Contribution of Active Iron Uptake to Acinetobacter baumannii Pathogenicity

Acinetobacter baumannii is an important nosocomial pathogen. Mechanisms that allow A. baumannii to cause human infection are still poorly understood.

ABSTRACT

Acinetobacter baumannii is an important nosocomial pathogen. Mechanisms that allow A. baumannii to cause human infection are still poorly understood. Iron is an essential nutrient for bacterial growth in vivo, and the multiplicity of iron uptake systems in A. baumannii suggests that iron acquisition contributes to the ability of A. baumannii to cause infection. In Gram-negative bacteria, active transport of ferrisiderophores and heme relies on the conserved TonB-ExbB-ExbD energy-transducing complex, while active uptake of ferrous iron is mediated by the Feo system. The A. baumannii genome invariably contains three tonB genes (tonB1, tonB2, and tonB3), whose role in iron uptake is poorly understood. Here, we generated A. baumannii mutants with knockout mutations in the feo and/or tonB gene. We report that tonB3 is essential for A. baumannii growth under iron-limiting conditions, whereas tonB1, tonB2, and feoB appear to be dispensable for ferric iron uptake. tonB3 deletion resulted in reduced intracellular iron content despite siderophore overproduction, supporting a key role of TonB3 in iron uptake. In contrast to the case for tonB1 and tonB2, the promoters of tonB3 and feo contain functional Fur boxes and are upregulated in iron-poor media. Both TonB3 and Feo systems are required for growth in complement-free human serum and contribute to resistance to the bactericidal activity of normal human serum, but only TonB3 appears to be essential for virulence in insect and mouse models of infection. Our findings highlight a central role of the TonB3 system for A. baumannii pathogenicity. Hence, TonB3 represents a promising target for novel antibacterial therapies and for the generation of attenuated vaccine strains.

INTRODUCTION

Over the last 20 years Acinetobacter baumannii has emerged as one of the most dreaded opportunistic pathogens in hospitals, being responsible for local and systemic infections, especially in immunocompromised and severely ill patients (1). While the genetic and functional basis of multidrug resistance in A. baumannii clinical isolates is matter of intensive research, the mechanisms of A. baumannii pathogenicity are still poorly understood.

Iron (Fe) is an essential nutrient for all living organisms, since it is required as a cofactor for several enzymes, such as those implicated in electron transport and in amino acid and DNA biosynthesis (2, 3). In aerobic environments, iron exists in the oxidized ferric form [Fe(III)], which aggregates in insoluble oxy-hydroxy polymers. Conversely, in anaerobic and/or reducing environments, the prevalent iron species is the more soluble ferrous form [Fe(II)].

It has been postulated that the ability to acquire iron from the environment contributes to A. baumannii pathobiology and virulence (4–6). Upon entry into the human host, A. baumannii is faced with the low level of free iron imposed by the hypoferremic response and by the presence of high-affinity iron-binding proteins (e.g., transferrin and lactoferrin) (7). To counteract iron starvation, A. baumannii has developed several iron acquisition strategies, such as the production of different siderophores which are variably present in different strains and likely account for Fe(III) scavenging from different sources (8). Production of siderophores is stimulated under iron-limiting conditions and repressed when sufficient iron is present. The Fur (ferric uptake regulator) repressor protein acts as the master regulator of iron homeostasis; in bacteria containing sufficient iron levels, the Fur-Fe(II) complex blocks transcription arising from Fur-controlled promoters, which conversely are transcribed during iron starvation due to detachment of apo-Fur from iron-repressible promoters (9).

In Gram-negative bacteria, Feo is the main system for Fe(II) uptake (10), and it consists of three proteins encoded by the feo operon: FeoA, a small cytosolic protein with still-unknown functions; FeoB, a large protein involved in active translocation of Fe(II) across the cytoplasmic membrane with a cytosolic N-terminal G-protein domain and a C-terminal integral inner membrane domain; and FeoC, a small cytosolic protein likely acting as transcriptional repressor (11).

Bacterial systems involved in Fe(III) acquisition (via either siderophores or heme) require the TonB energy transducing machinery, consisting of the TonB-ExbB-ExbD protein complex (12). This complex transduces the proton motive force (PMF) of the cytoplasmic membrane into energy required for high-affinity active transport of Fe(III)-loaded carriers across outer membrane transporter proteins into the periplasmic space (13). Structurally, TonB consists of a short hydrophobic N-terminal transmembrane domain associated with ExbB and ExbD proteins, a proline-rich linker domain and a C-terminal domain interacting with a variety of the outer membrane transporters (12, 14). Up to 21 putative TonB-dependent outer membrane transporter genes have been identified or predicted in A. baumannii genomes, most often associated with putative or confirmed ferri-siderophore and heme uptake genes (8). TonB-dependent transporter proteins are all characterized by a short conserved signature at the N terminus called TonB box. Once TonB proficiently interacts with the TonB box of an outer membrane transporter, translocation of the transporter-bound ligand into the periplasmic space occurs (14–16).

Although the TonB and Feo systems have extensively been studied in prototypic Gram-negative bacteria, including Escherichia coli and Pseudomonas aeruginosa (16–21), knowledge about these systems in A. baumannii is still limited. Three genes coding for TonB proteins have been identified in the chromosome of the A. baumannii type strain ATCC 19606^T, namely, tonB1, tonB2, and tonB3 (6). The tonB1 and tonB3 genes are components of typical tonB-exbB-exbD operons, while tonB2 is monocistronic (Fig. 1). In a seminal work by Luis Actis’ group, insertional mutagenesis suggested a modest contribution of tonB1 and tonB2 to bacterial growth under low-iron conditions (6). Until now, no data on the role of tonB3 in A. baumannii iron uptake and virulence have been available, mainly due to failure in generating tonB3 knockouts (6).

FIG 1.

Genetic organization of the A. baumannii ATCC 19606^T TonB and Feo systems. The ca. 3.9-Mb genome of A. baumannii ATCC 19606^T contains a feoABC operon and three tonB gene clusters. The tonB1 and tonB3 genes are part of typical tonB-exbB-exbD operons, while tonB2 is a monocistronic element. Nucleotide positioning of these gene loci on the A. baumannii ATCC 19606^T genome is indicated in bold. The three tonB genes and the feoA gene are preceded by putative promoter elements, whose nucleotide positioning on the A. baumannii ATCC 19606^T genome is indicated in italic.

