|Home | About | Journals | Submit | Contact Us | Français|
Staphylococci are the leading causes of endovascular infections worldwide. Commonly, these infections involve the formation of biofilms on the surface of biomaterials. Biofilms are a complex aggregation of bacteria commonly encapsulated by an adhesive exopolysaccharide matrix. In staphylococci, this exopolysaccharide matrix is composed of polysaccharide intercellular adhesin (PIA). PIA is synthesized when the tricarboxylic acid (TCA) cycle is repressed. The inverse correlation between PIA synthesis and TCA cycle activity led us to hypothesize that increasing TCA cycle activity would decrease PIA synthesis and biofilm formation and reduce virulence in a rabbit catheter-induced model of biofilm infection. TCA cycle activity can be induced by preventing staphylococci from exogenously acquiring a TCA cycle-derived amino acid necessary for growth. To determine if TCA cycle induction would decrease PIA synthesis in Staphylococcus aureus, the glutamine permease gene (glnP) was inactivated and TCA cycle activity, PIA accumulation, biofilm forming ability, and virulence in an experimental catheter-induced endovascular biofilm (endocarditis) model were determined. Inactivation of this major glutamine transporter increased TCA cycle activity, transiently decreased PIA synthesis, and significantly reduced in vivo virulence in the endocarditis model in terms of achievable bacterial densities in biofilm-associated cardiac vegetations, kidneys, and spleen. These data confirm the close linkage of TCA cycle activity and virulence factor production and establish that this metabolic linkage can be manipulated to alter infectious outcomes.
The number of patients in whom indwelling biomaterials are implanted annually (e.g., vascular access catheters) is steadily increasing (16). Moreover, the implantation of such biomaterials predisposes an individual to lifelong, device-related infective endocarditis (IE), an extremely serious complication with a high incidence of mortality (15, 36). The most common causative agents of IE are low-G+C gram-positive bacteria, with the predominant bacterial etiological agents being Staphylococcus aureus and coagulase-negative staphylococci (e.g., Staphylococcus epidermidis) (18). These biomaterial-associated infections frequently involve the formation of a bacterial biofilm, featuring an aggregation of bacteria usually encapsulated in an exopolysaccharide matrix (14). The exopolysaccharide provides structural stability to biofilms, enhanced adhesion to surfaces, and protection from host defenses and antibiotics (4, 35, 50). In staphylococci, this exopolysaccharide matrix is composed of polysaccharide intercellular adhesin (PIA) (31).
PIA is an N-acetylglucosamine polymer (31) whose biosynthesis requires the enzymes encoded by the intercellular adhesin operon (icaADBC) (24). Regulation of the ica operon is complex, involving at least two DNA binding proteins (IcaR and SarA) (3, 11, 23, 45, 47), the alternative sigma factor σB, and the tricarboxylic acid (TCA) cycle (32, 49). IcaR is homologous to the TetR family of transcriptional regulatory proteins and is a transcriptional repressor of icaADBC (11). SarA is a DNA binding protein (8) that affects the expression of many genes in S. aureus and S. epidermidis (5). σB affects PIA synthesis by controlling the expression of icaR (27). Inhibition of TCA cycle activity derepresses transcription of the ica operon, linking central metabolism to PIA synthesis (38, 49).
One function of the TCA cycle is to supply intermediates (i.e., oxaloacetate, α-ketoglutarate, and succinyl-coenzyme A) for biosynthesis. For this reason, transcription of genes encoding the first three enzymes of the TCA cycle is regulated by the availability of amino acids (21, 43). That is to say, during exponential growth when amino acids are exogenously available to staphylococci, TCA cycle activity is repressed. Derepression of TCA cycle activity occurs during the post-exponential-growth phase when the availability of exogenous amino acids becomes growth limiting. This amino acid-dependent repression of TCA cycle activity and the reciprocal relationship between TCA cycle activity and PIA biosynthesis led us to hypothesize that hindering exogenous amino acid uptake would increase TCA cycle activity, reducing PIA synthesis, limiting biofilm formation, and attenuating virulence in an in vivo biofilm model, catheter-induced experimental IE. To test this hypothesis, the major glutamine transport gene, the glutamine permease gene (glnP), in S. aureus strain UAMS-1 was inactivated and the effects on TCA cycle activity, PIA accumulation, the ability to form a biofilm, and the capacity to initiate and propagate IE were determined.
