Involvement of multiple distinct Bordetella receptor proteins in the utilization of iron liberated from transferrin by host catecholamine stress hormones

Summary

Bordetella bronchiseptica is a pathogen that can acquire iron using its native alcaligin siderophore system, but can also use the catechol xenosiderophore enterobactin via the BfeA outer membrane receptor. Transcription of bfeA is positively controlled by a regulator that requires induction by enterobactin. Catecholamine hormones also induce bfeA transcription and B. bronchiseptica can use the catecholamine norepinephrine for growth on transferrin. In this study, B. bronchiseptica was shown to use catecholamines to obtain iron from both transferrin and lactoferrin in the absence of siderophore. In the presence of siderophore, norepinephrine augmented transferrin utilization by B. bronchiseptica, as well as siderophore function in vitro. Genetic analysis identified BfrA, BfrD and BfrE as TonB dependent outer membrane catecholamine receptors. The BfeA enterobactin receptor was found to not be involved directly in catecholamine utilization; however, the BfrA, BfrD and BfrE catecholamine receptors could serve as receptors for enterobactin and its degradation product 2,3-dihydroxybenzoic acid. Thus, there is a functional link between enterobactin-dependent and catecholamine-dependent transferrin utilization. This investigation characterizes a new B. bronchiseptica mechanism for iron uptake from transferrin that uses host stress hormones that not only deliver iron directly to catecholamine receptors, but also potentiate siderophore activity by acting as iron shuttles.

Introduction

The Gram-negative bacterial species Bordetella bronchiseptica, Bordetella pertussis and Bordetella parapertussis are obligate pathogens of mammals (Parkhill et al., 2003; Mattoo and Cherry, 2005). B. bronchiseptica typically causes respiratory infections in nonhuman hosts and occasionally in immunocompromised humans, B. pertussis is the agent of human whooping cough or pertussis, and B. parapertussis causes respiratory disease in humans and sheep. A B. bronchiseptica-like strain is thought to be the ancestor of B. pertussis and B. parapertussis. These bacteria colonize the ciliated cells of the respiratory epithelium where they replicate and produce virulence factors that enable them to persist in the host. Most of the known Bordetella virulence factor genes are transcriptionally regulated by a two-component phosphorelay system consisting of the BvgS transmembrane sensor kinase and the BvgA DNA-binding response regulator (Weiss et al., 1983; Uhl and Miller; 1994).

Most cells have a requirement for iron and have evolved mechanisms to obtain this element, while at the same time regulating intracellular iron concentrations to avoid iron-catalysed oxidative damage. The mammalian host limits the amount of extracellular iron that may become available to microbial pathogens through homeostatic mechanisms such as production of the iron-binding glycoproteins transferrin (TF) and lactoferrin (LF) (Hentze et al.; 2004). For a pathogen, this iron restriction must be overcome for survival and multiplication in the host. Bacterial iron acquisition strategies include the synthesis and utilization of siderophores as well as the production of specific transporters enabling the use of xenosiderophores (Weinberg, 1978; Bullen, 1981; Neilands, 1995; Ratledge and Dover, 2000). In addition, bacteria may have systems allowing the uptake and use of heme iron sources and the direct removal of iron from receptor-bound TF and LF at the bacterial surface (Ratledge and Dover, 2000). Recently, catecholamine neuroendocrine hormones such as epinephrine and norepinephrine (NE) have been reported to remove the iron from TF and LF, and some bacterial species are then able to access this iron (Freestone et al., 2000; 2008; Sandrini et al. 2010), providing additional evidence that pathogens may exploit both microbially-produced and host-derived iron sources. To assimilate various iron sources, Gram-negative bacteria use TonB-dependent outer membrane receptors to transfer ferric iron, heme, or ferric chelates across the outer membrane to the periplasm (Postle and Larsen, 2007). Transport from the periplasm to the cytoplasm usually requires a periplasmic binding protein and cognate ATP binding cassette-type transporter.

B. pertussis, B. parapertussis and B. bronchiseptica produce and utilize the alcaligin siderophore (Moore et al., 1995; Brickman et al., 1996) and B. pertussis and B. bronchiseptica can also use the xenosiderophores enterobactin (Beall and Sanden, 1995), ferrichrome and desferrioxamine B (Beall and Hoenes, 1997). B. bronchiseptica has been reported to use the xenosiderophores ferrichrysin, ferrirubin, aerobactin, protochelin, schizokinen, ferricrocin, vicibactin, and pyoverdin (Pradel and Locht, 2001). These Bordetella species have genes encoding many predicted TonB-dependent iron acquisition systems (Table S1); however only three iron source utilization systems have been characterized to date: those for alcaligin, enterobactin and heme. Under iron-replete growth conditions the genes of these three iron systems are repressed by Fur, and under iron starvation conditions each system is positively regulated transcriptionally by a specific regulator with the cognate iron source acting as the inducer (Brickman et al., 2001; Vanderpool and Armstrong, 2003; Anderson and Armstrong, 2004; Brickman and Armstrong, 2009). Mouse respiratory infection studies demonstrated that B. pertussis mutants lacking any one of the outer membrane receptors for alcaligin, enterobactin or heme were attenuated for in vivo growth (Brickman et al., 2006; 2008; Brickman and Armstrong, 2007). In vivo expression studies determined that B. pertussis differentially regulates transcription of the alcaligin, enterobactin and heme utilization genes during infection, indicating that these iron sources are present in the host and are used at certain stages of infection (Brickman et al., 2008).

Enterobactin is a siderophore of the catechol structural class (Fig. S1) that is produced primarily by members of the Enterobacteriaceae such as Escherichia coli. Bordetellae do not produce enterobactin, but can use it to obtain iron via the BfeA TonB-dependent outer membrane receptor (Beall and Sanden, 1995). The bfeA gene is positively regulated by the AraC-like transcriptional regulator BfeR that requires enterobactin for induction (Anderson and Armstrong, 2004). We determined that other catechol compounds, in addition to enterobactin, were capable of acting as inducers for BfeR. These compounds included the salmochelin and corynebactin siderophores that were less potent inducers than enterobactin, and the NE, epinephrine and dopamine catecholamines (Fig. S1) that induced high levels of bfeA transcription, approximating the induction activity level of enterobactin (Anderson and Armstrong, 2006).

It is not obvious why the respiratory pathogenic Bordetellae have evolved the ability to use the enterobactin siderophore produced by gut bacteria, and it is not known whether enterobactin can be found in the respiratory tract. It was hypothesized that perhaps the natural substrate for the BfeA receptor is not enterobactin, but rather, a host-derived molecule such as a catecholamine. NE, when added to iron-depleted cultures containing serum (Anderson and Armstrong, 2006) or iron-loaded TF (Anderson and Armstrong, 2008), stimulated the growth of B. bronchiseptica. Although NE induces bfeA transcription, efforts to correlate the BfeA receptor with NE utilization or to identify any ferric NE receptor in B. bronchiseptica were not successful (Anderson and Armstrong, 2008). In the present study, we report the identification of three previously uncharacterized B. bronchiseptica outer membrane ferric catecholamine receptors. Furthermore, we present results indicating that BfeA is not a catecholamine receptor, and that catecholamines can act as iron shuttles, augmenting the activities of siderophores.