To gain further insight into the contribution of the TonB and Feo system to A. baumannii pathogenicity, we generated mutants with single and multiple mutations in the A. baumannii tonB and/or feo gene and tested them in insect and mammalian models of acute infection. Marked differences in the individual contributions of TonB and Feo systems to iron acquisition by A. baumannii were observed, with TonB3 being crucial for iron acquisition and pathogenicity. These findings encourage future exploitation of TonB3 druggability and pave the way for the generation of attenuated A. baumannii vaccine strains.

RESULTS

TonB3 is essential for A. baumannii growth under iron-limiting conditions.

To assess the contributions of the TonB and Feo systems to iron uptake by A. baumannii ATCC 19606^T, individual markerless tonB1, tonB2, tonB3, or feoB deletion mutants were generated as described in Materials and Methods (22). Multiple mutants lacking Feo and/or TonB were also obtained (Table 1). The wild type and isogenic iron uptake mutants were tested for their ability to grow under different conditions of iron availability, i.e., in M9 minimal medium and in M9 supplemented with either the iron chelator 2,2′-dipyridyl (DIP) or ferric chloride (FeCl₃) (Fig. 2). The growth of the ΔtonB1 and ΔtonB2 mutants did not differ from that of the parent strain, regardless of the test condition (Fig. 2). The isogenic ΔfeoB mutant also showed growth profiles similar to those of the wild type (Fig. 2), suggesting that ferrous iron acquisition is not essential for A. baumannii growth under these conditions. In contrast, the growth of the ΔtonB3 mutant was completely abrogated in both M9 (Fig. 2A) and M9 supplemented with 100 μM DIP (Fig. 2B). All multiple mutants carrying the tonB3 mutation (i.e., the ΔtonB3 ΔfeoB, ΔtonB1 ΔtonB2 ΔtonB3 and ΔtonB1 ΔtonB2 ΔtonB3 ΔfeoB mutants) were unable to grow under iron-limiting conditions (see Fig. S1A and B in the supplemental material). Growth of the tonB3 mutants was rescued by constitutive (Ptac-dependent) expression of tonB3 via the pME6031-derived plasmid pMEtonB3 (Fig. 2A and B) and by the exogenous provision of 100 μM FeCl₃ (Fig. 2C and S1C), even though growth of the tonB3 mutant in the presence of 100 μM FeCl₃ was delayed compared with that of the wild type (Fig. 2C and S1C). The empty vector pME6031, used as a control, did not affect the growth profile of A. baumannii ATCC 19606^T (data not shown). These results demonstrate that the ability of A. baumannii to grow in iron-poor media strictly depends on the TonB3 system, as opposed to the TonB1 and TonB2 systems, which appear to be dispensable under the test conditions.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Reference or source
Strains
A. baumannii
ATCC 19606T	Clinical isolate; type strain	ATCC
ΔfeoB mutant	ATCC 19606T containing a 1,681-bp deletion in the feoB gene	This study
ΔtonB1 mutant	ATCC 19606T containing a 693-bp deletion in the tonB1 gene	This study
ΔtonB2 mutant	ATCC 19606T containing a 585-bp deletion in the tonB2 gene	This study
ΔtonB3 mutant	ATCC 19606T containing a 852-bp deletion in the tonB3 gene	This study
ΔtonB3 ΔfeoB mutant	ΔtonB3 mutant containing a 1,681-bp deletion in the feoB gene	This study
ΔtonB1 ΔtonB2 ΔtonB3 mutant	ATCC 19606T with deletions in all three tonB genes	This study
ΔtonB1 ΔtonB ΔtonB3 ΔfeoB mutant	ATCC 19606T with deletions in all three tonB genes and containing a 1,681-bp deletion in the feoB gene	This study
ATCC 19606T(pME6031)	ATCC 19606T harboring pME6031, empty vector	This study
ΔtonB3(pMEtonB3) mutant	ΔtonB3 harboring pME6031, containing the tonB3 gene	This study
ΔfeoB(pMEfeoAB) mutant	ΔfeoB harboring pME6031, containing the feoAB genes	This study
E. coli
DH5α	recA1 endA1 hsdR17 supE44 thi-1 gyrA96 relA1 Δ(lacZYA-argF)U169 (ϕ80dlacZΔM15) F− Nalr	59
H1717	fhuF::λplacMu aroB araD139 ΔlacU169 rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR	27
Plasmids
pBIISK_sacB/kanR	Suicide vector for allelic replacement; Kmr	22
pME6031	Expression vector for genetic complementation; Ptac ΔlacIq; Tcr	60
pBluescript-II KS	Standard cloning vector; ColE1 replicon; Apr	Stratagene
pMP220	Broad-host-range, low-copy-number promoter-probe vector; Tcr	67
pMEtonB3	pME6031 containing the tonB3 gene (promoter and coding region); Tcr	This study
pMEfeoAB	pME6031 containing the feoAB genes (promoter and coding region); Tcr	This study
pBSPtonB1	pBS containing the tonB1 putative promoter region; Apr	This study
pBSPtonB2	pBS containing the tonB2 putative promoter region; Apr	This study
pBSPtonB3	pBS containing the tonB3 putative promoter region; Apr	This study
pBSPfeoA	pBS containing the feoA putative promoter region; Apr	This study
pBSPbasA	pBS containing the basA promoter region; Apr	26
pMPPtonB1	pMP220 containing the tonB1 putative promoter region; Tcr	This study
pMPPtonB2	pMP220 containing the tonB2 putative promoter region; Tcr	This study
pMPPtonB3	pMP220 containing the tonB3 putative promoter region; Tcr	This study
pMPPfeoA	pMP220 containing the feo putative promoter region; Tcr	This study
pMPPbasA	pMP220 containing the basA promoter region; Tcr	26

^aNal^r, nalidixic acid resistance; Km^r, kanamycin resistance; Tc^r, tetracycline resistance; Ap^r, ampicillin resistance.

