in our study we identified atmB as antigenic by screening with both human and mice sera; however, Lei et al. (28) were not able to elicit an immune response by immunizing CD-1 Swiss mice with purified recombinant AtmB. The reason for this is not clear; however, it has been shown that the susceptibility to GAS infection varies depending on the genetic background of mice (31). Therefore, it is possible that the strain of mice used by Lei et al. (28) did not mount an immune response to AtmB.

In addition to the host's genetic background, it appears that the genetic background of the bacterial strain is also playing an important role on the effects of atmB. Under two genetic backgrounds (M49 and M1 serotypes) the degree of attenuation of atmB was more pronounced in the M49 serotype strain (NZ131). The lesions in the mice inoculated with the atmB mutants of NZ131 were smaller and healed faster than those inoculated with the wild-type NZ131 (Fig. 7). However, a significant difference in the lesion size between GAS5448 and its corresponding atmB mutant was only observed at day 2 postinoculation (Fig. 7).

Of the seven genes characterized by real-time PCR, coaA, SPy1674, SPy1784, and SPy1733 were not upregulated in vivo relative to the highest level of expression in vitro (Fig. 6). There are three possible explanations for this result. First, since we measured in vivo gene expression at 24 and 48 h postinoculation, it is possible that we might have missed the time at which there were increased levels of expression of these genes relative to in vitro growth. Second, the patient and mice sera utilized in the present study were adsorbed against in vitro antigens derived from GAS cells grown to a single time point, i.e., late log or stationary phase. Third, the mRNA concentration might not reflect the amount of protein or antigen being produced if the regulation of these antigens is occurring posttranscriptionally. Nevertheless, the possible role of the aforementioned genes in invasive disease cannot be dismissed without further investigation.

coaA encodes a putative pantothenate kinase, which shares 71% identity with a pantothenate kinase gene in S. agalactiae. Pantothenate kinase (ATP-d-pantothenate 4′-phosphotransferase), present in both eukaryotes and prokaryotes, catalyzes the first step in the biosynthetic pathway leading to coenzyme A (CoA), an essential carrier that participates in the metabolism of fatty acids, carbohydrates, and amino acids (36). In Staphylococcus aureus, coaA was shown to be inhibited by a number of compounds and was suggested as a possible drug target for control of resistant isolates since there is little sequence homology between the prokaryotic and eukaryotic CoA biosynthetic enzymes (5). Since coaA encodes an essential enzyme for the synthesis of CoA and the present study has shown that coaA is expressed in vivo, it could offer an interesting and novel target for antibiotic therapy. SPy1733 is a hypothetical protein that shares 69% identity with a transcriptional regulator in S. mutans. A homologue of this gene in Bacillus subtilis was shown to act as an attenuator of the expression of the lytABC and lytR operons involved in autolysis (27). Moreover, mutants of the SPy1733 homologue in S. mutans were shown to be defective in biofilm formation (48). Investigations to study the role of SPy1733 in biofilm formation and invasive disease in GAS are currently under way.

Two putative ABC transporters, SPy1674 and SPy1784, were identified by IVIAT, the former with 74% identity to an ABC transporter in S. agalactiae and the latter with 50% identity to NisF of Lactococcus lactis. ABC transporters comprise a large number of paralogous protein families with diverse functions such as transport of nutrients and exclusion of antibiotics, thus conferring resistance (20). The substrate specificity of both of the ABC transporters identified in the present study is unknown. Nevertheless, a homolog of SPy1784 is nisF in L. lactis, which is a part of an 11-gene operon required for the production of the lantibiotic nisin and is proposed to contribute to immunity against nisin (37, 39). Lantibiotics are antimicrobial peptides produced by gram-positive bacteria that provide the producing strain with a competitive advantage by inhibiting related bacteria (30). SPy1784 might also be part of a lantibiotic producing operon based on its homology to nisF. Although neither SPy1674 nor SPy1784 were upregulated in vivo relative to the in vitro conditions of growth tested in the present study, SPy1674 was shown in another study to be upregulated in response to growth at 29°C relative to 37°C (41). Thus, it appears to respond to environmental temperature stimuli, which might explain why we saw roughly the same level of expression in vivo at 24 and 48 h (Fig. 5).

The present study has shown the successful application of IVIAT in the identification of IVI genes during invasive GAS disease. These genes range from those involved in metabolic activity to those of unknown function and provide us not only with insight into the host environment encountered by GAS but also facilitate characterization of genes with no known function. Additional support validating the successful application of IVIAT comes from the work of research groups that have used this technology to study the pathogenesis of Vibrio cholerae (18), V. vulnificus (24), Actinobacillus actinomycetemcomitans (3), Candida albicans (4), and Mycobacterium tuberculosis (8). We have identified three putative virulence factors (atmB, pbp1A, and tdcF) that are upregulated in vivo. In vivo analysis of insertionally inactivated mutants of atmB and pbp1A have shown the involvement of these genes in virulence, since the mutants were attenuated in virulence. Understanding the mode of action of these genes will contribute significantly to our knowledge of the adaptive mechanisms used by GAS and perhaps provide an explanation for the resurgence of invasive disease.