Legionella pneumophila Rhizoferrin Promotes Bacterial Biofilm Formation and Growth within Amoebae and Macrophages

ABSTRACT

Previously, we showed that Legionella pneumophila secretes rhizoferrin, a polycarboxylate siderophore that promotes bacterial growth in iron-deplete media and the murine lung. Yet, past studies failed to identify a role for the rhizoferrin biosynthetic gene (lbtA) in L. pneumophila infection of host cells, suggesting the siderophore’s importance was solely linked to extracellular survival. To test the possibility that rhizoferrin’s relevance to intracellular infection was missed due to functional redundancy with the ferrous iron transport (FeoB) pathway, we characterized a new mutant lacking both lbtA and feoB. This mutant was highly impaired for growth on bacteriological media that were only modestly depleted of iron, confirming that rhizoferrin-mediated ferric iron uptake and FeoB-mediated ferrous iron uptake are critical for iron acquisition. The lbtA feoB mutant, but not its lbtA-containing complement, was also highly defective for biofilm formation on plastic surfaces, demonstrating a new role for the L. pneumophila siderophore in extracellular survival. Finally, the lbtA feoB mutant, but not its complement containing lbtA, proved to be greatly impaired for growth in Acanthamoeba castellanii, Vermamoeba vermiformis, and human U937 cell macrophages, revealing that rhizoferrin does promote intracellular infection by L. pneumophila. Moreover, the application of purified rhizoferrin triggered cytokine production from the U937 cells. Rhizoferrin-associated genes were fully conserved across the many sequenced strains of L. pneumophila examined but were variably present among strains from the other species of Legionella. Outside of Legionella, the closest match to the L. pneumophila rhizoferrin genes was in Aquicella siphonis, another facultative intracellular parasite of amoebae.

INTRODUCTION

Legionella pneumophila is an aquatic bacterium that is best known as the agent of Legionnaires’ disease (LD) pneumonia (1, 2). In both natural and engineered water systems, Gram-negative L. pneumophila incorporates into biofilms and infects a wide range of amoebae, most notably species of Acanthamoeba and Vermamoeba (3–7). Besides providing a replicative niche, residence in amoebae affords L. pneumophila protection from biocides thereby enhancing the chance of the pathogen spreading to humans (8–12). After entrance into the lungs via the inhalation of contaminated water droplets (13, 14), L. pneumophila grows in macrophages, utilizing an infection pathway that mirrors the one operating within amoebae (15, 16). In both protozoan and mammalian host cells, L. pneumophila replicates within a membrane-bound phagosome (i.e., the Legionella-containing vacuole [LCV]), making use of a myriad of secreted proteins encoded by the Dot/Icm type IV secretion system and the Lsp type II secretion system (17–24). The incidence of LD has recently increased ~5-fold (25–27), with L. pneumophila remaining especially pathogenic for immunocompromised and elderly individuals (28, 29). Lately, L. pneumophila infection is also manifest in patients with COVID-19 (30–32). Thus, discerning the bacterial processes involved in L. pneumophila biofilm formation and infection of amoebae and macrophages remains an important endeavor.

Iron levels are an important parameter in plumbing systems associated with cases of LD, and the acquisition of iron is critical for the virulence of L. pneumophila (33, 34). In addition to utilizing hemin as an iron source, the Legionella bacterium uptakes both ferrous iron and ferric iron, with the assimilated metal serving as a cofactor for various important enzymes (33, 35). Like many other bacteria, L. pneumophila regulates its iron transport processes through the action of Fur, a transcriptional repressor of iron-transport genes when iron levels are sufficiently elevated (36–38). For ferrous iron acquisition, L. pneumophila utilizes the inner membrane transporter FeoB, and we have shown that mutants inactivated for feoB are impaired for growth in low-iron bacteriological media, intracellular infection of amoebae (Vermamoeba vermiformis, formerly Hartmannella vermiformis) and human macrophages (differentiated U937 cells), and virulence in a murine model of pneumonia (39). Some of the ferrous iron that feeds into the FeoB pathway is generated by the ferric reductase activity of the L. pneumophila melanin-like pigment (40, 41). For ferric iron uptake, L. pneumophila secretes the polycarboxylate siderophore rhizoferrin, which had been originally denoted as legiobactin (42, 43). Rhizoferrin is made when L. pneumophila strains are grown in a deferrated chemically defined medium (CDM) and is detected by either the Chrome Azurol S assay, which detects iron-chelating activity, or a Legionella-based bioassay (43–45). In L. pneumophila, the synthesis of rhizoferrin is mediated by the siderophore synthetase LbtA, which generates the siderophore by conjoining two citrates with putrescine (43, 44, 46). Export of the siderophore is mediated by the inner membrane protein LbtB, whereas the import of ferri-rhizoferrin is facilitated by the outer membrane receptor LbtU and the inner membrane transporter LbtC (43, 44, 47, 48). Based on our phenotypic analysis of a lbtA mutant, rhizoferrin is required for the optimal growth of L. pneumophila in low-iron media and in the lungs of A/J mice (47–49). However, even though lbtA is expressed by intracellular L. pneumophila, lbtA mutants were not found to be impaired for infection of either V. vermiformis, Acanthamoeba castellanii amoebae, human U937 cell macrophages, or explanted murine macrophages (47, 49, 50). These data implied that rhizoferrin is not relevant for intracellular growth by L. pneumophila and that the siderophore’s role during lung infection and environmental survival is strictly associated with extracellular bacteria (34).

Here, we considered an alternative hypothesis regarding the involvement of rhizoferrin in intracellular infection by L. pneumophila, i.e., the siderophore is complementary to/functionally redundant with the bacterium’s ferrous iron transport pathway. To test this hypothesis, we constructed a mutant lacking both lbtA and feoB, overcoming technical limitations that had precluded the prior isolation of such a mutant (50). The lbtA feoB mutant, unlike the lbtA mutant, proved to be greatly impaired for growth in multiple amoebal species and human macrophage cells, with its infectivity defect easily exceeding that of a feoB mutant. These findings, especially when combined with additional new data linking rhizoferrin to both biofilm formation and the elicitation of cytokines, demonstrate that siderophore production is of multifactorial importance for L. pneumophila in environmental and mammalian habitats.

RESULTS

Simultaneous loss of rhizoferrin and FeoB abolishes L. pneumophila growth in low-iron bacteriological media.