FIG 2.

Effect of tonB and feo deletion on A. baumannii growth. Growth curves of A. baumannii ATCC 19606^T and tonB and feo mutants in M9 minimal medium (A), M9 supplemented with 100 μM DIP (B), and M9 supplemented with 100 μM FeCl₃ (C) are shown. Values are the means from three independent experiments ± standard deviation.

Increased siderophore production is a hallmark of intracellular iron deficiency (23), and hence, the ability of the ΔtonB and Δfeo mutants to produce siderophores was assessed using the chrome azurol S (CAS) agar assay (24). We observed increased siderophore production for the single and multiple ΔtonB3 mutants compared with the parental strain, resulting in larger orange halos around the bacterial colonies (Fig. 3A; see Fig. S2A in the supplemental material). In the ΔtonB3 mutant, this phenotype was rescued by complementation with pMEtonB3 (Fig. 3A). Conversely, the tonB1, tonB2, and feoB mutations had no effect on the production of siderophores (Fig. 3A).

FIG 3.

Production of iron-chelating compounds and intracellular iron content of A. baumannii iron uptake mutants. (A) Production of iron-chelating compounds (siderophores) on CAS agar plates, as detectable by the formation of an orange halo surrounding the colonies of the indicated bacterial strains. Images are representative of three independent experiments giving similar results. (B) Amount of intracellular iron (normalized to the total protein content) in the indicated strains grown in M9 supplemented with 1 μM FeCl₃, determined by ICP-OES analysis (see Materials and Methods for experimental details). Values are the means from three independent experiments ± the standard deviation.

To directly correlate growth capabilities and siderophore production with intracellular iron levels, the total iron contents in ATCC 19606^T and the ΔtonB1, ΔtonB2, ΔtonB3, and ΔfeoB mutants was compared by means of inductively coupled plasma optical emission spectrometry (ICP-OES). Due to growth impairment of the tonB3 mutant under low-iron conditions, bacteria were precultured in M9 supplemented with 20 μM FeCl₃, washed twice with saline, diluted in M9 supplemented with 1 μM FeCl₃ (the minimum concentration supporting some growth of the tonB3 mutant), and then incubated at 37°C with shaking for 60 h (to obtain sufficient biomass of the tonB3 mutant for ICP-OES measurements of intracellular iron). In line with the prominent role of TonB3 in iron uptake and despite 1 μM FeCl₃ supplementation, the intracellular iron content was strongly reduced (more than 2-fold) in the ΔtonB3 mutant compared with the parent strain, and this phenotype was partially complemented by the pMEtonB3 plasmid (Fig. 3B). Conversely, the intracellular iron contents were comparable in the ΔtonB1, ΔtonB2, ΔfeoB, and parent strains (Fig. 3B).

Overall, these data demonstrate that TonB3-deficient A. baumannii is unable to grow in an iron-poor medium as a consequence of its inability to acquire iron from the environment, supporting a pivotal role of TonB3 in Fe(III) acquisition.

Iron controls the expression of the tonB3 and feo genes through the global regulator Fur.

Iron uptake systems are usually expressed under iron-limiting conditions. To investigate the expression of the tonB and feoB genes in response to iron concentration, the putative promoter regions of the tonB1 (P_tonB1), tonB3 (P_tonB3), and feo (P_feoA) operons and of the tonB2 gene (P_tonB2) (Fig. 1) were identified by in silico predictions using the BPROM tool of the SoftBerry suite for bacterial promoters (25), and large DNA fragments (310 to 1,067 nucleotides [nt]) encompassing the predicted promoter regions were cloned in the promoter-probe vector pMP220. Promoter activity was measured as β-galactosidase levels expressed by A. baumannii ATCC 19606^T carrying individual promoter fusions under iron-depleted (i.e., M9 plus 100 μM DIP) or iron-replete (i.e., M9 plus 100 μM FeCl₃) conditions (Fig. 4A). As a control, the pMP220 derivative plasmid containing the promoter region of the iron-regulated gene basA was used (26). Similar to the case for P_basA, the activity of both the P_tonB3 and P_feoA predicted promoters was increased by ca. 3-fold under iron-depleted relative to iron-replete conditions. Conversely, no iron regulation was observed for the predicted P_tonB1 and P_tonB2. Of note, β-galactosidase activity for the P_tonB1 putative promoter was barely detectable (<60 Miller units), suggesting that the tonB1 operon is poorly expressed, at least under the test conditions.

FIG 4.

Expression analysis of tonB and feo genes in A. baumannii ATCC 19606^T. (A) β-Galactosidase activity measured in A. baumannii ATCC 19606^T strains carrying the indicated plasmids, grown in M9 supplemented with 100 μM DIP (black bars) or with 100 μM FeCl₃ (gray bars). Mean values from three independent experiments ± standard deviation are shown. (B) Relative mRNA levels of the indicated genes, quantified by real-time PCR in A. baumannii ATCC 19606^T grown in M9 (white bars) or in M9 supplemented with 100 μM DIP (black bars), relative to those in the same strain grown in M9 supplemented with 100 μM FeCl₃. Mean values from three independent experiments ± standard deviation are shown. (C) The P_tonB3 (upper sequence) and P_feoA (lower sequence) putative promoter regions, with the predicted −35 and −10 sequences (underlined), the predicted ribosome-binding site (RBS) (underlined and in bold), the ATG start codon (green and in bold), and the predicted Fur box (red) shown; inverted repeats are indicated with arrows. Prediction of the −10 and −35 hexamers was based on RpoD-dependent E. coli promoters (consensus, TTGACA-N17 ± 1-TATAAT) (66). The tonB3 predicted promoter sequence is TTGcaA-N18-TATtAT, while the feoA predicted promoter sequence is TTGgaA-N16-TATcAT (lowercase denotes differences from the E. coli consensus). (D) FURTA. E. coli H1717 strains carrying the indicated plasmids were grown for 24 h on MacConkey agar supplemented with 100 μM FeSO₄. Images are representative of three independent experiments giving similar results (also see Fig. S3 in the supplemental material).