Results

NE-mediated acquisition of iron from TF and LF

Our previous studies demonstrated that the alcaligin deficient B. bronchiseptica ΔalcA strain BRM26 grew poorly in our iron-depleted defined SS medium culture system; addition of partially iron saturated TF or LF to the medium exacerbated this growth defect, presumably by chelation of residual iron in the system (Anderson and Armstrong, 2008). In that study, we found that B. bronchiseptica could utilize NE to obtain growth-promoting iron from TF by a mechanism that did not require the alcaligin siderophore, but NE-mediated iron acquisition from LF could not be demonstrated. Here, the ability of the B. bronchiseptica ΔalcA strain BRM26 to use NE to retrieve LF iron was reexamined, using apo-LF that we iron-saturated to 70%. In contrast with our 2008 results that used commercially available partially iron-saturated LF, new studies indicate that B. bronchiseptica can use NE to obtain the iron from LF (Fig. 1A). B. pertussis and B. parapertussis however demonstrated little, if any, NE-dependent growth simulation with TF as the iron source (Fig. 1B, Fig. 1C), and no apparent growth promotion by LF (data not shown). Testing of other strains of these two Bordetella species yielded similar results (data not shown). These results indicate that NE-mediated acquisition of both TF-and LF-bound iron in B. bronchiseptica can be siderophore-independent, and that B. pertussis and B. parapertussis are comparatively deficient in NE utilization.

Fig. 1. NE mediated utilization of LF or TF iron.

A. Siderophore-deficient ΔalcA B. bronchiseptica strain BRM26 was grown in iron replete Stainer Scholte (SS) defined medium (+Fe) or in iron-depleted SS medium with human LF (200 μg/ml) (-Fe +LF), or iron-depleted SS medium additionally supplemented with 50 μM NE (-Fe +LF +NE). Growth yields are reported as the means +/- standard deviation from triplicate cultures measured (by optical density at 600 nm) after 48 h of incubation. NE supplementation of LF cultures results in a significant increase in mean growth yield over LF alone (P = 0.0098).

B. Growth yields of B. pertussis strain UT25Sm1. UT25Sm1 was cultured in iron replete SS medium (+Fe) or in iron-depleted SS medium with human TF (200 μg/ml) (-Fe +TF) or with additional 50 μM NE supplementation (-Fe +TF +NE). Mean growth yields after 48 h are shown. Error bars indicate one standard deviation for triplicate cultures. Statistical significance : ns, not significant.

C. Growth yields of B. parapertussis strain ONT-1. ONT-1 was cultured as for B.pertussis in (B). Mean growth yields after 48 h are shown. Error bars indicate one standard deviation for triplicate cultures. Statistical significance : * , significant, P = 0.0368.

NE is a soluble iron chelator

Catecholamines and catechol monomers are known to chelate iron (Green et al., 1956; Jewett et al., 1997), although complex formation is not as effective as that of siderophores such as enterobactin (Harris et al., 1979; Gerard et al., 1999). In our iron depleted culture system, NE is not able to scavenge iron sufficient for Bordetella growth promotion (Anderson and Armstrong, 2008). When growth yields of alcaligin-deficient mutant BRM26 were determined using defined SS medium supplemented with various concentrations of NE and having 30% iron-saturated TF as the sole iron source (Fig. 2), a NE dose-dependent growth response was observed. At 24 h, BRM26 exhibited the best growth in the presence of 100 μM NE, whereas growth at concentrations ranging from 0 – 12.5 μM was minimal. By 48 h, good growth yields were obtained using 50 μM and 100 μM NE, compared to that in iron replete (36 μM Fe) medium, and the lowest NE concentration at which statistically significant growth stimulation could be detected was 6.25 μM. In these studies, growth stimulation was NE-dependent, since the Bordetella strain lacked the ability to produce the alcaligin siderophore and cultures without NE supplementation failed to grow. Iron replete control cultures supplemented with TF had high growth yields comparable to those of iron replete cultures lacking TF (data not shown).

Fig. 2.

NE dose-responsive growth yields of ΔalcA B. bronchiseptica strain BRM26. Cells were grown in iron-depleted SS medium with TF (200 μg/ml) and graded concentrations of NE (-Fe +TF +NE). Iron replete control cultures were grown in SS medium with 36 μM iron (+Fe). Growth yields were measured at 24h and 48h (means +/- standard deviation from triplicate cultures). At 48 h, the mean growth yields of all NE-containing cultures, except the 3.13 μM NE culture (P = 0.6784), were significantly greater than the cultures with no added NE (P ≤ 0.0069).

It was predicted that NE can function as a soluble chelator that removes iron from TF and subsequently diffuses to Bordetella cells for iron uptake and utilization. Early reports noted that TF and LF are able to bind to the Bordetella cell surface but no TF- or LF-specific receptor proteins have been identified (Redhead et al., 1987; Menozzi et al., 1991). To determine whether the bacteria require direct contact with TF for efficient NE-mediated iron uptake, B. bronchiseptica ΔalcA strain BRM26 was cultured in iron-depleted SS medium with TF that was either free in the medium or partitioned within a dialysis bag (8,000 dal molecular mass cutoff) immersed in the culture medium (Fig. 3). Without siderophore or NE supplementation of the culture medium, the bacteria showed no substantial growth regardless of whether the TF was free or sequestered in the dialysis bag (Fig. 3, -Fe samples). As predicted, addition of the alcaligin siderophore to the medium stimulated growth in both the free-TF and sequestered-TF cultures. Importantly, addition of NE alone to both free- and sequestered-TF cultures also stimulated growth, although NE promoted higher levels of growth in the free-TF than the sequestered-TF cultures. When both alcaligin and NE were supplied to TF-sequestered cultures, remarkably high levels of growth stimulation were observed, suggestive of a synergistic effect. Overall, these results showed that even in the absence of a siderophore, NE-mediated TF utilization did not require direct cell contact with TF; thus, NE is capable of functioning as a diffusible iron chelator for Bordetella cells. Although contact of the bacteria with TF was not necessary for NE-stimulated growth, direct contact resulted in higher growth yields. Regardless of whether TF binding to Bordetella cells is specific, interaction of TF with the bacterial surface may facilitate more efficient iron retrieval via siderophores or NE, perhaps simply by virtue of favorable diffusion kinetics associated with TF proximity to the cell.

Fig. 3.

NE-mediated TF utilization does not require cell contact with TF. B. bronchiseptica ΔalcA strain BRM26 was cultured in iron depleted SS medium (-Fe) with 123.7 μM alcaligin (-Fe +Alc), or 50 μM NE (-Fe +NE), or both (-Fe +NE +Alc), with TF either free in the medium (200 μg/ml) or contained within a dialysis bag (same total amount of TF as when free in the medium). Growth yields were measured after 48h. Iron replete control cultures were grown in SS medium containing TF (200 μg/ml) and 36 μM iron (+Fe).