A newly made lbtA feoB mutant (NU460) of L. pneumophila strain 130b exhibited a typical efficiency of plating on buffered charcoal yeast extract (BCYE) agar (Fig. 1A, left), the solid medium that is traditionally used to grow L. pneumophila (51). Indeed, the double mutant produced numbers of CFU that were like that of the parental wild type (WT) strain 130b, a lbtA mutant, and a feoB mutant. However, when strains were plated onto BCYE agar that lacked its usual iron supplement component (i.e., 330 μM ferric pyrophosphate) (51), the lbtA feoB mutant failed to form any colonies (Fig. 1A, right). In line with prior assays (39), the feoB mutant was sensitive to this level of reduction in iron availability (Fig. 1A, right). Nonetheless, the growth defect of the lbtA feoB mutant easily exceeded that of the feoB mutant by at least 1,000-fold. As alluded to above, the plating efficiency of an lbtA mutant is sensitive to iron limitation but that was only observed upon addition of exogenous iron chelators (e.g., deferoxamine) to the low-iron BCYE agar (44, 47, 48). Reintroduction of an intact copy of lbtA into the lbtA feoB mutant greatly improved the strain’s growth (Fig. 1A, right), confirming that the double mutant’s defect seen here was due in part to the loss of rhizoferrin. In agreement with these observations, a second, independently derived lbtA feoB mutant (NU461) showed the same inability to form CFU on BCYE agar lacking the iron supplement (data not shown). When we tested the strains for growth in buffered yeast extract (BYE) broth that had the customary ferric pyrophosphate supplement, all grew similarly (Fig. 1B, left). However, the lbtA feoB mutant, unlike WT, as well as the lbtA mutant and the feoB mutant, failed to grow when inoculated in BYE broth that lacked the added iron (Fig. 1B, right). Utilizing inductively coupled plasma mass spectrometry (ICP-MS), we determined that the level of iron in the base BYE broth was equal to 7.2 μM. Taken together, these data revealed the critical importance of having both FeoB and LbtA for L. pneumophila growth in lowered iron conditions. The fact that the double mutant was more impaired than either the feoB mutant or the lbtA mutant further affirmed that FeoB and LbtA operate on different iron acquisition pathways, namely, ferrous iron transport versus ferric iron uptake via rhizoferrin.

FIG 1.

Effect of lbtA and feoB mutations on L. pneumophila growth in bacteriological media with differing amounts of iron. (A) Following 3 d of growth on BCYE agar at 37°C, WT strain 130b (WT), lbtA mutant NU302 (lbtA), feoB mutant NU458 (feoB), lbtA feoB mutant NU460 (lbtA feoB), and NU460 containing a plasmid-carried lbtA (lbtA feoB/plbtA) were suspended in BYE broth to an OD₆₆₀ of 0.3, and then 10-μL aliquots taken from a series of 10-fold dilutions were spotted onto either BCYE agar that did (left) or did not (right) contain its standard supplement of ferric pyrophosphate. Following incubation for 5 days at 37°C, images were taken of the areas of bacterial growth. The differences in the efficiency of plating seen in the right panel were also evident when examining bacterial growth at 2, 3, or 4 days. (B) Bacterial strains, as indicated above, were suspended in BYE broth that did (left) or did not (right) contain the standard supplement of ferric pyrophosphate, and then aliquots were placed into the wells of a 96-well microtiter plate. Upon incubation at 37°C, bacterial growth was monitored by OD₆₆₀ readings taken every hour over the next 30 h. Data presented are the means and standard deviations from six technical replicates. In the medium lacking added iron supplement, the lbtA feoB mutant grew differently from the other strains at t = 3 h and beyond, ****, P < 0.0001. For panels A and B, the presented data are representative of the results from >3 independent experiments.

Rhizoferrin promotes L. pneumophila biofilm formation.

To begin to learn the importance of rhizoferrin along with ferrous iron transport in biofilm formation by L. pneumophila, we tested our panel of mutants for the ability to form a monospecies biofilm on a plastic surface. As done in many past studies with L. pneumophila and other biofilm-forming bacteria (52–59), we assessed biofilm formation using crystal violet staining. Since L. pneumophila may form biofilms more robustly when incubated at temperatures below 37°C (53), we did our assays at both 30°C and 37°C. The lbtA mutant achieved a WT level of biofilm at 30°C and 37°C (Fig. 2A), indicating that rhizoferrin production is not required for biofilms to form when L. pneumophila is incubated in standard BYE broth (i.e., containing a 330 μM ferric pyrophosphate supplement) under standard assay conditions. When incubated at 30°C but not at 37°C, the feoB mutant exhibited approximately one-half the level of biofilm that WT had (Fig. 2A), affirming that ferrous transport is required for optimal biofilm formation under some conditions (60). Most notably, the lbtA feoB mutant was even further impaired, i.e., at 30°C, it showed a level of crystal violet staining on the plastic surface that was like that of the no-bacteria control, and at 37°C, its biofilm was less than that of the feoB mutant (Fig. 2A). The greater defect shown by the double mutant was reversed upon reintroduction of an intact lbtA gene (Fig. 2A), suggesting that LbtA can also promote the ability of L. pneumophila to optimally form biofilms. When we quantified the biofilm biomass by staining with safranin (61), similar results were obtained (see Fig. S1 in the supplemental material), confirming that the difference between WT and mutant that we had initially observed was not an artifact of differential crystal violet binding. Since the lbtA feoB mutant grew normally in the iron-supplemented BYE medium that was used in this assay, based on CFU recovery (data not shown) and optical density (OD) measurements (Fig. 1B), the reduced level of biofilm observed here was also not simply due to a loss in numbers of mutant bacteria upon inoculation into the plastic wells. When the amount of supplemental iron in the assay medium was increased 2-fold (i.e., 660 μM ferric pyrophosphate), the lbtA feoB mutant gave a level of biofilm that was comparable to that of the feoB mutant (Fig. 2B), affirming that the reduced biofilm formation first observed was likely linked to an impairment in iron acquisition. Thus, these data indicated that the combination of rhizoferrin and ferrous iron transport is required for L. pneumophila to maximally form a biofilm even under conditions that would appear to be iron-replete and rich in other nutrients. In some other systems, biofilm-associated phenotypes can also be visualized by spotting an aliquot of a bacterial suspension unto an agar plate and then monitoring the pattern of outgrowth over the agar surface (62–65). When we performed a similar assay with our strains, the lbtA mutant and feoB mutant showed a central area of robust growth surrounded by a zone of spreading (Fig. 2C). Arguably, the extent of spreading seen for these two mutants was more pronounced than that of the parental WT strain. However, the lbtA feoB mutant displayed a pattern that was remarkably different from that of the other strains in both the central area and the region of spreading (Fig. 2C), further suggesting that a strain lacking both rhizoferrin and ferrous iron transport is impaired for biofilm formation. As before, the reintroduction of an intact copy of lbtA into the lbtA feoB mutant reversed its aberrant phenotype (Fig. 2C), confirming that this mutant’s defect was due in part to the loss of rhizoferrin.

