The Haemophilus influenzae hFbpABC Fe³⁺ Transporter: Analysis of the Membrane Permease and Development of a Gallium-Based Screen for Mutants

Ampicillin, 2,2′-dipyridyl (Dip), buffers, glucose, nitrilotriacetic acid (NTA), cetyltrimethylammonium bromide (CTAB), trans-1,2-diamoncyclohexane-N,N,N′,N′-tetraacetic acid (CDTA), phenylalanine, tyrosine, tryptophan, ferric nitrate, and gallium nitrate were all purchased from Sigma-Aldrich (St. Louis, MO). Nutrient broth (NB), Luria broth (LB), Bacto agar, and sterile supplement disks were purchased from Difco (Detroit, MI). Chelex-100 and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) reagents were purchased from Bio-Rad (Hercules, CA). ⁵⁵Ferric chloride was purchased from New England Nucleides (Boston, MA). Oligonucleotides were purchased from Invitrogen (Carlsbad, CA). Taq polymerase was purchased from Roche (Basel, Switzerland). Restriction enzymes were purchased from New England Biolabs (Beverly, MA). Nitrocellulose filters and scintillation fluid were purchased from Fisher (Pittsburgh, PA). E. coli strains and plasmids were obtained as described (Table 1).

Assays measuring growth sensitivity to the iron chelator Dip were performed under the following conditions. Single colonies were used to inoculate NB supplemented with 100 μg/ml ampicillin (NBamp₁₀₀), and cultures were incubated at 37°C with shaking at 250 rpm. Cells at mid-log growth (optical density at 600 nm, 0.5) were centrifuged, suspended in NBamp₁₀₀ containing 200 μM Dip (NBamp₁₀₀dip₂₀₀) top agar (0.7%), and seeded at 10⁶ CFU/plate on NBamp₁₀₀dip₂₀₀ agar. Plates were incubated in a water-jacketed incubator at 37°C, 5% CO₂ for 16 h. Following growth, plates were digitally scanned. Growth was evaluated as follows: −, no colonies; +, pinpoint colonies.

Assays measuring exogenous Fe³⁺(NTA)₂ growth rescue on media containing the Fe³⁺-specific chelator CDTA were performed under the following conditions. Single colonies were used to inoculate NBamp₁₀₀, and cultures were grown at 37°C with shaking at 250 rpm. Mid-log cells were centrifuged, suspended in NBamp₁₀₀ containing 600 μM CDTA (NBamp₁₀₀CDTA₆₀₀) top agar, and seeded at 10⁶ CFU/plate on NBamp₁₀₀CDTA₆₀₀ agar. Sterile supplement disks (6 mm) loaded with 100 mM Fe³⁺(NTA)₂ (10 μl) were applied, and the plates were incubated as described above. Following growth, plates were digitally scanned and zones of growth were measured.

Radiolabeled iron transport assays were performed as previously described (7). Briefly, M9 transport medium (6 mg/ml Na₂HPO₄, 3 mg/ml KH₂O₄, 0.5 mg/ml NaCl, 1 mg/ml NH₄C, 1 mM MgSO₄, 0.1 mM CaCl₂) was treated with 100 mg/liter Chelex-100 and supplemented with 2 mg/ml glucose, 0.1 mg/ml phenylalanine, 0.1 mg/ml tyrosine, and 0.1 mg/ml tryptophan prior to use. Fresh transformants were grown in LB supplemented with 100 μg/ml ampicillin (LBamp₁₀₀), seeded on NBamp₁₀₀dip₇₅, and grown at 37°C for 18 h. Cells were washed in transport medium, and following a 10-min preincubation at 37°C, radiolabeled Fe(NTA)₂ (3 × 10⁴ cpm/pmol) was added to a final concentration of 1 μM. At designated time points aliquots were removed, filtered through 0.45-μm-pore-size nitrocellulose filters, and washed with 100 mM LiCl. Activity (in counts per minute) was measured using a liquid scintillation counter (Packard, Billerica, MA).