The putative synergistic enhancement of B. bronchiseptica TF utilization by siderophore and NE together suggested that NE may augment siderophore function, or vice versa. A similar amplified B. bronchiseptica growth response was noted in our previous study using TF-containing cultures supplied with NE plus the enterobactin siderophore (Anderson and Armstrong, 2008). The chrome azurol S (CAS) universal siderophore assay was used to determine whether NE had iron-chelating activity similar to that of a siderophore and to examine whether NE or alcaligin had any functional influence on one another (Fig. 4). In the CAS assay (Schwyn and Neilands, 1987), siderophores can remove iron from the blue-colored CAS dye complex, resulting in a blue-to-orange color change that is measured as a decrease in absorbance at 630 nm wavelength. In this functional siderophore assay, the weak chelator, 5-sulfosalicylic acid (SSA), is a shuttle routinely used to speed the iron exchange to equilibrium. Purified alcaligin exhibited characteristic iron-binding activity in this assay and addition of the SSA shuttle increased the initial rate of alcaligin-mediated iron removal, consistent with previous results (Brickman et al., 1996). NE alone was not able to remove the iron from the CAS dye complex; however, when added along with the siderophore, it acted as a shuttle for alcaligin that was much more effective than SSA. These results suggest that although NE may not function as a bona fide siderophore in this assay system, it can augment the iron-binding activity of the alcaligin siderophore. Direct NE-mediated iron removal from TF is likely to be mechanistically different than iron removal from the CAS dye complex. In other experiments, Bordetella alcA mutant cultures supplied with NE did not produce a CAS-active compound, indicating that NE is not being used by Bordetella cells as a precursor for synthesis of another siderophore (data not shown). Taken together, the results suggest that NE can function as a pseudosiderophore to remove iron from TF (and likely LF) and deliver it to B. bronchiseptica directly or it can function as a shuttle to facilitate iron transfer to alcaligin or another utilizable siderophore, whereupon it is taken up by the cognate siderophore transport apparatus.

Fig. 4.

NE shuttles iron to alcaligin. The rate of iron removal from a chrome azurol S (CAS) dye-iron complex was monitored in vitro as a decrease in absorbance at 630 nm wavelength. 5-sulfosalicylic acid (SSA) is used in the standard CAS siderophore assay as a shuttle to increase the rate of iron exchange from the dye reagent to the siderophore. The CAS dye-iron complex reactions contained: no additions (Blank), 60 μM alcaligin (+Alc), 500 μM 5-sulfosalicylic acid shuttle (+SSA), 500 μM NE (+NE), 60 μM alcaligin and 500 μM 5-sulfosalicylic acid (+Alc +SSA), 60 μM alcaligin and 500 μM NE (+Alc +NE).

NE utilization requires TonB

Since the experimental results indicated that NE can act as a pseudosiderophore, it was postulated that a Bordetella TonB-dependent cell surface receptor may be involved in the uptake of the iron itself or of the ferric NE complex. Our previous study examined the ability of a B. bronchiseptica tonB mutant to grow in the presence of NE with TF as the sole iron source (Anderson and Armstrong, 2008). Although a growth defect was observed, it was thought that since the Bordetella strain was capable of alcaligin production, and that alcaligin would be predicted to remove any iron from NE, then the observed defect could be attributed to abrogation of ferric alcaligin uptake which depends on the TonB-dependent receptor protein FauA, rather than an effect on uptake via a putative NE transporter. In the present study, the B. bronchiseptica alcA mutant BRM26 was made tonB. The resulting double mutant, BRM46, was defective in alcaligin production and the ability to use hemin, alcaligin and enterobactin, consistent with a tonB defect (data not shown). BRM46 was tested for the ability to use NE for growth with TF as the iron source (Fig. 5). Whereas BRM26 showed typical NE-mediated growth in the absence of alcaligin production, NE-supplemented BRM46 cultures exhibited poor growth yields similar to those of TF-containing cultures that lacked NE. Genetic complementation of BRM46 with a wild-type copy of tonB restored high-level growth on TF via NE. These results implicated a TonB-dependent outer membrane receptor in NE-mediated TF utilization in Bordetella.

Fig. 5.

TonB is required for NE-mediated TF utilization. B. bronchiseptica strains BRM26 (alcA), BRM46 (alcA tonB) carrying the pRK415 plasmid vector and BRM46 (alcA tonB) carrying the tonB⁺ plasmid pRKton, were cultured in iron-depleted SS medium with TF (200 μg/ml), either with 50 μM NE (-Fe +TF +NE) or without NE (-Fe +TF) supplement. Iron replete control cultures contained 36 μM iron (+Fe). Mean growth yields were determined after 48h. Error bars indicate one standard deviation for triplicate cultures. Statistical significance: ns, not significant; ***, P ≤ 0.0006.

Identification of TonB-dependent outer membrane receptors

Annotation by the Sanger Centre, as well as our own bioinformatic analysis, predicted that B. bronchiseptica has 19 TonB-dependent receptor genes, B. pertussis has 15 and B. parapertussis has 14 (Table S1) (Parkhill et al., 2003; Anderson and Armstrong, 2008). In our previous study, each of the predicted B. bronchiseptica TonB-dependent receptor genes was insertionally inactivated and the resulting mutants tested for their ability to utilize NE to obtain TF-iron (Anderson and Armstrong, 2008). No single mutant was found to be defective. In the present study, given the possibility that the functional alcaligin siderophore system of the mutant strains might mask any NE utilization defects, a new bank of 19 insertion mutants in the BRM26 alcA background was constructed and tested for the ability to use NE to grow with TF as the sole iron source. It was predicted that the bfeA enterobactin receptor mutant would be defective for growth since NE and enterobactin share similar iron-binding catechol moieties and NE induces bfeA transcription. However, all 19 of the receptor mutants were able to grow in the NE-supplemented TF culture system (data not shown), indicating that there were multiple TonB dependent receptors for ferric NE.

Constructing strains with all possible combinations of multiple receptor gene mutations was not a realistic approach to identify the NE receptors. It was reasoned that since B. bronchiseptica had an exalted ability to use NE compared with B. pertussis and B. parapertussis, this difference could be due to receptors encoded by B. bronchiseptica-specific genes that are absent in the other two species. There are four predicted TonB-dependent receptor genes that are apparently unique to B. bronchiseptica: bfrA (BB4761), bfrZ (BB4744), BB1905, and BB0832. Plasmid-borne copies of these B. bronchiseptica genes were transferred by conjugation to B. pertussis strain UT25Sm1. The B. pertussis strain carrying the plasmid vector control demonstrated the typical modest growth yields in SS medium supplemented with TF and NE (Fig. 6). Of the four tested genes, only bfrA conferred the ability to promote NE utilization when carried by B. pertussis (Fig. 6 and data not shown); however, NE-dependent growth yields for the BfrA_Bb⁺ B. pertussis strain by 48 h were not as high as those typically observed for B. bronchiseptica.