To corroborate these data, real-time PCR analyses were performed on total RNA extracted from A. baumannii ATCC 19606^T grown in M9 and in M9 supplemented with either 100 μM DIP or 100 μM FeCl₃ (Fig. 4B). The mRNA levels of the tonB3, feoB, and basA genes were significantly increased under iron-limited conditions (i.e., in M9 and M9 plus 100 μM DIP relative to M9 plus 100 μM FeCl₃), thus confirming that the expression of tonB3 and feo genes is upregulated by iron starvation. Conversely, no significant difference was observed for tonB1 and tonB2 expression (Fig. 4B), in accordance with literature data (6). Raw data from the real-time PCR analyses confirmed low levels of tonB1 mRNA (high absolute threshold cycle [C_T] values at the threshold [data not shown]), consistent with low activity of the P_tonB1 promoter (Fig. 4A).

Iron homeostasis in Gram-negative bacteria is controlled primarily by the global transcriptional repressor Fur (9). Our in silico analysis revealed the presence of a putative Fur box in both P_tonB3 (encompassing the predicted −10/−35 sequences) and P_feoA (downstream of the predicted transcriptional start site) (Fig. 4C). Conversely, no Fur boxes could be identified in the P_tonB1 and P_tonB2 putative promoter regions. In vivo binding of Fur to P_tonB3 and P_feoA was investigated by the Fur titration assay (FURTA) (27). To this end, pBluescript (pBS)-derived plasmids containing the P_tonB1, P_tonB2, P_tonB3, and P_feoA promoter regions were generated. As controls, the empty vector pBS and a pBS-derived plasmid containing the Fur-controlled P_basA promoter were used. As expected, E. coli H1717 cells harboring pBSP_basA formed red colonies (Lac⁺) on MacConkey agar supplemented with 100 μM FeSO₄, in accordance with literature data (26), and the same phenotype was observed in the H1717 strains carrying pBSP_tonB3 or pBSP_feoA (Fig. 4D). Conversely, E. coli H1717 containing pBS, pBSP_tonB1, or pBSP_tonB2 generated white (Lac⁻) colonies in the same medium (Fig. 4D). Similar results were obtained for MacConkey agar plates supplemented with lower FeSO₄ concentrations (i.e., 25 and 50 μM; see Fig. S3 in the supplemental material). Indeed, the pBS, pBSP_tonB1, and pBSP_tonB2 constructs turned white at 25 μM FeSO₄, while pBSP_tonB3 and pBSP_feoA remained red up to 100 μM FeSO₄ (Fig. 4D and S3). Therefore, both the tonB3 and feo genes are repressed by the transcriptional regulator Fur when the intracellular iron content is high enough to enable Fur binding to the predicted Fur boxes located within the P_tonB3 and P_feoA promoters.

The tonB3 and feo mutations reduce A. baumannii growth in complement-free HS and increase susceptibility to the bactericidal activity of NHS.

To unravel the importance of iron uptake during A. baumannii infection, A. baumannii ATCC 19606^T and isogenic iron uptake-defective mutants were cultured in heat-inactivated human serum (HS), in which transferrin is expected to exert a bacteriostatic effect due to iron sequestration. HS somehow mimics the medium encountered by bacteria disseminating in biological fluids during a systemic infection. Since A. baumannii grows slowly in HS (26, 28), the bacterial cell density was determined at 48 h postinoculation (Fig. 5A). As expected, the ΔtonB3 strain was unable to grow in HS. Supplementation of HS with 100 μM FeCl₃ was not sufficient to chemically complement the tonB3 mutation, likely due to Fe(III) binding by serum transferrin. However, the cell density of the ΔtonB3 strain was restored to wild-type levels by exogenous provision of a higher FeCl₃ concentration (200 μM) or by plasmid-driven expression of a functional tonB3 gene (the ΔtonB3 pMEtonB3 strain). Likewise, no growth in HS was observed for the ΔtonB3 multiple mutants (i.e., the ΔtonB1 ΔtonB2 ΔtonB3 and ΔtonB1 ΔtonB2 ΔtonB3 ΔfeoB mutants) (see Fig. S4 in the supplemental material).

FIG 5.

Growth of A. baumannii iron uptake mutants in HS and their susceptibility to NHS. (A) The indicated bacterial strains were cultured in HS in the absence of FeCl₃ (white bars) or in the presence of 100 μM FeCl₃ (black bars) or 200 μM FeCl₃ (gray bars). Cell density (OD₆₀₀) was determined after 48 h of incubation at 37°C with shaking. Values are the means from three independent experiments ± the standard deviation. (B) Suspensions of the indicated bacterial strains were exposed to NHS at 37°C for 0 min (white bars), 30 min (black bars), or 120 min (gray bars) before determination of viable cell counts on LB agar plates. Data are the means from three independent experiments (± the standard deviation). Asterisks denote statistically significant differences, measured 30 min after NHS challenge, between ATCC 19606^T and the ΔtonB3 (****, P < 0.0001) or ΔfeoB (**, P < 0.01) mutant. Differences between ATCC 19606^T and the other strains are not statistically significant at 30 min. For all strains, no viable cell was detected after 120 min of incubation in NHS.

Growth of the ΔtonB1 and ΔtonB2 mutants in HS and in HS supplemented with 100 μM FeCl₃ was comparable to that of the parental strain (Fig. 5A), strengthening the evidence that these two systems are not primarily implicated in A. baumannii iron uptake.

In contrast to what was observed in M9 (Fig. 2), feoB deletion caused a 4-fold decrease of A. baumannii cell density in HS compared to that of the parental strain, and this growth defect could be complemented to wild-type levels both chemically (i.e., by addition of 100 μM FeCl₃ to HS) and genetically (i.e., by in trans expression of feoB from the pMEfeoAB plasmid) (Fig. 5A).

These results underscore the importance of TonB3 for iron uptake also in biological fluids and indicate that the Feo system contributes to bacterial proliferation in HS.