Strains and plasmids used in this study are listed in Table Table1.1. Escherichia coli strains were grown in 2× YT broth (39) or on 2× YT agar, and S. aureus strains were grown in tryptic soy broth (TSB) (BD Biosciences) or on TSB-containing 1.5% agar. Unless otherwise stated, all bacterial cultures were inoculated 1:200 from an overnight culture (normalized for growth) into TSB, incubated at 37°C, and aerated at 225 rpm with a flask-to-medium ratio of 10:1. Antibiotics were purchased from Fisher Scientific or Sigma Chemical and, when used, were used at the following concentrations: E. coli, ampicillin at 100 μg/ml; S. aureus, erythromycin at 5 to 8 μg/ml, chloramphenicol at 8 μg/ml, minocycline at 2 μg/ml, and tetracycline at 10 μg/ml.
The amino acid transporter to be inactivated was selected based on three inclusion criteria: (i) the amino acid must be extensively extracted by S. aureus during growth in a biofilm (55), (ii) there must be a complete biosynthetic pathway for the amino acid identified as being extracted, and (iii) the amino acid must be derived from one or more TCA cycle biosynthetic intermediates. These three criteria define an amino acid that is important during growth in a biofilm but that can be synthesized from the coordinated action of the TCA cycle and an amino acid biosynthetic pathway. Previously we determined that Gln, Ser, Pro, Gly, Thr, and Arg were the amino acids most extensively extracted by S. aureus strains UAMS-1 and UAMS-1182 during growth in biofilm flow cells (55). Of these amino acids, Ser and Gly are not synthesized from TCA cycle intermediates; therefore, they do not fulfill our inclusion criteria. Although Thr fulfills all inclusion criteria, it is transported by a multiple-amino-acid transporter; hence, inactivating this transporter would affect the transport of other amino acids. This left the transporters of Gln, Pro, and Arg as possible candidates for genetic inactivation. Mutational studies involving the Pro transporter gene (putP) and the Arg transporter gene (arcD) have been reported (1, 40, 55); hence, inactivating putP or arcD would be unnecessarily duplicative of prior research. In addition, the TCA cycle was not significantly induced in an arcD mutant strain (55) (data not shown), presumably due to the bacteria's ability to generate arginine from ornithine by acquiring exogenous glutamate. These considerations, taken together, led us to select the Gln permease gene (glnP) for inactivation and testing of our central hypothesis.
PCR primers (Table (Table2)2) were designed primarily using the S. aureus COL genome sequence (GenBank accession no. NC_002951). A 1.4-kb PCR product encompassing SACOL1916 (putative glutamine permease gene, glnP) was amplified using primers COL1916-f and COL1916-r and cloned into the SmaI site of pBluescript II KS(+) (Stratagene) to generate plasmid pYF-4. The ermB cassette of pEC4 (6) was amplified using primers ErmBNsiI-f and ErmBBbvI-r and ligated into pYF-4 digested with NsiI and BbvCI (this deletes 576 bp of glnP) to generate the plasmid pYF-5. The glnP::ermB fragment of pYF-5 was inserted into the SmaI site of the temperature-sensitive shuttle vector pTS1 (17) to create pYF-6. The temperature-sensitive plasmid pYF-6 was electroporated into S. aureus strain RN4220 and then introduced into strain UAMS-1 by 85 phage transduction. Strain UAMS-1 containing pYF-6 was used to construct the glnP mutant by the method of Foster (17). To minimize the possibility that any phenotypes were the result of random mutations occurring during the temperature shifts, the resulting glnP::ermB mutation was backcrossed into wild-type strain UAMS-1 using transducing phage 85. Mutants were confirmed by PCR and Southern blotting.
Plasmid pCL15 (Table (Table1)1) containing a Pspac promoter was used to construct the glnP complementation plasmid pYM6. A 1.5-kb promoterless glnP gene from S. aureus strain UAMS-1 was PCR amplified using primers HindIII-SD-glnP-f and EcoRI-glnP-r and ligated into plasmid pCL15, digested with HindIII and EcoRI. The ligation mixture was used to transform S. aureus strain RN4220. After pYM6 was isolated from RN4220, it was electroporated into S. aureus strain UAMS-1-glnP::ermB.