FIG 2.

Effect of lbtA and feoB mutations on L. pneumophila biofilm formation. (A) Following 3 d of growth on BCYE agar at 37°C, WT strain 130b (WT), lbtA mutant NU302 (lbtA), feoB mutant NU458 (feoB), lbtA feoB mutant NU460 (lbtA feoB), and NU460 containing a plasmid-carried lbtA (lbtA feoB/plbtA) were resuspended in standard BYE broth to an OD₆₆₀ of 0.2, and then aliquots of the suspensions were added into the wells of a 96-well, plastic microtiter plate. After 2 d of incubation at either 30°C (left panel) or 37°C (right panel), the amount of the biofilm formed by the bacteria was determined by staining with crystal violet as read at 600 nm. Control wells containing only the medium and no added bacteria were also included in the analysis. Data are presented as means with standard deviations of results from three independent experiments, with six technical replicates obtained in each experiment. (B) The indicated bacterial strains were assayed for biofilm formation at 30°C as described in panel A, with the medium in the wells being either BYE broth that has the standard iron supplement (i.e., 330 μM ferric pyrophosphate) added or BYE broth that has twice the standard iron supplement (330 μM ferric pyrophosphate) added. Data presented are the means and standard deviations from six technical replicates and are representative of three independent experiments. In panels A and B, asterisks indicate differences in the levels of biofilm formation between strains or between conditions: *, P < 0.05; **, P < 0.01; ****, P < 0.0001. (C) Bacterial strains, as indicated above, were suspended in BYE broth to an OD₆₆₀ of 0.2, and then 20-μL aliquots were spotted onto the surface of BCYE agar. Following incubation for 5 d at 37°C, images were taken of the areas of bacterial growth. The differences in growth patterns seen here were also evident when examining the bacteria at days 3 and 4. The presented data are representative of the results from three independent experiments.

Rhizoferrin potentiates L. pneumophila intracellular infection of amoebae.

To begin a reexamination of the importance of rhizoferrin in intracellular infection by L. pneumophila, we compared WT, lbtA mutant, feoB mutant, and lbtA feoB mutant for infection of A. castellanii amoebae. In agreement with prior analysis (47), the lbtA mutant behaved as WT did in standard cocultures (Fig. 3A). The feoB mutant also infected A. castellanii similarly to WT (Fig. 3A), akin to what it did before during coculture with V. vermiformis amoebae (39). In marked contrast, the lbtA feoB mutant barely increased its CFU over the 72-h course of the experiment, revealing ~1,000-fold reduced recovery compared to WT and the other two mutants (Fig. 3A). This phenotype was not seen for the lbtA feoB mutant complemented with an intact copy of lbtA (Fig. 3A), indicating, for the first time, that LbtA and its rhizoferrin product promote at least one form of intracellular infection by L. pneumophila. When we assayed specifically for bacterial entry into the amoebae, neither the lbtA feoB mutant nor the lbtA mutant or feoB mutant showed a defect (Fig. 3B), implying that the importance of LbtA and rhizoferrin occurs at a later stage of intracellular infection. When we infected A. castellanii monolayers with green fluorescent protein (GFP)-expressing L. pneumophila and specifically monitored replication within the intracellular compartment, the lbtA feoB mutant alone failed to show an increase in fluorescence during a primary round of infection (Fig. 3C). The inability of the lbtA feoB mutant to replicate was also evident when m-Cherry was utilized as the fluorescence reporter (Fig. S2A). Finally, similar results were obtained upon infection of V. vermiformis (Fig. 3D and Fig. S2B), another type of amoeba that is a key host for L. pneumophila in aquatic habitats (66, 67). Together, these data documented that rhizoferrin promotes L. pneumophila infection of multiple amoebae and does so during the replicative phase of the intracellular cycle. Given that the importance of rhizoferrin for intracellular infection was revealed by the characterization of a lbtA feoB mutant (being markedly more impaired than a feoB mutant), the role of the siderophore for L. pneumophila growth in amoebae quite likely includes ferric iron acquisition.

FIG 3.

Effect of lbtA and feoB mutations on L. pneumophila intracellular infection of amoebae. (A) A. castellanii were infected with either wild-type strain 130b (WT), lbtA mutant NU302 (lbtA), feoB mutant NU458 (feoB), lbtA feoB mutant NU460 (lbtA feoB), or NU460 containing plasmid-carried lbtA (lbtA feoB/plbtA). At 0, 24, 48, and 72 h postinoculation, net bacterial growth (number of CFU at the denoted time/number of CFU at t = 0) was determined by plating aliquots of the supernatant on BCYE agar. Since L. pneumophila does not grow in the assay medium, any increases in CFU are the result of growth in the acanthamoebae. (B) A. castellanii monolayers were inoculated with GFP-expressing WT or mutant bacteria, as noted above. Following a 1-h incubation that allowed for bacterial entry into the host cells, a trypan blue solution was added to quench the GFP signal from remaining extracellular bacteria. Entry was reported as the percentage of trypan blue resistant GFP signal, which corresponds to internalized L. pneumophila. For panels A and B, the data are presented as means with standard deviations of results from four (A) or three (B) independent experiments, with four (A) or six (B) technical replicates in each experiment. (C and D) Monolayers of A. castellanii (C) or V. vermiformis (D) were infected with the GFP-expressing WT and mutant bacteria indicated above. After centrifugation and a 1-h incubation to facilitate uptake, gentamicin was added to kill remaining extracellular legionellae. The fluorescence from the intracellular bacteria was then monitored kinetically every 30 min for the next 21 h, and the fluorescence values obtained were normalized to the GFP signal at t = 0 following antibiotic treatment. Since this assay only monitors intracellular bacteria, the GFP signals for some of the strains level off (C) or begin to decline (D) due to the end of the primary round of intracellular growth, lysis of the spent amoebae, and release of legionellae into the culture supernatant. This is in contrast to the assay depicted in panel A, which detects ever-increasing numbers of some strains due to multiple rounds of intracellular growth and lysis of more and more spent amoebae. In panels C and D, the values presented are the means and standard deviations obtained from six technical replicates and are representative of the results obtained from four (C) or three (D) independent experiments. In panels A, C, and D, the asterisks indicate differences between the lbtA feoB mutant and the other strains: ***, P < 0.001; ****, P < 0.0001.