Fig. 6.

The B. bronchiseptica bfrA gene promotes NE-mediated TF utilization in B. pertussis. B. pertussis strain UT25Sm1 carrying the pRK415 plasmid vector or the plasmid-borne bfrA gene (pRK/bfrA) was grown in iron-depleted SS medium with TF (200 μg/ml) (-Fe +TF), or medium additionally supplemented with 50 μM NE (-Fe +TF +NE). Iron replete control cultures contained 36 μM iron and grew to an OD₆₀₀ of approximately 4.0 (data not shown). Mean growth yields were determined after 48h. Error bars indicate one standard deviation for triplicate cultures. Statistical significance: ***, P = 0.0005.

With the identification of BfrA as a presumptive NE receptor, and the likelihood that the BfeA enterobactin receptor was also a NE receptor candidate, a B. bronchiseptica bfeA bfrA mutant was constructed in the BRM26 alcA background. Strains BRM65 (alcA bfeA) and BRM66 (alcA bfeA bfrA) were examined for the ability to grow using NE and TF (Fig. 7). As predicted, iron-replete cultures of both strains grew well over the 44 h culture period and cultures lacking NE grew poorly. Compared with the BfrA⁺ strain BRM65, the bfrA mutant strain BRM66 exhibited a diminished growth rate and yield in SS medium containing both NE and TF. Interestingly, despite its decreased growth, the BRM66 alcA bfeA bfrA mutant did not demonstrate a null NE-dependent growth phenotype. This result indicated not only that BfrA was not the sole NE receptor, but that even if BfeA were also involved, there were still additional NE receptors.

Fig. 7.

A B. bronchiseptica mutant lacking BfrA exhibits reduced growth in the presence of TF and NE. Strains BRM65 (alcA bfeA) and BRM66 (alcA bfeA bfrA) were cultured in iron-depleted SS medium with TF (200 μg/ml), either with (-Fe + TF+ NE) or without (-Fe +TF) 50 μM NE supplementation. Iron replete control cultures contained 36 μM iron (+Fe). Growth was measured by optical density over a 44-h time period.

Of the four B. bronchiseptica-specific, putative NE receptor genes tested- bfrA (BB4761), bfrZ (BB4744), BB1905, and BB0832- only the bfrA gene conferred NE utilization upon B. pertussis. This result suggested that B. pertussis has catecholamine receptors in common with B. bronchiseptica but they are insufficiently expressed in the in vitro culture system used. Expressing the cloned B. bronchiseptica bfeRAB enterobactin transport gene cluster in B. pertussis failed to impart enhanced NE utilization (data not shown), suggesting that B. bronchiseptica BfeA is likely not an NE receptor. In depth bioinformatic analyses were repeated on all of the B. bronchiseptica TonB-dependent receptor genes. As noted by Beall and Hoenes (1997), BfrA was found to be most similar to the IrgA and Cir receptors for enterobactin and its catechol derivatives including the biosynthetic intermediate 2,3-dihydroxybenzoic acid (DHBA) and the enterobactin degradation product 2,3-dihydroxybenzoylserine (DHBS) (Table S2) (Hantke, 1990; Mey et al., 2002). The IrgA and Cir receptors are distinct from the well-characterized FepA-type enterobactin receptors (Bäumler et al., 1998; Rabsch et al., 1999). The Bordetella BfrD and BfrE putative receptors showed remarkable similarity to the E. coli Fiu receptor for monomeric catechols such as DHBS (Hantke, 1990). Based on amino acid sequence similarities of predicted Bordetella receptors with known catechol and enterobactin receptors, after BfrA, the top NE receptor candidates were identified as BfrD, BfrE and BfeA.

Combinations of single, double, triple and quadruple mutations were constructed in the BRM26 ΔalcA background (Table S3). The bfrD and bfrE genes are adjacent to one another on the chromosome, allowing both genes to be easily deleted simultaneously. The mutants were tested in NE growth stimulation assays with TF as the iron source (Fig. 8). Similar to the BRM66 alcA bfeA bfrA strain results (Fig. 7), an alcA mutant lacking BfeA, BfrD and BfrE (BRM68) was still able to utilize NE. Subsequent inactivation of bfrA in BRM68, yielding quadruple receptor gene mutant strain BRM69 (alcA bfeA bfrA bfrDE), finally resulted in the elimination of the NE-dependent TF utilization phenotype. To confirm that BfeA, BfrA, BfrD and BfrE were involved in NE utilization, BRM69 was genetically complemented with the individual B. bronchiseptica genes carried on plasmids (data not shown). Individually, the bfrA, bfrD, and bfrE genes were able to restore NE utilization to mutant BRM69. However, the bfeA enterobactin receptor gene did not complement the mutant. Based on this result, a strain that was alcA bfrA bfrDE (BRM71) was constructed and shown to be fully defective in the use of NE (Fig. 9). Since it was possible that BfeA could be a receptor for ferric catecholamines other than NE, BfeA⁺ (BRM71, alcA bfrA bfrDE) and BfeA^- (BRM69, alcA bfeA bfrA bfrDE) strains were compared in growth studies that used TF as the iron source (data not shown). The alcaligin-deficient parent strain BRM26 was able to use epinephrine, dopamine and L-3,4-dihydroxyphenylalanine (L-DOPA) for growth, but the two mutant strains BRM71 and BRM69 could not use these catecholamines regardless of the presence or absence of the BfeA receptor. These experiments identified BfrA, BfrD and BfrE as receptors functioning in catecholamine utilization in B. bronchiseptica and established that the BfeA enterobactin receptor is not involved.

Fig. 8.

Identification of a B. bronchiseptica receptor mutant unable to utilize NE. Strains BRM68 (alcA bfeA bfrDE) and BRM69 (alcA bfeA bfrA bfrDE) were cultured in iron replete SS medium (+Fe) and in iron-depleted SS medium with TF (200 μg/ml) (-Fe +TF), or additionally supplemented with 50 μM NE (-Fe +TF +NE). Growth yields were measured after 48h (mean +/- standard deviation from triplicate cultures).

Fig. 9.

A B. bronchiseptica mutant lacking BfrA, BfrD and BfrE is unable to utilize NE: the BfeA enterobactin receptor is not involved. Parent strain BRM26 and receptor mutant derivatives BRM65, BRM67, BRM70, and BRM71 were cultured in iron replete SS medium (+Fe) and in iron-depleted SS medium with TF (200 μg/ml) (-Fe +TF), or with both TF and 50 μM NE (-Fe +TF +NE). Genotypes are designated for each strain. Growth yields were measured after 48h (mean +/- standard deviation from triplicate cultures). Statistical significance: P values for all strains’ mean growth yields (-Fe + TF + NE), ranged from 0.0001 to 0.0007.