To determine if the glnP gene codes for a glutamine transporter, S. aureus strains were grown overnight in a chemically defined medium lacking glutamate and glutamine (26) and diluted into sterile medium containing different concentrations of the glutamine analogue γ-l-glutamyl hydrazide (Sigma Chemical) to an optical density at 600 nm (OD600) of 0.005. Cultures were grown at 37°C with shaking (225 rpm) for 12 h. Bacterial densities were determined by measuring OD600.
To confirm activation of the TCA cycle, we performed two enzymatic assays querying the oxidative and reductive branches of the TCA cycle. Isocitrate dehydrogenase (oxidative branch) enzymatic activity assays were performed as described previously (43). Fumarase (reductive branch) activity was measured as described previously (25). Protein concentrations were determined by the modified Lowry assay (Pierce Chemical).
TCA cycle activity requires NAD; therefore, an increase in TCA cycle activity can alter the NAD+/NADH redox balance. The intracellular concentrations of NAD+ and NADH were determined using an enzymatic cycling assay kit (Biovision). Total NAD+ and NADH concentrations were normalized to the bacterial density, and determinations of the concentrations were performed in quadruplicate for three independent experiments.
PIA accumulation was determined as described previously (55). The data are presented as the percentage differences in PIA accumulation relative to that by UAMS-1 at 2 h.
Decreasing TCA cycle activity increases the transcription or message stability of the global regulator RNAIII (41, 42), and RNAIII regulates the expression of secreted (e.g., serine protease, encoded by sspA) and cell-associated (e.g., protein A, encoded by spa) virulence determinants. To determine if increasing TCA cycle activity affected RNAIII message levels, Northern blot analysis of transcripts was performed as described previously (39), except that total RNA was isolated using the FastRNA Pro Blue kit (Qbiogene) and purified using an RNeasy kit (Qiagen). Probes for Northern blotting were generated by PCR amplification of unique internal regions of RNAIII, spa, cna, and fnbA (Table (Table2)2) and labeled using the North2South random prime labeling kit (Pierce). Detection was performed using the chemiluminescent nucleic acid detection module (Pierce).
To assess the effect of TCA cycle-mediated RNAIII changes on serine protease gene transcription, a reporter plasmid was constructed. Similarly, to validate changes in PIA biosynthesis, an icaA reporter plasmid was constructed. To construct the sspA reporter plasmid, an 822-bp region of strain UAMS-1 genomic DNA, containing the sspA promoter and the first 189 nucleotides of sspA, was amplified by PCR using primers sspA(SA)-PstI-f and sspA(SA)-BamHI-r (Table (Table2).2). The PCR product was ligated to the xylE gene of pLL38 (7) after digestion of both with PstI and BamHI to generate the reporter plasmid pMRS13. The icaA-xylE transcriptional fusion was made by PCR amplification of a 394-bp region containing the promoter and the first 216 nucleotides of icaA using primers icaA-P1 and icaA-P2 (Table (Table2).2). The resulting PCR product was ligated to the xylE gene in pLL38 at the PstI-BamHI sites to generate pML3783. Reporter plasmids were introduced into strains UAMS-1 and UAMS1-glnP::ermB by phage transduction. Catechol 2,3-dioxygenase activity was assayed as described previously (7). Promoter activity was defined as the absorbance at 375 nm for 1 OD660 unit and expressed relative to the mean activity present in strain UAMS-1 at 2 h postinoculation.
As mentioned, RNAIII regulates protein A expression. To determine if TCA cycle-mediated changes in RNAIII levels affected protein A biosynthesis, protein A was collected as described previously (51) and Western blot analysis was performed as described previously (46).
The primary attachment assay was performed as described by Lim et al. (29). Briefly, bacterial cultures (2 h postinoculation) were diluted into TSB to yield approximately 300 CFU. Bacteria were poured onto polystyrene petri dishes (Fisher Scientific) and incubated at 37°C for 30 min. Following incubation, the petri dishes were rinsed three times with sterile phosphate-buffered saline (pH 7.5) and covered with 15 ml of TSB containing 0.8% agar maintained at 48°C. The percentage of bacteria attached to the polystyrene was defined as the number of CFU remaining in petri dishes after washing compared to the number of CFU in control plates. The experiment was repeated three times.