Rhizoferrin promotes L. pneumophila intracellular infection of macrophages.

To assess the importance of rhizoferrin in L. pneumophila infection of macrophages, we utilized our updated panel of mutants to infect U937 cells, a human cell line that can be differentiated to a macrophage-like state and therefore is commonly used in the field (68–72). Similar to past studies (39, 44), the feoB mutant displayed a relatively modest defect in CFU recovery, whereas the lbtA mutant behaved as parental WT did (Fig. 4A). More importantly, the newly made lbtA feoB mutant, but not its complement containing intact lbtA, did not increase its CFU within the U937 cell monolayer (Fig. 4A). Although not impaired for entry into the macrophages (Fig. 4B), the double mutant proved to be specifically impaired for intracellular replication when we infected the U937 cells with GFP-expressing bacteria (Fig. 4C) or m-Cherry expressing legionellae (Fig. S2C) and monitored changes in intracellular fluorescence. Together, these data demonstrated, for the first time, that rhizoferrin promotes L. pneumophila infection of and replication within macrophage host cells. As was the case for rhizoferrin during L. pneumophila infection of amoebae, the role of rhizoferrin during L. pneumophila infection of macrophages most likely includes ferric iron acquisition.

FIG 4.

Effects of lbtA and feoB mutations on intracellular infection of human macrophages by L. pneumophila. (A) Monolayers of PMA-differentiated U937 cells were infected with WT strain 130b (WT), lbtA mutant NU302 (lbtA), feoB mutant NU458 (feoB), lbtA feoB mutant NU460 (lbtA feoB), or NU460 containing plasmid-carried lbtA (lbtA feoB/plbtA). After a 2-h incubation period to allow for bacterial entry, the infected wells were washed to remove remaining extracellular bacteria, and then at that time (i.e., t = 0) and at 24, 48, and 72 h postinoculation, the macrophages were lysed and bacterial numbers in the infected wells determined by plating for CFU on BCYE agar. At 0, 24, 48, and 72 h postinoculation, net bacterial growth (number of CFU at the denoted time/number of CFU at t = 0) was determined by plating aliquots of the supernatant on BCYE agar. Since L. pneumophila does not grow in the assay medium, any increases in CFU are the result of growth within the macrophages. (B) Differentiated U937 cells were infected with GFP-expressing WT or mutant bacteria and following a period of bacterial entry into the host cells, a trypan blue solution was added to quench the GFP signal from remaining extracellular bacteria. Entry was reported as the percentage of trypan blue-resistant GFP signal, which corresponds to internalized legionellae. For panels A and B, the data are presented as means with standard deviations of results from four (A) or three (B) independent experiments, with four (A) or six (B) technical replicates in each experiment. (C) Differentiated U937 cell monolayers were infected with GFP-expressing WT and mutant bacteria, as indicated above. After centrifugation and a 1-h incubation to facilitate uptake, gentamicin was added to kill remaining extracellular legionellae. The fluorescence from the intracellular bacteria was then monitored kinetically every hour for the next 24 h, and the fluorescence values obtained were normalized to the GFP signal at t = 0 following antibiotic treatment. (D) Monolayers of PMA-differentiated U937 cells were infected with WT strain 130b (WT), lbtA mutant NU302 (lbtA), feoB mutant NU458 (feoB), lbtA feoB mutant NU460 (lbtA feoB), mavN mutant lpw_30711 (mavN), mavN feoB mutant NU462 (mavN feoB), or mavN lbtA mutant NU464 (mavN lbtA) in standard RPMI + 10% fetal bovine serum (FBS) medium (left panel) or in standard RPMI + FBS medium that was supplemented with 100 μM ferric ammonium citrate (right panel), and then at 72 h postinoculation, net bacterial growth was determined as described in panel A. In panels C and D, the values presented are the means and standard deviations obtained from six (C) and four (D) technical replicates and are representative of the results obtained from three independent experiments. Asterisks indicate instances where a strain was impaired relative to the other indicated strains: **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Analysis of mutants lacking rhizoferrin and the MavN transporter.

MavN (IroT) is an effector of the Dot/Icm type IV secretion system that is incorporated into the LCV membrane where it most notably mediates the transport of ferrous (but not ferric) iron from the host cell cytoplasm into the lumen of the LCV (70, 73, 74). Thus, mavN mutants are highly impaired for replication within macrophages (e.g., human U937 cells) and amoebae (e.g., A. castellanii) (70, 75). This defect was fully offset by the addition of exogenous ferric iron into the Legionella-macrophage coculture (70, 75) (Fig. 4D) and partly offset by the addition of a similar amount of ferric iron into the Legionella-Acanthamoeba coculture (Fig. S3). A newly made mavN feoB double mutant behaved similarly to a mavN mutant especially regarding the effects of added ferric iron (Fig. 4D and Fig. S3), which is compatible with the fact that FeoB, like MavN, promotes the acquisition of ferrous iron, albeit at different stages. In contrast, a newly made mavN lbtA double mutant was more impaired under standard infection conditions than the mavN mutant and mavN feoB mutant were (Fig. 4D and Fig. S3), which is compatible with the mavN lbtA mutant having defects in both ferrous iron acquisition and rhizoferrin-mediated ferric iron acquisition. Indeed, the mavN lbtA mutant, unlike the mavN mutant and the mavN feoB mutant, was only partly rescued for growth in macrophages by the addition of exogenous ferric iron (Fig. 4D) and was not rescued at all by the addition of a similar amount of iron to the Legionella-Acanthamoeba coculture (Fig. S3). Interestingly, a newly made mavN lbtA feoB mutant appeared to be more impaired for intracellular survival than both the mavN lbtA mutant and mavN feoB mutant were, and it exhibited no growth-stimulation after the addition of exogenous ferric iron into the Legionella-macrophage cocultures (Fig. S4). We tested whether the impaired growth of the mavN lbtA mutant (Fig. 4D) could be better rescued by the addition of a combination of Fe³⁺, Mn²⁺, and Zn²⁺ versus the addition of Fe³⁺ alone (as we did in Fig. 4D). Not surprisingly, the addition of Mn²⁺ alone or Zn²⁺ alone did not stimulate the growth of the mavN lbtA mutant, since it did not stimulate the growth of a mavN mutant either (Fig. S5), as had been previously observed (73). However, the addition of the combination of Fe³⁺, Mn²⁺, and Zn²⁺ did not stimulate the mutant’s growth any better than did the addition of Fe³⁺ alone (Fig. S5).

Rhizoferrin triggers the production of cytokines.