Besides iron uptake, features that facilitate the persistence of A. baumannii in the host include the capacities to adhere to biotic and abiotic surfaces, to form biofilm, and to resist to the complement-mediated killing of normal human serum (NHS) (29). Thus, we evaluated these virulence-related traits in TonB and Feo mutants. Briefly, biofilm formation and membrane properties linked to surface adherence (i.e., outer membrane stability and hydrophobicity) were comparable for A. baumannii ATCC 19606^T and all tested mutants (see Fig. S5 in the supplemental material). Conversely, the ΔtonB3 mutant and, to a lesser extent, the ΔfeoB mutant displayed increased susceptibility to the (complement-dependent) bactericidal activity of NHS compared with that of the parent strain and the ΔtonB1 and ΔtonB2 mutants (Fig. 5B). In fact, exposure to 90% NHS for 30 min caused a ca. 3-log reduction in the viability of wild-type A. baumannii ATCC 19606^T and the ΔtonB1 and ΔtonB2 mutants relative to that of preexposed cultures (0 min), while viability of the ΔfeoB mutant was reduced by ca. 4 logs and all the cells of the ΔtonB3 mutant were killed by NHS in 30 min (Fig. 5B). Reversal to wild-type-level susceptibility to NHS was observed for both the ΔfeoB and ΔtonB3 mutants upon complementation with the pMEfeoAB and pMEtonB3 plasmids, respectively (Fig. 5B). Notably, all the tested strains were killed by NHS exposure for 120 min (Fig. 5B).

As a whole, the above findings denote a primary implication of tonB3 in A. baumannii Fe(III) uptake and resistance to complement-dependent killing in human serum.

TonB3 is essential for A. baumannii virulence in insect and mammalian infection models.

It has previously been observed that siderophore systems are involved in A. baumannii virulence (4, 6, 30). Here, the role of iron uptake mutants in A. baumannii virulence was initially screened by using the Galleria mellonella infection model, since it was previously used to investigate the virulence of the ΔtonB1 and ΔtonB2 mutant strains (6). Larvae were infected with ca. 1 × 10⁶ CFU of the parental strain A. baumannii ATCC 19606^T or the derivative iron uptake mutants, and viability was monitored daily for 72 h (Fig. 6A; see Fig. S6 in the supplemental material).

FIG 6.

Impact of the TonB and Feo systems on A. baumannii virulence in animal models. (A) G. mellonella larvae (n = 30) were injected with ca. 1 × 10⁶ CFU of the indicated strains. As a control, larvae were injected with saline or not injected. Larvae were incubated at 37°C and monitored daily for 72 h. ***, P < 0.0001 (log rank test). (B) Groups of five BALB/c mice were inoculated i.p. with ca. 10⁵ CFU of the indicated strains. Their clinical signs and survival were monitored daily for 5 days. The inocula were 0.9 × 10⁵ CFU for ATCC 19606^T, 0.6 × 10⁵ CFU for the ΔfeoB mutant, 0.4 × 10⁵ CFU for the ΔfeoB(pMEfeoAB) mutant, 0.8 × 10⁵ CFU for the ΔtonB3 mutant, and 0.5 × 10⁵ CFU for the ΔtonB3(pMEtonB3) mutant. **, P < 0.001 (log rank test).

About 60% of caterpillars infected with the parental strain ATCC 19606^T were killed 3 days after injection (Fig. 6A). Interestingly, the tonB3 deletion mutant killed only 10% of the injected larvae after 72 h (P < 0.0001), while deletion of tonB1 and tonB2 had no effect on A. baumannii ATCC 19606^T lethality (Fig. 6A). The killing ability of the ΔfeoB mutant was slightly reduced (50% of larvae killed 72 h postinjection) (Fig. 6A) compared with that of the parental strain, but this difference was not statistically significant (P = 0.2256). Only minor, nonsignificant differences were observed in the percentages of larvae killed by the ΔtonB3 single mutant or by multiple mutants with tonB3 deletion (Fig. S6). These results demonstrate that the TonB3 system is strictly required for A. baumannii lethality in G. mellonella, as opposed to the TonB1, TonB2, and Feo systems.

To corroborate the key role of TonB3 in A. baumannii ATCC 19606^T pathogenicity, the virulence of the parental strain and of iron uptake mutants was compared in a mouse model of intraperitoneal (i.p.) infection (31). Preliminary screenings were performed on a small number of mice (1 to 3 per strain) to select for iron uptake systems relevant to A. baumannii virulence in this infection model. It was noticed that only the mice infected with A. baumannii mutants with deletions in the tonB3 gene survived the challenge (see Table S1 in the supplemental material). Although the ΔfeoB strain killed the infected mice in the preliminary experiment, this mutant was further investigated due to its growth defect in HS. Therefore, in a subsequent experiment, groups of 5 mice were inoculated i.p. with ca. 10⁵ CFU of A. baumannii ATCC 19606^T or of the ΔtonB3 or ΔfeoB mutant carrying or not carrying the complementing pMEtonB3 or pMEfeoAB plasmid, respectively. Survival of the infected mice was monitored every 24 h for 5 days. As shown in Fig. 6B, the tonB3 mutation completely abrogated A. baumannii virulence in mice (100% survival), and genetic complementation with pMEtonB3 fully restored lethality. Conversely, all the mice infected with the ΔfeoB or ΔfeoB(pMEfeoAB) strain died within 24 h after the challenge, like the mice infected with the wild-type strain (Fig. 6B). These in vivo data provide conclusive evidence that TonB3 is essential for A. baumannii ATCC 19606^T virulence in different animal models, whereas the TonB1, TonB2, and Feo systems appear to be dispensable.

DISCUSSION

Colonization and infection are strictly dependent on the ability of pathogenic bacteria to acquire iron from their host (7). In turn, mammals respond to the infection by nutritional immunity, i.e., increasing iron sequestration and storage to withhold this essential metal from invading pathogens (32). As a consequence, bacterial pathogens have evolved numerous mechanisms to scavenge iron from host proteins during the infection.