S. aureus was grown in flow cell chambers (Stovall Life Sciences) as described previously (55). To assess bacterial growth, 12 h postinoculation and every 4 h thereafter, effluent samples were collected and the pH and lactic acid accumulation were measured.
A well-characterized rabbit model of IE (12, 52) was used to study three infection outcomes of S. aureus UAMS-1 (parental strain) versus UAMS-1-glnP::ermB: (i) early bacteremia clearance, (ii) initial vegetation colonization, and (iii) intrinsic virulence. In brief, female New Zealand White rabbits (Irish Farms Products and Services) underwent transcarotid-transaortic valve catheterization (12, 52, 53). Subsequent infection challenge studies are described below.
Animals were challenged by the intravenous (i.v.) injection of S. aureus strain UAMS-1 or UAMS-1-glnP::ermB (109 CFU/animal) at 24 h postcatheterization. At 1 and 30 min postchallenge, blood samples were obtained for quantitative culture. Previous studies with this model suggest that initial clearance of the bloodstream maximally occurs between 30 and 60 min postchallenge (12). In addition, to assess vegetation colonization, animals were sacrificed by a rapid i.v. injection of sodium pentobarbital (200 mg/kg of body weight; Abbott Laboratories) at 30 min postchallenge; all vegetations were removed and processed for quantitative culture on TSB-containing 1.5% agar plates and incubated at 37°C for 24 h. S. aureus densities were expressed as mean log10 CFU per gram of vegetation ± standard deviations.
The intrinsic virulence of IE can be measured as a composite of (i) induction rates of IE over an inoculum challenge range and (ii) target tissue bacterial densities at 48 h after infection at the 95% infective dose inoculum. Thus, animals were challenged at 104 or 105 CFU (the inoculum range encompassing the 95% infective dose for most S. aureus strains in this model ) i.v. with strain UAMS-1 or UAMS-1-glnP::ermB at 24 h postcatheterization. At 48 h after infection, all animals were euthanized, and their cardiac vegetations, kidneys, and spleens were removed and quantitatively cultured as detailed above.
All experiments involving animals were reviewed and approved by the LA Biomedical Research Institute's Institutional Animal Care and Use Committee and comply with Animal Welfare Legislation and NIH guidelines and policies.
The statistical significance of changes between wild-type and mutant strains (e.g., NADH concentrations) was assessed with Student's t test. P values less than 0.05 were considered significant.
The TCA cycle can be induced by inhibiting the ability of staphylococci to acquire a TCA cycle-derived amino acid, thus forcing the bacteria to synthesize that amino acid for growth. To do this, a 576-bp portion of the putative Gln permease gene (glnP, orf SACOL1916) was replaced in S. aureus strain UAMS-1 with an ermB cassette and TCA cycle activity was assessed. Inactivation of glnP did not alter the growth rate, growth yield, or the pH profile of the culture medium (Fig. (Fig.1A),1A), demonstrating the absence of any growth defects in the glnP mutant strain relative to strain UAMS-1. It is possible that phenotypic differences could occur in different media and growth conditions. To confirm that orf SACOL1916 encodes the glutamine permease GlnP, the susceptibilities of strains UAMS-1 and UAMS-1- glnP::ermB to the toxic glutamine analog γ-l-glutamyl hydrazide were determined (Fig. (Fig.1B).1B). As expected, glnP inactivation significantly (P < 0.01) decreased the susceptibility of strain UAMS-1-glnP::ermB to γ-l-glutamyl hydrazide, strongly suggesting that orf SACOL1916 encodes a Gln transporter. Complementation of strain UAMS-1-glnP::ermB with plasmid pYM6 restores susceptibility to γ-l-glutamyl hydrazide equivalent to that of parental strain UAMS-1 (Fig. (Fig.1B).1B). Of note, although the glnP mutant strain is less susceptible to γ-l-glutamyl hydrazide than is the parental strain, it does remain susceptible at high concentrations of γ-l-glutamyl hydrazide, suggesting that there is a lower-affinity transporter capable of transporting glutamine.