Some microbial siderophores, such as deferoxamine and enterobactin, can elicit cytokine production (e.g., IL-6) by chelating the labile iron pool in macrophages and/or epithelial cells (76–80). Proinflammatory cytokines such as IL-6 and TNF-α are linked to L. pneumophila pneumonia (81–95). Thus, we tested whether rhizoferrin might also be a trigger for cytokine production, adopting the approach used to study the other siderophores noted above. When added to uninfected U937 cell macrophages, purified rhizoferrin triggered increases in IL-6 and TNF-α levels (Fig. 5). For IL-6, rhizoferrin was more effective at triggering IL-6 than was the same amount of purified deferoxamine or enterobactin (Fig. 5), even though the other two siderophores have a greater affinity for iron than does rhizoferrin (96). Thus, the elevations in cytokine triggered by rhizoferrin are likely not just a by-product of iron-chelation in the tissue culture medium. In sum, in addition to promoting bacterial replication and biofilm formation, rhizoferrin is a potent inducer of host cytokines.

FIG 5.

Effect of rhizoferrin on cytokine production by macrophages. U937 cells were treated with either 10% RPMI vehicle control (CTL), or 50 μM rhizoferrin (RF), deferoxamine (DFX), or enterobactin (ENT) in 10% RPMI. Supernatants were collected at 12 h and 24 h post treatment, and secreted IL-6 (top panel) and TNF-α (bottom panel) were determined by ELISA. Graphs show the average cytokine levels pooled from three independent experiments, each done in technical triplicate, with standard errors. Cytokine levels elicited upon treatment were normalized to the value for the CTL. Asterisks indicate points at which the values for the siderophore-treated samples were different from those for samples from CTL:*, P < 0.05; **, P < 0.01; ****, P < 0.0001.

Prevalence of rhizoferrin-related genes among L. pneumophila strains, other Legionella species, and other members of the order Legionellales.

Given the major expansion of the genome database that has occurred since lbtA and the other rhizoferrin-specific genes were first identified (44, 47, 48), we performed a bioinformatic assessment of the prevalence of the siderophore genes in currently sequenced strains of Legionella bacteria. The lbtA biosynthetic gene and its neighboring siderophore transport genes, lbtB, lbtC, and lbtU, were found in 57/57 L. pneumophila strains examined, with the corresponding proteins having 95.8% to 98.5% average amino acid (aa) identity to those of strain 130b (Table S1). Thus, rhizoferrin production and usage appears to be a conserved feature of the L. pneumophila species. The feoB and mavN/iroT genes were also 100% conserved in L. pneumophila strains, as were the genes encoding Fur and the LbtU paralog LbtP (97) (Table S1). Homologs of lbtA, lbtB, lbtC, and lbtU were also all present in the genomes of 14/58 other species of Legionella examined, with the corresponding proteins having 48% to 60% average aa identity to those of strain 130b (Fig. 6 and Table S2). Hence, rhizoferrin production and utilization likely occurs in non-pneumophila species of Legionella but is far from being universal within the genus. There were 12 other Legionella species that only lacked the gene encoding the LbtU receptor for rhizoferrin (Fig. 6 and Table S2); but they, like all Legionella species, encoded the LbtU-paralog LbtP and therefore might also produce and utilize the siderophore. In contrast to the somewhat limited distribution of the rhizoferrin-related genes, feoB, mavN/iroT, and fur were 100% conserved across all the Legionella species examined, with the one exception being the absence of mavN in L. geestiana (Fig. 6 and Table S2). Looking at other sequenced bacteria that are phylogenetically close to the Legionella genus, i.e., other species that are in the order Legionellales (98–100), Aquicella siphonis stood out for having lbtABC and lbtU (Fig. 6 and Table S3).

FIG 6.

Distribution of genes encoding rhizoferrin and other iron acquisition functions within the Legionella and related genera. (Left) A list of fully sequenced members of the order Legionellales, their phylogenetic relationships based upon data from whole-genome sequencing, and their association with human disease. A maximum-likelihood phylogenetic tree was constructed from the concatenated amino acid sequences derived from 78 near-universal single-copy genes. Bar, 0.1 amino acid substitutions per site. Legionella species colored in red have been associated with human disease, and those in black have not (yet) been linked to disease. Appearing at the top of the list are non-Legionella species (brown) that belong to other genera within the Legionellales. (Right) The presence (blue-colored square)/absence (white square) of the genes indicated across the top row in all sequenced members of the order Legionellales was determined using BLASTP. In the case of L. spiritensis, the lbtC gene was present but contained a frameshift mutation (light blue square).

DISCUSSION

Previously, we reported that intratracheal inoculation of a lbtA mutant (but not a complemented derivative) into the lungs of A/J mice results in ~10-fold fewer CFU compared to inoculation of the parental WT strain, indicating that rhizoferrin enhances the growth and survival of L. pneumophila in the mammalian lung (49). Our current data signal that the in vivo importance of rhizoferrin is likely due, at least in part, to the siderophore’s ability to enhance L. pneumophila iron acquisition during intracellular growth within macrophages. This role for rhizoferrin likely persists as host defense systems are triggered, e.g., γ-interferon-activated macrophages become less permissive for legionellae due to lowered iron pools (101, 102). Additional, noncanonical roles for siderophores that have been reported in other bacterial systems and that might also help explain the role of rhizoferrin during L. pneumophila intracellular infection include providing protection from oxidative stress and acting as a signaling molecule (103). Another question for future investigation is deciphering how the Fe³⁺ that engages rhizoferrin comes to reside within the LCV. Since MavN is known to transport Fe²⁺, Zn²⁺, Co²⁺, and Mn²⁺, but not Fe³⁺, into the LCV (73, 74), we posit that L. pneumophila orchestrates the placement of another metal importer(s) into the vacuolar membrane. Although our data clearly indicate a role for L. pneumophila rhizoferrin in intramacrophage growth, it remains possible that the siderophore also helps legionellae to survive within extracellular spaces in infected alveoli. Unlike a variety of other siderophores, rhizoferrin is not bound by lipocalin-2, a siderophore-neutralizing, antimicrobial factor that occurs in the lungs (104–106). Thus, rhizoferrin may help extracellular L. pneumophila to obtain iron despite the presence of circulating lipocalin-2. Finally, as shown here, the elicitation of IL-6 and TNF-α by rhizoferrin means that the L. pneumophila siderophore likely also helps promote an inflammatory response within the lungs, which initially would foster bacterial growth by bringing more macrophages (i.e., permissive host cells) into the local environment but over time dampens bacterial persistence (107, 108).