Ferrous iron uptake appears to be more important than ferric iron transport for a number of bacteria, including enteropathogenic E. coli (33), Helicobacter pylori (34), and Clostridium perfringens (35), suggesting that in low-oxygen or low-pH environments (e.g., the intestine or gastric mucosa), Fe(II) uptake could be the preferred pathway for bacterial iron acquisition. This is not the case for the aerobic species A. baumannii, which is found predominantly in oxygen-rich environments. In contrast to Fe(III), Fe(II) is thought to passively diffuse across the porins on the Gram-negative outer membrane (11), thus not requiring a TonB-energy transduction component to reach the periplasm. However, at neutral pH and in the presence of oxygen, Fe(II) is rapidly oxidized to Fe(III), which is likely to be the main iron form encountered by A. baumannii in aerobic environments, suggesting a predominant role of Fe(III) over Fe(II) acquisition in this pathogen. Accordingly, our results demonstrate that feoB deletion does not impair A. baumannii growth in an iron-poor medium (Fig. 2A), even after supplementation with the iron-chelating agent DIP (Fig. 2B). We also observed that knocking out the Feo system neither affects siderophore production (Fig. 3A) nor reduces A. baumannii virulence (Fig. 6). Of note, all in vitro assays described in this study were performed under aerobic conditions in the absence of reducing agents; thus, only a minimal amount of ferrous iron would be available for transport by the Feo system in our experimental settings. Intriguingly, we found that the Feo system was required for full growth of A. baumannii in HS and for resistance to the bactericidal activity of NHS, as also suggested by previous work (36) (Fig. 5). These phenotypes could be attributed to host-derived antimicrobial peptides, since the feoB mutant was previously shown to be hypersensitive to human serum complement and to polymyxin B (which somehow mimics the activity of antimicrobial peptides) compared to the parental strain (36). Considering the significant mortality (34%) associated with A. baumannii bloodstream infection in nosocomial settings (ranking third after Candida spp. and P. aeruginosa) (37), the possible role of the Feo system in ferrous iron acquisition during systemic infection cannot be excluded.

Assays in M9 minimal medium revealed that deletion of either tonB1 or tonB2 does not affect A. baumannii growth under iron-limiting conditions, as opposed to the deletion of tonB3, which completely abrogated growth (Fig. 2; see Fig. S1 in the supplemental material). These data demonstrate the essential role of TonB3 in sustaining A. baumannii growth under conditions of iron starvation, arguing for a functional predominance of the TonB3-dependent over the TonB1- and TonB2-dependent iron uptake system. The key role of TonB3 in Fe(III) uptake is also corroborated by retarded growth of the tonB3-deficient strains even in the presence of 100 μM FeCl₃ (Fig. 2C and S1), although still-unknown functions of TonB3, unrelated to iron uptake, could contribute to this phenotype. Altogether, these findings add novel insights to previously published work, not only confirming the minor role played by the individual TonB1 and TonB2 systems in iron uptake by A. baumannii (6) but also providing an experimental proof of the formerly envisaged prominent role of TonB3 (6).

Besides A. baumannii, many Gram-negative bacteria harbor multiple genes coding for TonB proteins (6, 38–40). TonB proteins are known for providing energy to different high-affinity transport systems, which allow bacteria to acquire several nutrients, such as vitamin B₁₂ (41), ferric siderophores (42), and hemin and heme (43). A number of TonB-dependent transporters have previously been identified in A. baumannii genomes (8), suggesting that the TonB complex could serve as a polyvalent energy coupler for functioning of multiple TonB-dependent transporters and that multiple TonB orthologs may accomplish different functions. Although TonB1 and TonB2 seem to be dispensable for growth under the experimental conditions used in this work, they could be involved in other TonB-dependent membrane-associated processes. Indeed, TonB2 seems to be involved in A. baumannii adhesion to human alveolar epithelial cells (6). Similar results were previously described for the opportunistic human pathogen P. aeruginosa, which carries three tonB homologs. It has been shown that in P. aeruginosa, only tonB1 (the gene orthologous to A. baumannii tonB3) is required for growth under iron limitation (17), while tonB3 is involved mainly in motility and pilus assembly (44).

While outer membrane transporters functionally associated with individual A. baumannii TonB proteins are not well defined at present, it can be speculated that TonB3 can serve as an energy coupler for more than one transporter, given that 20 putative TonB-dependent transporters have been identified in the annotated A. baumannii ATCC 19606^T genome (https://www.ncbi.nlm.nih.gov/genome/proteins/403?genome_assembly_id=165902), many of them being genetically associated with siderophore and heme receptors (8). A plausible candidate for TonB3-dependent transport is the acinetobactin siderophore receptor BauA (45, 46), since an entA mutant impaired in acinetobactin biosynthesis did not grown in iron-poor media (6), thus showing a phenotype similar to that of the tonB3 mutant. Since deletion mutants for the three A. baumannii ATCC 19606^T tonB paralogs are available, functional associations between individual TonB-dependent transporters and TonB systems have now become feasible.

Experimental evidence obtained from FURTA and mRNA quantification indicates that both tonB3 and feo are iron regulated, consistent with the identification in their promoter sequences of putative Fur boxes showing similarity with the A. baumannii Fur consensus (Fig. 4) (47). Being impaired in iron uptake, the tonB3 mutant contains lower intracellular iron levels and overproduces siderophores (Fig. 3) as a compensatory response to iron scarcity and dysregulated Fur function.

No substantial difference in outer membrane stability, hydrophobicity, or biofilm formation was observed between iron uptake mutants (see Fig. S4 in the supplemental material). However, it should be taken into account that in these experimental settings the ΔtonB3 mutant required supplementation of iron to the medium, albeit at a low concentration, to achieve sufficient growth. Notably, both the feoB and tonB3 mutants showed faster and more significant killing by NHS than the wild type and other tonB mutants. Indeed, the tonB3 deletion increased A. baumannii susceptibility to NHS more severely than feoB deletion (Fig. 5B), further supporting the key role played by the TonB3 system in host colonization. Future experiments are required to clarify the molecular mechanisms underlying increased susceptibility of the ΔtonB3 mutant to NHS.