As stated, the TCA cycle is repressed during growth in nutrient-rich conditions. The carbon backbone of glutamate is the TCA cycle intermediate α-ketoglutarate; therefore, if Gln transport is hindered, then bacteria will increase synthesis of α-ketoglutarate to offset the decrease in Gln transport. This increase in synthesis of α-ketoglutarate means that there will be an increase in TCA cycle activity at a time when the TCA is normally repressed, which is analogous to the TCA cycle repression during biofilm growth under nutrient-rich conditions. To assess the effect of glnP inactivation on TCA cycle activity under nutrient-rich conditions that are most similar to the nutrient-rich conditions of a biofilm, the exponential- to early-postexponential-growth-phase TCA cycle activity in strains UAMS-1 and UAMS-1-glnP::ermB was assessed by measuring the activities of one enzyme in the oxidative branch (isocitrate dehydrogenase) and one enzyme in the reductive branch (fumarase) and by measuring the intracellular concentrations of NADH (Fig. 1C, D, and F). Consistent with the central hypothesis, glnP inactivation increased TCA cycle activity relative to that of the isogenic parental strain, UAMS-1 (Fig. (Fig.1C),1C), under conditions that are similar to the nutrient-rich conditions encountered in a biofilm flow cell. Complementation of strain UAMS-1-glnP::ermB with a plasmid-borne copy of glnP restored TCA cycle activity to levels consistent with that for the wild-type strain, UAMS-1 (Fig. (Fig.1D).1D). As expected, increased TCA cycle activity was accompanied by a transient increase in the intracellular concentration of NADH (Fig. (Fig.1F).1F). Overall, these data confirm that selective inhibition of amino acid transport increased TCA cycle activity and altered the metabolic status of the bacteria and that this altered metabolic status was due to glnP inactivation.
Our central hypothesis predicts that increased TCA cycle activity will decrease PIA biosynthesis. To determine if increased TCA cycle activity altered transcription of icaADBC and/or synthesis of PIA, the relative amounts of cell-associated PIA produced by strains UAMS-1 and UAMS-1-glnP::ermB were determined using a PIA immunoblot assay, while icaADBC transcription was assessed using a PicaA-xylE reporter plasmid (Fig. 2A to C). As hypothesized, increased TCA cycle activity correlated with decreased icaADBC transcription (Fig. (Fig.2A)2A) and PIA accumulation (Fig. 2B and C); however, these results persisted only while the intracellular NADH concentration in strain UAMS-1-glnP:: ermB was greatest (2 h) (Fig. (Fig.1D).1D). To assess if glnP inactivation would affect biofilm formation, the gross morphology of biofilms was assessed by determining the growth of strains UAMS-1 and UAMS-1-glnP::ermB in biofilm flow cells under nutrient-rich conditions (Fig. (Fig.3).3). These nutrient-rich conditions are similar to those that exist during the exponential growth phase of planktonic cultures (Fig. 1C and D). Inactivation of glnP delayed biofilm maturation; however, the morphologies of the biofilms formed by the wild-type and glnP mutant strains were similar after approximately 40 to 48 h of growth (Fig. (Fig.33).