The pronounced functional defect exhibited by a mavN lbtA feoB mutant (relative to both a mavN lbtA mutant and a mavN feoB mutant) provided another clue that, during intracellular infection, MavN/FeoB and/or LbtA are involved in processes that extend beyond iron acquisition. Several prior observations had indicated potential roles for MavN and LbtA in Mn²⁺ and Zn²⁺ acquisition, i.e., (i) MavN, when reconstituted into proteoliposomes, could mediate the transport of Mn²⁺ and Zn²⁺ (in addition to Fe²⁺); (ii) the addition of Mn²⁺ or Zn²⁺ when in combination with ferric iron stimulates the intracellular growth of a mavN mutant more so than the addition of ferric iron alone; and (iii) purified rhizoferrin can chelate Mn²⁺ and Zn²⁺ (in addition to Fe³⁺ and several other forms of metals) (73, 96, 109–111). However, the fact that the addition of a combination of Fe³⁺, Mn²⁺, and Zn²⁺ did not stimulate a mavN lbtA mutant’s growth any better than did the addition of Fe³⁺ alone suggested that, during intracellular infection of the U937 cell macrophages by L. pneumophila, lbtA (i.e., rhizoferrin) and MavN may promote the acquisition of another metal that is not Mn²⁺ or Zn²⁺.

Although the initial impetus for our current study was to pursue the connection between rhizoferrin and L. pneumophila growth in macrophages, we have documented that rhizoferrin also promotes the growth of legionellae within A. castellanii and V. vermiformis. This finding, especially when coupled with past studies linking FeoB, MavN/IroT, and Nramp to bacterial growth in both macrophages and amoebae (39, 75, 112), indicates that iron pools and iron transport processes can be similar in mammalian versus protozoan cells. Although siderophores are extensively known for promoting bacterial and fungal growth under extracellular/planktonic conditions, the instances connecting a siderophore with microbial growth within a host cell are relatively few (113). Prior examples of bacteria using siderophores to enhance intracellular infection include Brucella species making dihydroxybenzoic acid, Mycobacterium species producing mycobactins, and Shigella species utilizing aerobactin (114–118). Yet, the past example most akin to our current findings is rhizoferrin promoting the intracellular growth of Francisella tularensis in murine macrophages (J774 cells) and a human liver carcinoma cell line (HepG2) (119, 120). In this case, the role for the siderophore was also only revealed by examining a mutant that lacked both rhizoferrin and FeoB. Thus, adding our results to that of past studies, the importance of siderophores during intracellular infection clearly encompasses a range of (i) siderophore structures (e.g., hydroxamates to catecholates to carboxylates); (ii) targeted host cells (e.g., amoebae to macrophages to epithelial cells); and (iii) niches within the infected host cells (i.e., modified phagosomes versus “free” in the host cytoplasm).

Another key experimental result obtained here was the finding that rhizoferrin promotes the ability of L. pneumophila to form a biofilm. Indeed, to our knowledge, this is the first report of rhizoferrin enhancing biofilm formation. In contrast to some other bacterial systems, such as Pseudomonas and Staphylococcus (121–124), relatively little is known about the relationship between iron and Legionella biofilm formation (125, 126). In one past study, high iron levels (i.e., five times the amount of ferric pyrophosphate that is in standard BYE) impeded L. pneumophila biofilm formation (53). In another study, the addition of the ferrous iron chelator dipyridyl increased biofilm formation, possibly by combatting the effect of iron on the generation of reactive oxygen species (60). Thus, our finding that rhizoferrin as well as FeoB-mediated ferrous transport promote the formation of a L. pneumophila biofilm on plastics is a notable advance and should prompt further investigations of their roles using biofilms formed under a wider range of conditions, including different substrata, media, flow dynamics, and temperatures (7, 125, 127–130). The incorporation of other microbial species into the L. pneumophila biofilm would be particularly interesting since some non-rhizoferrin-producing bacteria can nonetheless use (“steal”) it (131–133). Combining our biofilm results with those obtained from the amoebal infections, we can now infer that rhizoferrin plays a major role in L. pneumophila growth and spread within both natural and human-made water systems, which are the source for pathogen transmission to humans. Moreover, it is possible that inhibiting this siderophore pathway might be an alternative means to controlling L. pneumophila persistence, analogously to what has been proposed for addressing other pathogens that produce siderophores and/or assimilate other (host) iron chelates (134–138).

Our bioinformatic analysis concerning the prevalence of lbtA and its related genes suggests that rhizoferrin promotes the growth and spread of various other Legionella species, some of which are also human pathogens, as noted in Fig. 6. That some Legionella species, such as L. birminghamensis, L. brunensis, L. feeleii, and L. micdadei, do not encode lbtA (nor lbtB, lbtC, and lbtU) yet secrete a siderophore-like activity when grown in an iron-deplete CDM (139, 140) suggests that another siderophore is produced by some members of the Legionella genus. Looking beyond Legionella, rhizoferrin was first identified as a fungal siderophore, being made by many zygomycetes (141–150). Subsequently, it was found to also be produced by the bacterium Ralstonia pickettii and then Francisella species, as mentioned above (46, 151, 152). We now provide genomic evidence that rhizoferrin is also made by at least one species of Aquicella, which, like L. pneumophila, is an intracellular parasite of amoebae (153, 154).

In sum, our data indicate that rhizoferrin, a siderophore that spans diverse genera, has multifactorial importance for L. pneumophila, encompassing planktonic growth in low-iron conditions, biofilm formation, intracellular infection of amoebal and mammalian host cells, cytokine elicitation, and intrapulmonary growth.

MATERIALS AND METHODS

Bacteria, amoebae, macrophages, and media.

L. pneumophila strain 130b (American Type Culture Collection [ATCC] strain BAA-74; also known as strain AA100 or Wadsworth) served as wild type and parent for all mutants (155). The L. pneumophila lbtA mutant NU302 used here was previously described as was the L. pneumophila mavN mutant lpw_30711 (44, 75). These strains and all newly made mutants (below) were routinely grown at 37°C on BCYE agar or in BYE broth (51). A. castellanii (ATCC 30234) and V. vermiformis (ATCC 50237) amoebae were axenically maintained as before (156, 157). Human U937 cells (ATCC CRL-1593.2) were maintained and differentiated to a macrophage-like state using phorbol myristate acetate, as before (71, 72). Unless otherwise noted, chemicals were from Sigma-Aldrich.

Mutant constructions and complemented mutants.