Receptor substrate preference

To determine if there was selectivity for different catecholamines among the BfrA, BfrD and BfrE receptors, the BRM69 quadruple receptor gene mutant strain carrying individual plasmid-borne receptor genes was examined for its ability to utilize NE, epinephrine, dopamine and L-DOPA (Fig. 10). As predicted, the BRM26 parent strain grew in the presence of TF and all of the aforementioned catecholamines, and the BRM69 null mutant carrying the plasmid vector control did not grow. Each of the bfrA, bfrD, and bfrE genes was able to complement BRM69, promoting growth in all of the catecholamine cultures and indicating that each of the three receptors can utilize all four of the catecholamines tested. Epinephrine was significantly less effective (P ≤ 0.0129) than either NE, dopamine, or L-DOPA in mediating TF utilization; NE, dopamine, and L-DOPA yielded comparable growth with TF.

Fig. 10.

The BfrA, BfrD and BfrE receptors are each capable of utilizing NE, epinephrine, dopamine and L-DOPA. Parent strain BRM26 and the null receptor mutant BRM69 (alcA bfeA bfrA bfrDE) carrying plasmids bearing bfrA, bfrE or bfrD were cultured in iron replete SS medium (+Fe) and in iron-depleted SS medium with TF (200 μg/ml) (-Fe +TF), or with both TF and 50 μM NE (-Fe +TF +NE), TF and 50 μM epinephrine (-Fe +TF +Epi), TF and 50 μM dopamine (-Fe +TF +Dop), and TF and 50 μM L-DOPA (-Fe +TF +L- DOPA). Growth yields were measured after 48h (means +/- standard deviation from triplicate cultures). Epinephrine was significantly less effective than either NE, dopamine, or L-DOPA in mediating TF utilization (P ≤ 0.0129)

Since the newly identified receptors share significant similarity with E. coli receptors for enterobactin and other catechols, the growth of Bordetella mutants differentially expressing the receptor genes was examined in catechol-containing cultures. In SS medium containing TF, enterobactin and NE, the BRM26 parent strain grew well, as did a bfeA mutant derivative (BRM65) and a bfrA bfrDE mutant (BRM71) (Fig. 11A). BRM26 growth yield in the presence of both siderophore and NE was enhanced compared with the growth observed in cultures containing only NE (at 36 h, a 32% increase) (Fig. 11B) or enterobactin (at 36 h, a 26% increase) (Fig. 11C), consistent with previous results (Anderson and Armstrong, 2008) and the proposed NE-siderophore synergism. As predicted, in cultures containing TF and NE, only strains producing catecholamine receptors were able to grow (Fig. 11B). In medium containing TF and only enterobactin (i.e., lacking NE), the receptor-proficient BRM26 grew, as did the strains producing any of the known enterobactin or catecholamine receptors, including the bfeA enterobactin receptor mutant BRM65 (Fig. 11C). The BRM69 mutant lacking all known enterobactin and catecholamine receptors showed poor growth in all of the media. All four strains grew poorly in TF-containing medium that lacked NE or enterobactin supplementation (Fig. 11D). Together, these experiments demonstrated that there is reciprocity between enterobactin-dependent and catecholamine-dependent TF utilization: NE augments enterobactin- and BfeA receptor-dependent TF utilization and conversely, enterobactin enhances delivery of TF iron via the catecholamine-dependent uptake system. Furthermore, that a bfeA enterobactin receptor mutant could still grow using enterobactin, but the quadruple receptor mutant could not, suggested a role for BfrA, BfrD and/or BfrE in utilization of enterobactin or its breakdown products.

Fig. 11.

Enterobactin-dependent and catecholamine-dependent TF utilization. Parent strain BRM26 and mutant derivatives BRM65, BRM71 and BRM69 were cultured in iron-depleted SS medium with TF (200 μg/ml) as the sole iron source. Those cultures were further supplemented with both 5 μM enterobactin and 50 μM NE (A), NE alone (B), enterobactin alone (C), or were not supplemented (D). Growth was measured by optical density over a 36-h time period.

In growth yield experiments using cultures supplemented with TF and enterobactin, the BRM65 bfeA enterobactin receptor mutant grew as in the previously described experiments, as did strains lacking 1) bfeA and bfrA (BRM66) and 2) bfeA and bfrDE (BRM68) (Fig. 12A). Similar patterns of growth for these strains were noted when the cultures contained DHBS rather than enterobactin. These results indicate that BfrA, BfrD and/or BfrE can serve as enterobactin and DHBS receptors in B. bronchiseptica. The receptor-proficient BRM26 strain was incapable of utilizing the enterobactin biosynthetic precursor DHBA to obtain iron from TF (data not shown), indicating that either DHBA is not recognized by any of the four catechol receptors or that DHBA is unable to remove the iron from TF in this experimental system.

Fig. 12.

The BfrA, BfrD and BfrE catecholamine receptors function in both enterobactin and DHBS utilization of TF iron. Parent strain BRM26 and mutant derivatives BRM65, BRM66, BRM68 and BRM69 were grown in the SS test media (A). BRM26 and the mutant BRM69 (alcA bfeA bfrA bfrDE) carrying plasmids bearing bfeA, bfrA, bfrE or bfrD were cultured in SS media (B). Culture media: iron replete SS medium (+Fe) and in iron-depleted SS medium with TF (200 μg/ml) (-Fe +TF), or with both TF and 50 μM DHBS (-Fe +TF +DHBS), TF and 5 μM enterobactin (-Fe +TF +Ent). Mean growth yields after 48 h are shown. Error bars indicate one standard deviation for triplicate cultures.

To determine whether enterobactin and DHBS utilization in the BRM66 alcA bfeA bfrA mutant was due to BfrD, BfrE or both receptors, the quadruple receptor gene mutant BRM69 carrying bfeA⁺, bfrA⁺, bfrD⁺ or bfrE⁺ plasmids was assessed for TF-dependent growth mediated by enterobactin or DHBS (Fig. 12B). BRM69 expressing plasmid-borne bfeA, bfrA, bfrD or bfrE genes was able to use enterobactin to obtain growth-promoting iron from TF, compared with BRM69 carrying the plasmid vector control. In the DHBS-containing cultures, all strains grew except BRM69 (vector control) and BRM69 carrying the bfeA⁺ plasmid, indicating that the BfeA enterobactin receptor is incapable of DHBS transport in this culture system and is specific for the tris-catecholate enterobactin. In sum, B. bronchiseptica cells producing BfeA, BfrA, BfrD or BfrE can use enterobactin for growth on ferric TF as the sole iron source. BfrA, BfrD and BfrE, the newly discovered catecholamine receptors, are also able to recognise the DHBS catechol.