TCA cycle activity is linked with the synthesis of virulence factors and virulence factor regulators (42, 43), including RNAIII, part of the agr quorum sensing system (37). The transcription and/or stability of RNAIII is increased in an S. aureus TCA cycle mutant (42), raising the possibility that increased TCA cycle activity (Fig. (Fig.1C)1C) could decrease RNAIII transcript levels and alter the temporal pattern of virulence factor synthesis. To determine if increased TCA cycle activity altered RNAIII levels in strain UAMS-1-glnP::ermB relative to those in the parental strain, UAMS-1, Northern blot analysis of RNAIII transcript levels was performed (Fig. (Fig.4A).4A). As expected, increased TCA cycle activity correlated with decreased RNAIII transcript levels relative to those in the parental strain (Fig. (Fig.4A).4A). RNAIII is a negative regulator of adhesin biosynthesis (37) and a positive regulator of secreted virulence determinants (41); hence, glnP inactivation should increase synthesis of adhesins such as protein A (encoded by spa), collagen adhesin (encoded by cna), and fibronectin binding protein A (encoded by fnbA) and decrease transcription of secreted proteins like serine protease (encoded by sspA). Consistent with decreased RNAIII levels, glnP inactivation increased the exponential-phase synthesis of protein A (Fig. (Fig.4B),4B), moderately increased transcription of fnbA and cna (data not shown), and decreased transcription of sspA (Fig. (Fig.4C).4C). These data are consistent with strain UAMS-1-glnP::ermB having an increased ability to adhere to surfaces. To determine if glnP inactivation enhanced bacterial adherence to polystyrene, a polystyrene primary attachment assay was performed with strains UAMS-1 and UAMS-1-glnP::ermB. Levels of primary attachment of strains UAMS-1 and UAMS-1-glnP::ermB to polystyrene were equivalent (data not shown). These data demonstrate that the altered bacterial metabolic status changed the temporal pattern of virulence factor synthesis.
No significant differences in the rates of early bacteremia clearance were observed in rabbits challenged with the parental strain, UAMS-1, or its glnP mutant at either 1 or 30 min postinfection (data not shown). Similarly, no significant differences were observed in the extent of initial colonization of vegetations between strains UAMS-1 and UAMS-1-glnP::ermB (5.60 ± 0.4 and 5.99 ± 0.36 log10 CFU/g vegetation, respectively).
At the 104 and 105 CFU challenge inocula of S. aureus strains UAMS-1 or the isogenic glnP mutant, all catheterized animals developed IE. Importantly, animals individually infected with strain UAMS-1 had significantly higher bacterial densities in all three target tissues than animals infected with strain UAMS-1-glnP::ermB (Table (Table3)3) (P < 0.05).
Inactivation of glnP in strain UAMS-1 increased TCA cycle activity at a time when it is normally repressed (Fig. (Fig.1C).1C). Consistent with our central hypothesis, inactivation of glnP also correlated with transient decreases in PIA synthesis (Fig. (Fig.1C)1C) and RNAIII levels (Fig. (Fig.4A),4A), delayed maturation of biofilms (Fig. (Fig.3),3), and significantly reduced bacterial densities in cardiac vegetations, kidneys, and spleens in an endocarditis model. Precedence for amino acid permease inactivation attenuating S. aureus virulence was shown for the high-affinity proline permease (encoded by putP) (1). Inactivation of putP significantly reduced bacterial densities in cardiac vegetations in an experimental endocarditis model; however, no data are available regarding the effects on virulence factor synthesis. Similar to glnP inactivation, inhibition of arginine transport by deletion of the arginine/ornithine antiporter (encoded by arcD) decreased PIA accumulation; however, deletion of arcD did not affect S. aureus virulence in a catheter-based murine infection model (55). These data demonstrate that S. aureus amino acid transporter inactivation does not produce a uniform effect on virulence, and this may be due to the differing metabolic changes necessitated by the specific transporter inactivated.
The TCA cycle supplies biosynthetic intermediates, reducing potential, and a small amount of ATP. In nutrient-rich growth conditions, the bacterial demand for biosynthetic intermediates is supplied exogenously; hence, TCA cycle activity is very low (10, 43, 44). These same nutrient-rich growth conditions occur during growth in a biofilm flow cell. When environmental conditions change and nutrients become growth limiting, staphylococci increase TCA cycle activity and catabolize nonpreferred carbon sources such as acetate (42). In vitro, the transition from nutrient-rich conditions to nutrient-limited conditions usually occurs concomitant with the transition from exponential growth to postexponential growth. As stated, PIA is synthesized when nutrients are abundant (13); in contrast, TCA cycle activity is repressed under these same conditions (20, 21, 43). In addition to this inverse correlation between PIA synthesis and TCA cycle activity, TCA cycle activity actually represses icaADBC transcription and PIA biosynthesis (38, 49). This causal relationship suggested that derepressing TCA cycle activity would decrease PIA synthesis. Normally, staphylococci derepress TCA cycle activity when nutrients become growth limiting; however, TCA cycle activity can be induced by withdrawing a nutrient from the culture medium or hindering the transport of that nutrient. The current study, using the latter strategy, shows that inactivation of glutamine permease resulted in increased TCA cycle activity relative to that of the parental strain, UAMS-1 (Fig. 1C and D). The increased TCA cycle activity altered the bacterial metabolic status (Fig. (Fig.1F)1F) and the temporal pattern of virulence factor synthesis and transiently decreased PIA synthesis (Fig. 2A to C and 4A to C). These data suggested that manipulating the bacterial metabolic status would decrease virulence. To test this suggestion, we chose to use a rabbit IE model. This model is thought to be the “gold standard” for experimental endovascular infections because it closely mimics its human counterpart microbiologically, immunologically, histopathologically, pathogenetically, and anatomically (in terms of organ involvement) (2).