New mutants containing deletions in lbtA and/or feoB were constructed by using a form of allelic exchange. To begin, mutagenized alleles of feoB and lbtA were generated using overlap extension PCR (OE-PCR) as previously done (71, 158). Approximately 900-bp fragments of the 5′ and 3′ regions flanking the open reading frames (ORFs) of feoB and lbtA were PCR amplified from 130b DNA using Platinum SuperFi II DNA polymerase (Thermo Fisher Scientific) and primers AL13 and AL14 for 5′ feoB, AL15 and AL16 for 3′ feoB, AL17 and AL18 for 5′ lbtA, and AL19 and AL20 for 3′ lbtA (see Table S4 for these and all other primers listed below). A kanamycin (Kn) resistance cassette flanked by Flp recombination target sites was PCR-amplified from pKD4 (158) using primers AL21 and AL22 for use in the eventual mutation of feoB and AL23 and AL24 for eventual mutation of lbtA. We then performed two-step OE-PCR to combine the 5′ and 3′ regions of feoB and the 5′ and 3′ regions of lbtA with the respective Kn resistance cassette. PCR products corresponding to the correct target size were gel purified, and 5 μg of linear DNA containing the recombinant allele carrying the antibiotic cassette were used for natural transformation (75) of either WT strain 130b or a previously made mutant of 130b as indicated below. Bacteria putatively containing an inactivated feoB or lbtA gene were obtained by plating the transformation mixture onto BCYE agar containing Kn. Verification of an altered feoB gene was done by PCR using the primer pair AL25 and AL26, and confirmation of a mutated lbtA was done by PCR using the primer pair AL27 and AL28. When the mutagenized feoB allele was introduced into WT strain 130b, the new feoB mutants obtained were designated NU458 and NU459. To construct a mutant lacking both feoB and lbtA, we introduced the mutagenized feoB allele into the previously made lbtA mutant NU302 (44). The genotype of the lbtA feoB mutants (NU460, NU461) was verified by PCR using primers AL25 and AL26. To make a mutant lacking feoB and mavN, we introduced mutagenized feoB into the previously made mavN mutant (75), and then verified the genotype of the new double mutants (NU462, NU463) by PCR using primers AL25 and AL26. To construct a mutant lacking both lbtA and mavN, we introduced the above-mutagenized lbtA allele into the mavN mutant, and then confirmed the genotype of the new double mutants (NU464 and NU465) by PCR using primers AL27 and AL28. Finally, to construct a mutant lacking mavN, lbtA, and feoB, we introduced a mutagenized mavN allele (75) into lbtA feoB mutant NU460 and then confirmed the genotype of the new triple mutants (NU466 and NU467) by PCR using primers AL29 and AL30.

Complementation analysis of the lbtA feoB mutant was done by reintroducing an intact copy of lbtA on a plasmid, as has been done before for other 130b mutants (159). To that end, a 2,132-bp fragment containing the intact lbtA ORF with its ~200 bp upstream and downstream (but no other gene) was amplified from 130b DNA using primers AL31 and AL32. Next, the product was digested with KpnI and HindIII, the lbtA-containing fragment was cloned into pMMBGent, which encodes gentamicin resistance and is stably maintained by L. pneumophila (160). The resulting plasmid (i.e., plbtA) was electroporated (155) into the lbtA feoB mutants NU460 and NU461, and transformants containing plbtA were obtained by plating on BCYE agar containing gentamicin.

Extracellular growth assays.

To assess the relative ability of L. pneumophila strains to grow under low-iron conditions, bacteria were examined for their efficiency of plating on BCYE agar containing its standard added iron supplement (i.e., 330 μM ferric pyrophosphate) versus BCYE agar lacking that supplement (47, 48). The extracellular growth capacity of strains was further quantified by measuring changes in the optical density at 660 nm (OD₆₆₀) of BYE cultures containing or not containing that ferric pyrophosphate supplement, using a DU720 spectrophotometer (Beckman Coulter) (4). Finally, bacteria that had been grown for ~12 h in 30 mL of standard BYE broth (in a 125-mL flask) with shaking at 37°C were resuspended in fresh BYE broth with or without added ferric pyrophosphate to an OD₆₆₀ equal to approximately 0.3, and then 200-μL aliquots of the suspensions were added into the wells of a 96-well (flat bottom), clear polystyrene (tissue-culture treated) microtiter plate (Fisher cat. no. FB012931). Upon incubation at 37°C with shaking, bacterial growth was monitored for 30 h by measuring absorbance at 660 nm every hour in a Synergy H1 plate reader (Biotek). Bacterial growth was further assessed by plating aliquots from the inoculated wells for CFU on BCYE agar.

Measurement of iron and other metals in bacteriological media.

Quantification of iron (Fe), manganese (Mn), copper (Cu), and zinc (Zn) levels in the non-iron-supplemented BYE medium was accomplished using ICP-MS (161, 162) and is reported in Table S5. Specifically, 1.0 mL of the medium was added to a preweighed metal-free centrifuge tube and weighed. The samples were then digested in 500 μL of concentrated trace-grade nitric acid (>69%, Thermo Fisher Scientific) and placed at 65°C for 1 h to allow for complete sample digestion. Ultrapure H₂O (18.2 MΩ·cm) was then added to produce a solution of 5.0% nitric acid in a total volume of 10 mL. Quantitative standards were made using a mixed elemental standard (Inorganic Ventures) containing 100 μg/mL each of Mn, Fe, Cu, and Zn, which was diluted to create a 100-ng/g mixed element standard in 5.0% nitric acid (vol/vol) in a total volume of 50 mL. A solution of 5.0% nitric acid (vol/vol) was used as the calibration blank. ICP-MS was done on a computer-controlled (QTEGRA software) Thermo iCapQ ICP-MS (Thermo Fisher Scientific) operating in kinetic energy discrimination mode and equipped with an ESI SC-2DX PrepFAST autosampler. Internal standard was added in-line using the PrepFAST system and consisted of 1 ng/mL of a mixed element solution (IV-ICPMS-71D) containing Bi, In, ⁶Li, Sc, Tb, Y (Inorganic Ventures). Online dilution was carried out by the PrepFAST system and used to make a calibration curve consisting of 100, 50, 25, 10, 2, and 1 ppm Mn, Fe, Cu, and Zn. Each sample was acquired using 1 survey run (10 sweeps) and 3 main (peak jumping) runs (40 sweeps). Isotopes selected for analysis were ⁵⁵Mn, ^56,57Fe, ^63,65Cu, and ^66,68Zn, and as internal standards for data interpolation and machine stability, ⁴⁵Sc, ⁸⁹Y, and ¹¹⁵In.

Biofilm assays.