Discussion

Lyte made the early observations of the growth-enhancing effects of catecholamines on bacteria (Lyte and Ernst, 1992) and coined the term “microbial endocrinology” to describe the intersection of the fields of microbiology and neurobiology (Lyte, 2004). The association of sepsis and increased susceptibility to infection with catecholamine treatment has been observed clinically in human patients as well as in animal models of infection (Lyte, 2004; Freestone et al., 2008). Catecholamines are produced in the adrenal medulla and concentrations in the pM to nM range can be found in plasma, with higher concentrations at neuronal synapses and in highly innervated tissues such as the gut. To our knowledge, the levels of catecholamines on the respiratory mucosal surface are not precisely known, although it is known that serum exudes from the circulatory system onto the respiratory epithelial surface in both healthy and diseased states (Persson et al., 1991). NE production, excretion and responsiveness have been reported in a variety of immune cell types, and stimulation of neutrophils and alveolar macrophages by lipopolysaccharide induces transcription of catecholamine synthesis genes and the release of NE and epinephrine (Flierl et al., 2008). Our NE titration experiments showed B. bronchiseptica in vitro growth promotion at concentrations as low as 6.25 μM. It is possible that on the respiratory epithelium, Bordetella cells may be exposed to nutritionally relevant catecholamine concentrations in the local microenvironment. Therefore, respiratory pathogens such as B. bronchiseptica may be exposed to catecholamines during infection, either on the mucosal surface via serum exudation or through interaction with immune cells.

Catecholamines have been reported not only to possess in vitro growth-promoting activity for bacteria, but also to influence bacterial virulence. For example, pretreatment of chicks with NE renders them more susceptible to colonization by Salmonella enterica (Methner et al., 2008) and NE was found to increase adherence of E. coli 0157:H7 to bovine intestinal mucosa (Vlisidou et al., 2004) and to colonic mucosae of pigs (Green et al., 2004). NE and epinephrine also activate the quorum sensing autoinducer-3 QseBC system of enterohemorrhagic E. coli that regulates the expression of certain virulence genes (Clarke et al. 2006). Both growth and expression of type III secretion system-1 genes in Vibrio parahaemolyticus are enhanced by NE (Nakano et al., 2007) and in Borrelia burgdorferi, expression of the membrane protein OspA, important for tick colonization, is increased in the presence of NE or epinephrine (Scheckelhoff et al. 2007).

Catecholamines bind iron in vitro (Green et al., 1956; Jewett et al., 1997; Gerard et al., 1999) and they participate in feedback inhibition of their own biosynthesis by binding a ferric iron cofactor of tyrosine hydroxylase (Andersson et al., 1988). Bidentate catechols similar to NE can remove iron from TF and transfer it to the siderophore desferrioxamine (Devanur et al, 2008), and NE was demonstrated to remove the iron from TF and LF (Freestone et al., 2000). Previous studies demonstrated a role for catecholamines such as NE for growth of E. coli (Burton et al., 2002; Freestone et al., 2003) and Salmonella enteriditis (Methner et al., 2008) in serum-containing medium. However, use of NE by these organisms required enterobactin, DHBA or DHBS. It is possible that in these studies, NE acted as a shuttle to augment siderophore function, consistent with our results and a hypothesis proferred by Freestone and coworkers (Freestone et al., 2003). Recently, NE was demonstrated to form a complex with TF or LF and remove the iron by a reductive mechanism (Sandrini et al., 2010). The study also reported that E. coli mutants defective in either TonB or enterobactin biosynthesis showed decreased NE-mediated iron uptake compared with the wild-type strain. In contrast, the present investigation demonstrated that a siderophore is not required by Bordetella cells to utilize NE to obtain iron from TF. Thus, for B. bronchiseptica, NE acts as a soluble iron chelator, in a manner that is functionally similar to that of a siderophore. Listeria monocytogenes, which produces no known siderophores, was reported to utilize catecholamines to obtain iron from a weak chemical chelator, tropolone; the mechanism appeared to involve iron reduction at the cell surface (Coulanges et al.,1997). Campylobacter jejuni also does not produce siderophores, but exhibits CfrA enterobactin receptor-dependent growth using NE on agar containing desferrioxamine as an iron chelator (Zeng et al., 2009). A recent study described Staphylococcus aureus siderophore mutants that could use catecholamines to obtain growth-promoting iron from TF, and showed that the Sst cytoplasmic membrane transporter for catechol siderophores was involved in this process (Beasley et al., 2011). Our report describes Bordetella outer membrane receptors for catecholamines; however the presumed required cytoplasmic membrane transport proteins remain unknown and no predicted candidates are encoded near the receptor genes bfrA, bfrD or bfrE.

Under iron starvation growth conditions, Bordetella bfeA transcription and production of the BfeA enterobactin receptor is increased in the presence of enterobactin (Anderson and Armstrong, 2004; 2006). This induction response, requiring the transcriptional regulator BfeR, can also be stimulated by catecholamines. Based on these induction studies and the structural similarity between catecholamines and enterobactin, it was hypothesized that BfeA may also function in iron uptake mediated by NE and other catecholamines. Surprisingly, the present analysis demonstrated that the BfeA receptor is not involved in ferric catecholamine utilization, although catecholamines clearly augment enterobactin-mediated TF utilization (Anderson & Armstrong, 2008; this study). That catecholamines induce transcription of the Bordetella enterobactin receptor gene may simply reflect the structural similarity of these catechol compounds to the enterobactin inducer, acting via the positive regulator BfeR, or may indicate that catecholamines serve as host cues to induce bfeA transcription. Since the B. pertussis BfeA receptor was found to be important for growth in a mouse model of respiratory infection (Brickman et al., 2008), in vivo induction of bfeA transcription by stress hormones may be advantageous to the pathogen.