In some bacterial species, tonB mutants exhibit a dramatic attenuation of virulence compared to the parental strains (21, 48–51). In this study, animal experiments revealed that mutants with deletions in the tonB3 gene are less virulent than the wild type, whereas tonB1 and tonB2 mutations do not affect A. baumannii virulence (Fig. 6). The negligible effect of single tonB1 or tonB2 mutation on killing of G. mellonella larvae by A. baumannii is in agreement with literature data (6) and emphasizes the prominent role of TonB3, and hence of iron uptake, in this animal model of infection. G. mellonella is considered a suitable organism for bacterial pathogenicity screening, since this insect model of infection bypass the logistical, ethical, and financial barriers of mammalian models. However, the use of mammals for evaluating the virulence of microbial pathogens provides more complete information on the host-pathogen interactions. In our case, a good correlation exists between virulence in G. mellonella larvae and in mice, corroborating the utility of G. mellonella as a “screening tool” to select a limited number of strains to be subsequently tested in mice.

Many studies have exploited the possibility of developing antibacterial strategies targeting siderophores biosynthesis (52, 53). However, in many pathogens, including A. baumannii, the use of inhibitors of specific siderophore biosynthetic enzymes could be problematic due to the multiplicity of iron acquisition systems (8). This functional redundancy complicates the identification and development of drugs that successfully inhibit bacterial growth by targeting iron uptake. Common players in the iron acquisition machinery, rather than individual iron acquisition systems, are more promising targets for broad-range therapeutic approaches. Indeed, antibacterial compounds targeting iron acquisition (i.e., TonB) have already been explored in uropathogenic E. coli (54), and molecules targeting TonB systems gave encouraging results when tested on A. baumannii (55). As to TonB3 druggability, it is important to underline that tonB3 is among the genes whose expression is upregulated during bacteremia (56) and that no homologs of TonB3 have so far been identified in mammals, hopefully limiting toxicity issues related to the administration of TonB3 inhibitors. Moreover, some studies indicate that avirulent tonB mutants represent suitable backbone strains for future vaccine development against Klebsiella pneumoniae, Burkholderia mallei, and Burkholderia cenocepacia infections (51, 57, 58). Given the increasing antibiotic resistances of A. baumannii and the essentiality of iron for A. baumannii growth in vivo, the data presented in this study highlight TonB3 as a suitable target for the development of new antimicrobial compounds and pave the way to testing tonB3 mutants as attenuated A. baumannii vaccines.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The strains and plasmids used in this study are listed in Table 1. A. baumannii ATCC 19606^T, E. coli DH5α, and E. coli H1717 were grown at 37°C in Luria-Bertani broth (LB) and LB agar (LA) or in M9 minimal medium containing 20 mM sodium succinate as the carbon source (59). Ampicillin (Ap), kanamycin (Km), and tetracycline (Tc) were added when required at the following concentration: for E. coli, 100 μg/ml Ap, 20 μg/ml Km, and 12.5 μg/ml Tc; for A. baumannii, 50 μg/ml Km and 50 μg/ml Tc.

Markerless mutagenesis and genetic complementation.

Markerless A. baumannii iron uptake mutants were generated as previously described (22). Briefly, upstream and downstream DNA regions of feoB, tonB1, tonB2, and tonB3 (ca. 1,500 bp each) were cloned in the suicide vector pBIISK_sacB/kanR. The resulting plasmids were used for allelic exchange in A. baumannii ATCC 19606^T (22). Km-sensitive colonies were verified by PCR using appropriate primer pairs (see Table S2 in the supplemental material). For the tonB3 mutant, growth media were supplemented with 50 μM FeSO₄. Mutants were complemented with the pMEfeoAB and pMEtonB3 plasmids, generated by cloning the feoAB and tonB3 genes under control of the constitutive Ptac promoter in the pME6031 vector (60), respectively. Additional details on the generation of A. baumannii mutant strains and plasmid construction are given in the supplemental material.

FURTA and β-galactosidase activity assay.

The DNA fragments encompassing the putative promoter region of the tonB1-exbB1-exbD1.1-exbD1.2, tonB2, tonB3-exbB3-exbD3, and feoABC operons (Fig. 1) were obtained by PCR amplification with primers (no. 29 to 36) listed in Table S2 and cloned at the EcoRI-BamHI restriction sites of the pBluescript-II KS (pBS) vector to yield pBSP_tonB1, pBSP_tonB2, pBSP_tonB3, and pBSP_feoA (Table 1). These plasmids were introduced into E. coli H1717 to assess the FURTA phenotype, as described previously 27. Briefly, 1-ml bacterial cultures grown overnight at 37°C in LB were washed twice with saline and diluted to obtain ca. 10⁸ cells/ml. Ten microliters of the resulting bacterial suspensions was spotted on MacConkey agar plates supplemented with 100 μM FeSO₄ (Fig. 4D) or with FeSO₄ concentrations ranging from 0 μM to 50 μM (see Fig. S3 in the supplemental material). Plates were incubated at 37°C for 24 h. Putative promoter regions of the tonB1, tonB2, tonB3, feoA, and basA genes were PCR amplified with primers listed in Table S2 and cloned in the pMP220 vector for transcriptional fusions. The resulting plasmids, namely, pMPP_tonB1, pMPP_tonB2, pMPP_tonB3, pMPP_feoA, and pMPP_basA (26), were independently introduced in A. baumannii ATCC 19606^T by electroporation, and transformants were selected on LA plates containing 50 μg/ml Tc. Promoter activities were assessed by measuring β-galactosidase (LacZ) expression levels. For this purpose, A. baumannii ATCC 19606^T carrying the different plasmids was grown at 37°C for 16 h in M9 with 100 μM DIP or 100 μM FeCl₃.

The LacZ activity was determined spectrophotometrically using o-nitrophenyl-β-d-galactopyranoside (ONPG) as the substrate after permeabilization of bacterial cells. Cell permeabilization was as follows: 1-ml bacterial cultures were pelleted by centrifugation and resuspended in 500 μl of sterile saline containing 250 μg/ml lysozyme, and the resulting cell suspensions were incubated for 15 min at room before addition of 30 μl lysis buffer (10% SDS, 0.02 M MnCl₂, toluene, and 2-β-mercaptoethanol) and subsequent incubation for 30 min at 37°C. LacZ activity is expressed as Miller units. Experiments were conducted in triplicate.

Real-time PCR analysis.