The Gene Editor System (Promega, Madison, WI) was used to perform site-directed mutagenesis. Plasmid DNA was isolated from single putative mutants using standard miniprep techniques and was subjected to automated DNA sequencing using an ABI Prism 3100 sequencer operated by the Department of Molecular Genetics and Biochemistry Shared Resources Facility at the University of Pittsburgh.

H-1443 E. coli cells harboring the pAHIO plasmid were grown to mid-log phase in LBamp₁₀₀. Cells were centrifuged, suspended in LBamp₁₀₀, and seeded at 10⁹ CFU/plate in parallel on NBamp₁₀₀dip₇₅ and NBamp₁₀₀dip₇₅ containing 100 μM Ga(NO₃) (NBamp₁₀₀dip₇₅Ga₁₀₀) agar. Plates were incubated as described above. Growth of these cells on NBamp₁₀₀dip₇₅ resulted in slightly smaller colonies than those of pAHIO/H-1443 grown on NBamp₁₀₀ (without Dip), while the same cells on Ga³⁺-containing medium (NBamp₁₀₀dip₇₅Ga₁₀₀) exhibited a highly growth-suppressed microcolony phenotype. On NBamp₁₀₀dip₇₅Ga₁₀₀ plates, however, approximately 1 × 10⁻⁸ colonies (1 to 10 CFU/plate) exhibited a larger macrocolony phenotype among a lawn of microcolony growth. One of these gallium-resistant macrocolonies was chosen as a putative mutant for further study.

In the original development of the screen, a single putative mutant was selected and grown in LBamp₁₀₀, and plasmid was isolated using standard miniprep techniques. Purified plasmid was used to transform fresh H-1443 E. coli, and cells were seeded in parallel on NBamp₁₀₀dip₇₅ and NBamp₁₀₀dip₇₅Ga₁₀₀ plates. Upon passage of the gallium resistance phenotype, a single colony was selected and grown in LBamp₁₀₀ broth to stationary phase. The culture was divided into three aliquots. Aliquot one was used to isolate plasmid DNA; a 0.5-μg aliquot of this DNA was run on a 1% agarose electrophoresis gel and compared to wild-type pAHIO plasmid. The other two aliquots were subjected to FbpA protein analysis. Aliquot two was centrifuged, washed in cold phosphate-buffered saline (PBS), suspended in 30 mM Tris, pH 8.0, 20% sucrose, 1 mM EDTA, and shaken at 25°C for 5 min. The suspensions were pelleted, suspended in 5 mM ice-cold MgCl₂, and shaken for 10 min at 4°C. The suspensions were again pelleted, and the supernatant was saved as the periplasmic osmotic shock fluid. Aliquot three was centrifuged, washed with PBS, suspended in 400 mM Tris, pH 8.0, 2% CTAB, and shaken at 37°C for 1 h. The suspension was pelleted, and the supernatant was saved as the soluble whole-cell lysate. The soluble periplasmic fraction and the soluble whole-cell fraction were subjected to SDS-PAGE on a 12% gel. The gels were stained with Coomassie blue and subjected to densitometry using a Kodak Imagestation 1000 (Rochester, NY).

The initial gallium-resistant mutant was screened using the above methods to determine whether (i) the phenotype was linked to the pAHIO plasmid, (ii) the phenotype was not due to any major insertions or deletions within pAHIO, or (iii) the mutation was not within or upstream of the hitA gene resulting in lack of expression, truncation, or misprocessing of the hFbpA protein. Upon satisfying these criteria, the hitBC locus within the mutant plasmid was subjected to automated DNA sequencing as described above.