The present study identified BfrA, BfrD and BfrE as catecholamine receptors. bfrA was first reported as a B. bronchiseptica iron-repressed gene encoding a putative TonB-dependent outer membrane receptor that is absent in B. pertussis and B. parapertussis (Beall and Hoenes, 1997). That study noted that a bfrA mutant was still able to use alcaligin, enterobactin or DHBS as iron sources. BfrA is most similar to outer membrane catechol receptors of the IrgA group and to the E. coli Cir receptor for monomeric catechols such as DHBS and DHBA (Hantke, 1990). In Vibrio cholerae, IrgA is a receptor for enterobactin and possibly its intermediates or breakdown products (Mey et al., 2002). The C. jejuni CfrA enterobactin receptor is another member of the IrgA catechol receptor group (Zeng et al., 2009), as is the Iha receptor, found in strains of uropathogenic E. coli (Leveille et al., 2006). The Bordetella BfrD and BfrE catecholamine receptors are most similar to the E. coli Fiu receptor for DHBS and DHBA (Hantke, 1990). BfrD is an outer membrane protein produced by virulent Bvg⁺ phenotypic phase, but not Bvg^- phase, B. pertussis strains (Passerini de Rossi et al., 1999; Antoine et al., 2000). The adjacent bfrE gene was identified as transcriptionally Bvg independent, and encoding a predicted receptor protein with 56.7% identity to BfrD (Antoine et al., 2000). A subsequent study verified that BfrD production was Bvg activated and modestly iron repressed, although it was noted that bfrD does not have a well-conserved predicted Fur repressor binding site (Passerini de Rossi et al., 2003). Consistent with these observations, subsequent B. pertussis and B. bronchiseptica microarray analyses did not identify bfrD or bfrE as iron-repressed genes (Brickman et al., 2011) but confirmed that bfrD is a member of the BvgAS regulon (Cummings et al., 2006). Similarity searches of sequences from Achromobacter xylosoxidans A8 and Achromobacter piechaudii, which are taxonomically related to Bordetella species, revealed each Achromobacter species to have a single gene (AXYL01473; ZP_06686349, respectively) encoding a predicted receptor with highest similarity to BfrE (80% identity), rather than BfrD (57% identity). Bordetella petrii also has a single gene (Bpet1692) that encodes a protein 80% identical to BfrE and 56% identical to BfrD. No Bordetella avium receptor gene was identified that specified a protein with high-scoring similarity to either BfrD or BfrE. Based on the presence of a single bfrE - like gene in Achromobacter species and in B. petrii, it is likely that the bfrE gene was shared among these species as well as the progenitor of the mammalian-pathogenic Bordetella species, whereupon the gene was duplicated in the pathogen ancestor and evolved to become the Bvg-regulated bfrD gene. Since both B. pertussis and B. parapertussis have an intact bfrD gene and B. pertussis has bfrE (B. parapertussis bfrE is an annotated pseudogene), it is unknown why these species utilize catecholamines relatively poorly in comparison to B. bronchiseptica. It is possible that they lack the requisite cytoplasmic membrane transport machinery to internalize ferric catecholamines. It is also conceivable that the bfrD genes of B. pertussis and B. parapertussis, as well as B. pertussis bfrE, are not sufficiently expressed so as to confer effective catecholamine utilization, at least in the experimental system used in this study.

Members of the Enterobacteriaceae have multiple TonB-dependent receptors for the uptake of ferric catechols; these include FepA (enterobactin, myxochelin C, DHBS), IroN (salmochelin, enterobactin, bacillibactin, DHBS), Iha (enterobactin, DHBS), and the Cir and Fiu receptors that transport ferric catechols such as DHBS. For example, a single Salmonella strain can produce several of these receptors (Rabsch et al., 1999), suggesting that a diverse catechol receptor repertoire may be needed for successful growth in a host. Many bacterial species that do not synthesize enterobactin have evolved receptors to use it as well as its degradation products (Thulasiraman et al., 1998). Bordetella cells use enterobactin via BfeA, which exhibits highest similarity to FepA and our present analysis identified three receptors for ferric catecholamines. Thus the presence of four catechol receptors in the obligate pathogen B. bronchiseptica suggests that catechols, either derived from the resident flora or provided by the host, are present in the respiratory tract. Furthermore, the co-regulation of bfrD with other Bvg-controlled virulence genes suggests that this receptor is important in vivo.

LF is the principal iron-binding glycoprotein on host mucosal surfaces, and TF is also present on the respiratory tract mucosae of humans (Stites et al., 1995). Bordetella toxins, such as the tracheal cytotoxin, are known to kill host cells and damage the respiratory epithelium (Flak and Goldman, 1999). It is reasonable to hypothesize that during infection, Bordetella cells have access to heme, as well as LF and TF, catecholamine hormones, and any xenosiderophores produced by microbial flora. Our studies revealed that NE has strong iron shuttling activity and can potentiate the activity of siderophores in the presence of ferric TF. In the host, Bordetella cells may initially use existing xenosiderophores to obtain iron from LF and TF until they can synthesize effective concentrations of alcaligin. In this environment, host catecholamines may function as shuttles that enhance ferric iron uptake from any utilizable siderophores that are available. In the absence of siderophores, catecholamines would be predicted to remove the iron from TF and LF and provide it directly to B. bronchiseptica via any or all of the newly identified receptors.

Experimental Procedures

Bacterial strains and culture conditions

Bordetella strains used in this study are listed in Table S3. E. coli DH5α (Invitrogen, Carlsbad, CA) was used in routine cloning and in conjugations. Luria Bertani broth and agar (Sambrook et al., 1989) were used for the cultivation of E. coli. Bordetella strains were grown on Bordet Gengou agar (Bordet and Gengou, 1906) containing 20% (B. pertussis, B. parapertussis) or 7.5% (B. bronchiseptica) sheep blood, or Luria Bertani agar (B. bronchiseptica). Modified Stainer-Scholte (SS) medium (Schneider and Parker, 1982; Stainer and Scholte, 1970) was used as a chemically defined medium for the cultivation of Bordetella strains. SS basal medium was rendered iron-depleted by treatment with Chelex 100 (BioRad, Richmond, CA) and iron-replete by supplementation with 36 μM FeSO₄ (Armstrong and Clements, 1993). The siderophores alcaligin (Brickman et al., 1996) and enterobactin (Anderson and Armstrong, 2004) were purified as described previously. Monomeric 2,3-dihydroxybenzoylserine (DHBS) was from EMC Microcollections (Tuebingen, Germany) and 2,3-dihydroxybenzoic acid (DHBA) was obtained from Sigma-Aldrich (St. Louis, MO). Stock solutions of norepinephrine bitartrate salt, epinephrine, dopamine and L-3,4-dihydroxyphenylalanine (L-DOPA) (Sigma-Aldrich) were made fresh and added to SS cultures at the specified concentrations. The purified apo- forms of human TF and LF were purchased from Sigma-Aldrich and loaded with ferric iron as described (Agiato and Dyer, 1992). Bordetella growth in SS medium was monitored by optical density using a spectrophotometer (at 600 nm wavelength) or a Klett-Summerson colorimeter. Media were supplemented with the following concentrations of antibiotics as appropriate: ampicillin, 100 μg/ml; chloramphenicol, 30 μg/ml; gentamicin, 10 μg/ml; kanamycin, 50 μg/ml; nalidixic acid, 35 μg/ml; tetracycline, 15 μg/ml.

Siderophore assay

Siderophore activity was determined using the chrome azurol S (CAS) assay (Schwyn and Neilands, 1987). Reaction mixtures contained various combinations of 60 μM alcaligin, 500 μM 5-sulfosalicylic acid (SSA) shuttle and 500 μM norepinephrine with the CAS dye reagent. The absorbance of the reaction mixtures at 630 nm wavelength was monitored over time. Results typical of two independent experiments are shown.