Total RNA was extracted from 5-ml cultures of the parental ATCC 19606^T strain grown for 16 h at 37°C in M9 and in M9 supplemented with 100 μM FeCl₃ or 100 μM DIP. Briefly, bacterial cells were pelleted by centrifugation at 4,500 × g for 20 min, and total RNA extraction was performed using the miRNeasy minikit (Qiagen), including the on-column DNase I digestion step. Eluted RNAs were treated for 1 h at 37°C with Turbo DNase (Ambion), following the manufacturer’s instructions. DNase I was removed with the RNeasy column purification kit (Qiagen). The absence of chromosomal DNA was verified by PCR with primers pairs 49 and 50 (Table S2) (61). cDNA synthesis was performed using the iScript Reverse Transcription Supermix for the reverse transcription-quantitative PCR (RT-qPCR) kit (Bio-Rad). Real-time PCRs were performed using the iTaq Universal SYBR Green Supermix (Bio-Rad) and primers 37 to 48 (Table S2). recA was used as the internal control to normalize the real-time PCR data and to calculate the relative fold change in gene expression by using the 2^−ΔΔCT method. The analysis was performed in three technical replicates.

CAS agar assay.

The ability of A. baumannii ATCC 19606^T and isogenic iron uptake mutants to produce iron chelators (siderophores) was investigated using the CAS agar assay (24). Briefly, 1 ml of bacterial culture grown at 37°C in LB was washed with sterile saline and diluted to obtain ca. 10⁸ cells/ml. Ten microliters of this bacterial suspension was spotted on CAS agar plates and incubated for up to 48 h at 37°C. The halo around each spot provided a semiquantitative estimation of the amount of released siderophores.

Measurement of intracellular iron content.

Intracellular iron content was measured according to a procedure described previously (62). Briefly, cells grown in M9 supplemented with 20 μM FeCl₃ were collected by centrifugation, washed twice with saline, diluted in 200 ml of M9 supplemented with 1 μM FeCl₃ to an optical density at 600 nm (OD₆₀₀) of 0.05, and incubated for 60 h at 37°C with shaking at 200 rpm. Cells were then collected by centrifugation, washed with saline, lysed in HNO₃, and analyzed by ICP-OES with an ICP-OES 710 Varian spectrometer (Agilent Technologies). In parallel, total protein content in the same cultures was evaluated by using the Coomassie protein assay reagent (Sigma-Aldrich). Iron levels determined by ICP-OES were normalized to the total protein content of each sample. Results are the means from triplicate experiments.

Assays performed in HS and in NHS.

An existing stock of normal human serum (NHS) pooled from healthy donors (Policlinico Umberto I, Sapienza University of Rome) was used (28). For the growth assays in heat-inactivated human serum (HS), complement was inactivated by incubation at 56°C for 30 min, and the bulk of HS was sterilized by filtration as previously described (28) and then stored at 4°C until used. Growth of A. baumannii in HS was assessed in microtiter plates at 37°C with moderate shaking. Bacteria were grown overnight at 37°C in LB, washed with sterile saline, and then diluted to an OD₆₀₀ of 0.01 in 200 μl of HS supplemented or not with 100 or 200 μM FeCl₃. Growth was measured spectrophotometrically (OD₆₀₀) in a Wallac 1420 Victor^3V multilabel plate reader (Perkin Elmer) at 48 h postinoculation.

Susceptibility to the bactericidal activity of NHS was assessed as previously described (63). Briefly, bacterial cells grown in 3 ml of LB for 16 h at 37°C were diluted in sterile saline to obtain ca. 1 × 10⁷ CFU/ml. Ten microliters of bacterial suspension was added to 90 μl of freshly sampled NHS in 96-well microtiter plates. Viable cell counts were determined at different times by plating 10-fold serial dilutions (10⁰ to 10⁻⁴) on LB agar plates.

In vivo infection assays.

The G. mellonella larva killing assay was performed as previously described (6, 64). Bacterial cells were cultured in LB for 16 h at 37°C with shaking, collected by centrifugation, washed twice, and diluted in saline to ca. 10⁸ CFU/ml. A 1-ml BD Plastipak insulin syringe with a 0.3-mm needle, mounted on a Tridak stepper pipette, was used to inject ca. 1 × 10⁶ bacterial cells (10-μl inoculation) into the hemocoel of each caterpillar through the last left proleg. G. mellonella larvae were incubated at 37°C, and their viability was monitored every 24 h for 3 days. The experiment was performed twice using 15 larvae for experimental groups (n = 30).

Six- to 8-week-old specific-pathogen-free female BALB/c mice were purchased from Charles Rivers Laboratories (St. Constant, QC, Canada). The mice were housed and used in accordance with the recommendations of the Canadian Council on Animal Care Guide to the Care and Use of Experimental Animals. This study and all animal care/use protocols were approved (AUP no. 2012.12) by the Human Health Therapeutics Animal Care Committee (HHT-ACC), National Research Council of Canada.

Fresh inocula were prepared for each experiment from the frozen stocks of A. baumannii ATCC 19606^T and derivative mutants, as previously described (65). Briefly, bacteria were grown overnight on cysteine heart agar (CHA) plates, and a portion was transferred into brain heart infusion broth and incubated at 37°C at 200 rpm for 3 to 4 h until an OD₆₀₀ of 0.85 was reached. Bacterial cells were then centrifuged and suspended in 0.85% saline at 10× the desired inoculation concentration. Immediately before the inoculation, the bacteria were further diluted 1:10 in 5% porcine mucin (Sigma-Aldrich), in order to obtain a final inoculum of 10⁴ to 10⁶ CFU per mouse (31). The inoculum concentration was confirmed by plate counting. Groups of mice were intraperitoneally (i.p.) inoculated with 10⁴ to 10⁶ CFU of different A. baumannii strains in 0.5 ml, and clinical signs and survival were observed and recorded once or twice daily for 5 days.

Statistical analysis.

Statistical analysis was performed with the GraphPad Instat software. For the NHS assay, comparisons between groups were performed using the Student t test. Survival curves for the G. mellonella and mouse killing assays were generated by the Kaplan-Meier method and analyzed by the log rank test. Differences having a P value of ≤0.05 were considered statistically significant.