Biofilm formation by strains of L. pneumophila was first examined on plastic surfaces as before (56, 61). Bacteria that had been grown on standard BCYE agar for 3 d at 37°C were resuspended in fresh BYE broth to an OD₆₆₀ of 0.2, and then 200-μL aliquots of the suspensions were added into the wells of a 96-well (flat bottom), clear polystyrene (tissue-culture treated) microtiter plate (Fisher cat. no. FB012931). After 2 d of static incubation at 30°C or 37°C, 40 μL of 0.25% crystal violet (dissolved in ddH₂0) or 1.25% safranin (dissolved in ddH₂0) were added for 15 min at room temperature. After the wells were washed three times with 300 μL of ddH₂0 using a multichannel pipettor, the plates were allowed to air dry for 30 min at room temperature as indicated. At this point, a L. pneumophila biofilm was visible as a blue ring (from crystal violet staining) or a red ring (from safranin staining) around the walls of the inoculated wells, analogous to what has been described for other bacterial biofilms (58, 61). To quantitate the mass of the biofilm, 300 μL of 30% acetic acid were added into the wells to solubilize the crystal violet or safranin for 15 min, and then 200-μL samples were transferred into a new microtiter plate and read at 600 nm for the crystal violet treated samples or 530 nm for the safranin treated samples using a Synergy H1 plate reader (Biotek). To assess bacterial growth patterns on an agar surface, legionellae that had been grown for ~12 h in 30 mL of BYE broth (in a 125-mL flask) with shaking at 37°C were resuspended in PBS to an OD₆₆₀ of 0.2, and then 20-μL aliquots were spotted onto BCYE agar. Following incubation for 5 d at 37°C, photographic images of the bacterial growth were obtained.

Intracellular infection assays.

A. castellanii and V. vermamoeba were inoculated with L. pneumophila and intracellular infection monitored, as before (156, 163, 164). Briefly, bacteria that had been grown on BCYE agar for 3 d at 37°C were inoculated onto amoebal monolayers at a multiplicity of infection (MOI) equal to 0.1, and then immediately (i.e., “t = 0”) and at 24, 48, and 72 h postinoculation, aliquots taken from the culture supernatants were assessed for bacterial numbers by plating for CFU on BCYE agar. Differentiated U937 cells were inoculated with L. pneumophila strains and intracellular infection examined as before (71, 163). Briefly, bacteria that had been grown on BCYE agar for 3 d at 37°C were added into wells containing the U937 cells at an MOI equal to 0.5, and the bacteria centrifuged onto the monolayers (i.e., 250 × g, for 5 min). After a 2-h incubation period to allow for bacterial entry, the infected wells were washed to remove remaining extracellular bacteria. At that time (i.e., “t = 0”) and at 24, 48, and 72 h postinoculation, the monolayers were lysed with saponin, and aliquots taken from the lysates were assessed for bacterial numbers by plating for CFU on BCYE agar. To assess the effect of added iron, manganese, and zinc on bacterial intracellular growth, differentiated U937 cells or A. castellanii amoebae were pretreated with 100 μM ferric ammonium citrate, manganese (II) sulfate, or zinc chloride alone or in combination (in the tissue culture medium) for 24 h and then infected as noted above while maintaining added metals in the medium (70).

To judge intracellular replication more specifically, A. castellanii, V. vermiformis, and U937 cells were infected with GFP- or mCherry-expressing legionellae at an MOI equal to 20 and increases in intracellular fluorescence were monitored over time, as previously done (157, 165). To that end, a GFP-expressing plasmid was introduced into WT and the mutant L. pneumophila by electroporation, as before (71, 165). The mCherry-expressing plasmid pON.mCherry (Addgene plasmid no. 84821, as a gift from Howard Shuman) (166) was also introduced into L. pneumophila strains by electroporation. To monitor bacterial entry into host cells, we utilized the trypan blue quenching-based entry assay, which we and others have done before for L. pneumophila (71, 157, 167–169). Thus, monolayers of A. castellanii or differentiated U937 cells were seeded into black clear-bottom, 96-well microtiter plates (Falcon), and GFP-expressing L. pneumophila strains added at an MOI of 20 (for amoebae) or 50 (for macrophages) to six replicate wells. After synchronizing the infection by centrifuging the plates, entry was allowed to occur for 1 h (for amoebae) and 5 min (for macrophages). A fluorescence reading at excitation 485 nm/emission 530 nm was then taken to gauge the initial number of fluorescent legionellae in the wells. After removing the medium containing the remaining extracellular bacteria, trypan blue was added to quench residual extracellular fluorescence, and then ultimately another fluorescence reading was taken to discern the amount of trypan blue-resistant fluorescence (i.e., intracellular legionellae) remaining in the wells. Entry was defined as percent trypan blue-resistant fluorescence between the GFP prior to and after quenching, after subtracting out background fluorescence from uninfected wells.

Cytokine assays.

To quantify cytokines secreted by U937 cell macrophages, tissue culture supernatants obtained at the indicated time points were analyzed by ELISA, as previously described (72, 170). Following the manufacturer’s recommended protocol, single ELISA plates were used to assay either human IL-6 (eBioscience, no. 88-70066-77) or human TNF-α (eBioscience, no. 88-7346-88), with the quantification of cytokine levels determined using a plate reader (Biotek Synergy H1). To gauge the effect of rhizoferrin, the U937 cells were treated with either vehicle control (10% RPMI, from Corning) or non-iron-loaded rhizoferrin (EMC Microcollections) at a concentration of 50 μM in 10% RPMI. For comparison, we separately added the same amount of deferoxamine or enterobactin, which is the level of added siderophore that has been utilized in a variety of past studies examining the effect of siderophore on cytokines (76–78, 80).

In silico analyses.

Using sequences from the draft genome of L. pneumophila strain 130b (171) as the query, BLASTP at the NCBI was done to search the genome database for proteins with primary sequence similarity to LbtA, LbtB, LbtC, LbtU, LbtP, FeoB, MavN/IroT, and Fur, as done in the past for other proteins (20, 159, 165). Homologs were predicted using an E value cutoff of < 1 × 10⁻²⁰ and an alignment length of at least 80% (172). A maximum-likelihood phylogenetic tree was made in RaxML (LG + Γ+F model) (173) from the concatenated amino acid sequences derived from 78 near-universal single-copy genes (174).

Statistical methods.

As indicated in the Figure legends, each sample or condition was assessed using ≥3 technical replicates, and the resultant values obtained were presented as the means and standard deviations or the means and standard errors. P values were determined by the Student's t test, with the Bonferroni correction applied to account for multiple t tests being performed (165). Repeat experiments (biological replicates) were routinely done for confirmation, as noted in the Figure legends.