Growth stimulation assays

Catecholamine-mediated growth stimulation by TF (30% iron saturated) or LF (70% iron saturated) was measured as growth in liquid SS cultures. Where noted, cultures contained alcaligin (123.7 μM), enterobactin (5 μM), DHBS (50 μM) or DHBA (50 μM). B. bronchiseptica strains were grown on agar for 24 h, then subcultured into iron-replete SS medium. After growth of this seed culture for 18 hours at 35°C in a shaking incubator (250 rpm), the cells were harvested and washed with SS basal medium lacking iron. The cells were then subcultured to an initial optical density (600nm) of 0.01 to iron-replete medium or iron-depleted medium containing 10 mM sodium bicarbonate and 200 μg/ml TF or 200 μg/ml LF (approximately 2.5 μM TF or LF), with or without 50 μM catecholamine (NE, epinephrine, dopamine or L-DOPA). For Fig. 3 experiments, the 30% iron saturated TF (1 ml of a 3 mg/ml solution) was placed within a sterile dialysis bag (8,000 molecular mass cutoff) that was then submerged in 15 ml of the iron-depleted SS medium. For B. pertussis strain UT25Sm1 and B. parapertussis strain ONT1, seed cultures were initiated using inocula from agar plates, and grown in iron replete SS for 24 hours, shaking, at 35°C. The cells were harvested, washed, and subcultured to an initial O.D.₆₀₀ of 0.02 to iron-replete or iron-depleted medium containing TF or LF, with or without catecholamine, as described above for B. bronchiseptica. The optical densities of the Bordetella cultures were measured 24 and 48 hours post inoculation. The growth yield results reported are the means of triplicate cultures and each experiment was performed at least twice independently. Statistical analysis of growth yield data used Student's t-test (paired, two-tailed, hypothesized difference = 0), with StatView version 4.51 software (Abacus Concepts, Inc.). The dialysis membrane and growth curve experiments are representative of at least two experimental trials.

General genetic methods

Standard methods were used for the construction of recombinant plasmids (Sambrook et al., 1989). Plasmid DNA was transferred to E. coli strains by electroporation and to Bordetella strains by conjugation using E. coli DH5α as the donor strain, with mobilization functions provided by plasmid pRK2013 (Figurski and Helinski, 1979). Broad host range plasmid pRK415 (Keen et al., 1988) was the cloning vector used in Bordetella strains. Plasmids pGEM3Z, pGEM7Z (Promega, Madison, WI) and pBluescript (Stratagene/Agilent Technologies, Santa Clara, CA) were used as general cloning vectors in E. coli. Construction of B. bronchiseptica mutants by allelic exchange employed suicide plasmids pEG7 (Cotter and Miller, 1997) or pRE112 (Edwards et al.,1998), and construction of B. bronchiseptica mutants by plasmid integration used pEG7 or pSS1129 (Stibitz, 1994). Bordetella transconjugants were selected on agar containing the appropriate antibiotics and colicin B as described previously (Brickman and Armstrong, 1996a). Oligonucleotide primers (Table S4) were purchased from Integrated DNA Technologies (Coralville, IA). For expression of B. bronchiseptica genes in B. pertussis, the sequences of the B. bronchiseptica bfeRAB genes and those encoding the predicted TonB-dependent receptors BfrA, BfrZ and BB0832 and BB1905 were amplified from strain RB50 by polymerase chain reaction (PCR), subcloned to pRK415 and the resulting plasmids conjugated to B. pertussis strain UT25Sm1. B. bronchiseptica genes tonB, bfrA and bfrE were amplified by PCR, cloned to pRK415, and used in mutant complementation studies. bfrD was expressed from the 238 bp bfrE promoter region (primers pbfrE1 and pbfrE2) cloned upstream of the coding region. The Bordetella genome sequences (Genbank accession numbers: B. pertussis Tohama I, NC_002929; B. parapertussis NC_002928; B. bronchiseptica RB50, NC_002927) were accessed at the GeneDB website (http://www.genedb.org/), developed and maintained by the Sanger Institute's Pathogen Sequencing Unit. BLASTP analyses were conducted using the server at the National Center for Biotechnology Information (http://blast.ncbi.nlm.nih.gov/Blast.cgi). DNA and protein analyses and PCR primer design used the Lasergene software package version 5.53 (DNASTAR, Inc., Madison, WI).

Mutant construction

Insertion mutants

B. bronchiseptica mutants defective in each known or predicted TonB-dependent receptor gene were constructed using plasmids described previously (Brickman and Armstrong, 1999; Anderson and Armstrong, 2008). Suicide plasmids were introduced to the alcA mutant strains BRM1 (Armstrong and Clements, 1993) or BRM26 by conjugation, and integrants were selected based on gentamicin resistance. Integration sites were confirmed by PCR analysis.

tonB mutant

An in-frame 627-bp tonB deletion was made by PCR amplification of a region comprising 1022-bp upstream of tonB (BB1980) with 89-bp of the 5′ end of the gene from RB50 chromosomal DNA and cloning this fragment into pGEM3Z as an SstI-BamHI fragment, yielding plasmid p3Z164. A PCR product comprising of 1095-bp downstream of tonB with 91-bp at the 3' end of the gene was cloned into p3Z164 as a BamHI-XbaI fragment to make plasmid p3Z165. The resulting ΔtonB allele was subcloned to pRE112 as an SstI-XbaI fragment. The plasmid was delivered to B. bronchiseptica alcA mutant BRM26 and following allelic exchange, mutant BRM46 was isolated and confirmed to have the tonB mutation by PCR.

bfeA mutant

Construction of the allelic exchange plasmid carrying bfeA with a 582-bp in-frame deletion mutation has been described (Anderson and Armstrong, 2008). The plasmid was transferred to B. bronchiseptica BRM26 and allelic exchange mutant BRM65 was isolated and confirmed to have the bfeA mutation by PCR.

bfrA mutant

A region containing bfrA along with flanking sequence on either side was PCR amplified and cloned into pBluescript II KS+, yielding plasmid pKS+/bfrA1#4. An in-frame 810-bp bfrA deletion was made by NcoI restriction digestion and religation of plasmid pKS+/bfrA1#4. The resulting ΔbfrA allele was subcloned to pEG7 as a BamHI-HindIII fragment along with the sacBR cassette as a BamHI fragment. The plasmid was delivered to B. bronchiseptica strains and following allelic exchange, mutants were isolated and their bfrA mutations confirmed by PCR.

bfrDE mutant

The bfrD and bfrE genes are adjacent on the chromosome and a single plasmid construct was designed to mutate both genes simultaneously. A region consisting of the first 76 bp of the bfrD gene and extending 1681 bp upstream was PCR amplified and cloned into pGEM7Z as an XbaI-HindIII fragment, resulting in plasmid p7Z5. A PCR product comprising 1160-bp downstream of bfrE and 34-bp of the 3′ end of the gene was cloned into p7Z5 as a HindIII-SstI fragment, yielding p7Z6. The resulting ΔbfrDE fragment possessed a deletion mutation spanning both genes, removing 4393 bp from the coding sequences. The ΔbfrDE fragment was subcloned to pBluescript II KS+ as an SstI-PstI fragment then to the allelic exchange vector pEG7 as an EcoRI fragment along with the sacBR cassette from pEG18.3 as a BamHI fragment. The plasmid was conjugated to B. bronchiseptica strains and putative mutants were isolated and their bfrDE mutations confirmed by PCR.