There are at least four important phases in the pathogenesis of endovascular infections: (i) initial seeding of the bloodstream and survival therein, (ii) colonization of damaged endothelial surfaces, (iii) survival and progression of infection at damaged endothelial sites and subsequent hematogenous dissemination to other target organs, and (iv) reseeding of damaged endothelial sites. The experimental IE model allows one to dissect most of these phases and compare the capacity of parental bacterial strains with those of their relevant isogenic mutants defective in specific phenotypes of interest. Thus, we were able to compare the abilities of the parental UAMS-1 and its glnP mutant strain to execute the above endovascular pathogenetic stages. Since the glnP mutation causes a delay in biofilm formation secondary to TCA cycle activation and PIA suppression in vitro and the catheter-induced IE model above is dependent at least in part on biofilm-related vegetation formation, this provided an ideal experimental milieu. Several interesting observations emerged from these in vivo studies. First, the bacteremic clearance and initial endothelial site colonization phases were not impacted by the glnP mutation. As bloodstream clearance is probably most dependent on susceptibility or resistance to innate host defenses such as opsonophagocytosis and antimicrobial peptide-induced killing (9, 19, 33, 48, 54), this outcome is not unexpected. Moreover, since the colonization phase is related to expression of a cadre of adhesins (9, 52), the noneffect of glnP inactivation on this pathogenetic stage is also not surprising. Of note, in the intrinsic virulence experiments involving parental strain or glnP mutant challenge, a differential fitness advantage of the parental strain versus the glnP mutant within catheter-related cardiac vegetations was observed (Table (Table3).3). With a reduction in glutamine uptake and activation of the TCA cycle, the delay in biofilm formation within vegetations should enable increased access of both cellular and secreted host defenses (e.g., platelet microbicidal proteins ) at this site, translating into enhanced clearance of the glnP mutant from this location. Finally, in the two major hematogenous dissemination target organs in this model (i.e., kidneys and spleens), achievable bacterial densities in animals infected with the parental strain significantly exceeded those in animals infected with the glnP mutant. This outcome undoubtedly reflects the larger bacterial burden in biofilm-associated vegetations of parental-strain-infected animals available for subsequent septic embolization to such target organs.
The close linkage of the TCA cycle and virulence determinant synthesis (42, 43, 49, 55) can be exploited to modify the temporal pattern of virulence factor synthesis (Fig. 2A to C and 4A to C) and attenuate virulence (Table (Table3).3). In addition, increasing TCA cycle activity could potentially increase the in vivo susceptibility of S. aureus to bactericidal antibiotics (28). Taken together, these data suggest that vaccines directed against unique epitopes of bacterial amino acid transporters, such as GlnP, could provide therapeutic benefits beyond facilitating immune system recognition of the infectious agent. Specifically, vaccines directed against amino acid transporters capable of inducing TCA cycle activity would theoretically facilitate protection/treatment by three different means: (i) conventional immune-mediated recognition and killing of bacteria, (ii) attenuation of virulence due to disruption of the temporal pattern of virulence factor synthesis, and (iii) enhancement of killing by bactericidal antibiotics (28).
G.A.S. was supported by funding from the National Institute of General Medical Sciences, the American Heart Association, and funds provided through the Hatch Act. Y.Q.X. was supported in part by the American Heart Association (0465142Y), and A.S.B. was supported by an NIH grant (AI39108).
We declare no conflicts of interest.
Editor: A. Camilli
Published ahead of print on 10 August 2009.