Upon establishing conditions for gallium selection of pAHIO mutants, directed mutations localized to the hitB gene were generated by subcloning putative mutant hitB genes into the pAHIO backbone. This required engineering unique restriction sites into the 5′ and 3′ ends of the hitB gene within the pAHIO plasmid. A unique NsiI 5′ hitB site was inserted at 1525 and a MluI site at 3109 was deleted, yielding a unique MluI site at 3064 at the 3′ end of hitB and resulting in the pNΔMHIO plasmid. To enrich the initial pool of putative mutants, a PCR-based random mutagenesis approach was used. The GeneMorph kit (Stratagene, La Jolla, CA) was used along with primers specific to the 5′ and 3′ ends of hitB incorporating NsiI and MluI restriction sites under PCR conditions consistent with low-mutation frequency (0 to 3 mutations/kb). The mutagenic PCR product was restricted with NsiI and MluI, purified, and subcloned into the identical sites within pNΔMHIO. Ligated plasmid was used to transform fresh H-1443 E. coli cells. Putative clones were selected, grown in LBamp₁₀₀ to mid-log phase, and seeded on NBamp₁₀₀dip₇₅Ga₁₀₀ plates. Clones exhibiting positive growth on NBamp₁₀₀dip₇₅Ga₁₀₀ were selected and grown in LBamp₁₀₀ to stationary phase, and plasmid was isolated and used to transform fresh H-1443. Upon passage of the gallium resistance phenotype, mutant DNA was isolated and the hitB locus was subjected to DNA sequencing as described above.

A transposon-mediated β-lactamase fusion strategy was employed to monitor hFbpB point mutant protein expression levels. Specifically, the EZ-TN5 β-lactamase fusion kit (Epicenter, Madison, WI) was used to generate random insertions within pACYCHIO, a pACYC184-based construct with hitABC inserted between the EcoRV and BamHI sites of the Tet^r gene (7) (Table 1). Following a standardized transposon insertion reaction, TOP10 (recA) E. coli cells were transformed and plated on LB containing 40 to 70 μg/ml of ampicillin to select for Amp^r insertion clones. Putative clones were selected and subjected to HindIII restriction analysis. All clones that demonstrated the presence of the 1.2-kb EZ-TN5 blaM insert and appeared to be unique were submitted for sequencing.

One such Amp^r clone which possessed an in-frame blaM insertion at nucleotide 1098 within the hitB gene (designated pB366) was selected and subjected to overlap extension PCR mutagenesis using the NdeI-5′, RevQEc-3′, ForQEc-5′, and BspHI-3′ primers (Table 1). This method replaced the blaM translational stop codon with a glutamine codon (TAA>CAA) and introduced an EcoRV restriction site directly downstream of this missense site. The final 3.7-kb PCR product was digested with NdeI-5′ and BspHI-3′ and ligated into compatible sites within pACYCHIO to generate pB366QE. The downstream portion of the hitB coding region was PCR amplified from the pB366 plasmid using the EcRV5410-5′ and BspHI-3′ primers to generate a 1.9-kb product with an introduced 5′ EcoRV restriction site. This PCR product was digested with EcoRV and BspHI and ligated into a compatible site within pB366QE to generate pB366BLAM. The resulting pB366BLAM plasmid encoded the hFbpB-BlaM fusion protein, with β-lactamase inserted between residues 366 and 374 of the hFbpB permease.

Significant hFbpB point mutants which demonstrated null or attenuated iron transport phenotypes were introduced into both pB366BLAM and pACYCHIO (control) backgrounds and tested for positive growth on ampicillin-containing media. Point mutations were generated in pACYCIO using the Gene-Editor mutagenesis kit (Stratagene) and then were PCR amplified and subcloned into the pB366BLAM plasmid. All mutations were verified by sequencing. Cells were grown in LB containing 30 μg/ml chloramphenicol to mid-log phase, diluted in LB, and then plated on LB agar containing ampicillin ranging in concentrations from 40 to 100 μg/ml. Growth of the point mutants was monitored and compared to that of both wild-type pB366BLAM and the pACYCHIO negative controls.

Optimal alignment of the permease amino acid sequences was performed using ClustalW 2.0 (http://align.genome.jp/ ), and shading was done using AMAS (33). Set 1 included nFbpB (N. gonorrhoeae GenBank gi, 1098688), nFbpB (N. meningitidis gi, 7379559), hFbpB (H. influenzae gi, 619399), SfuB (Serratia marcescens gi, 1173433), YfuB (Yersinia enterocolitica gi, 619574), and MhFbpB (Mannheimia haemolytica gi, 3978165). Sequence sets 2 to 4 were obtained by BLAST analysis using hFbpB as the search sequence and selecting those proteins that fit the classification: Set 2, iron siderophore; Set 3, thiosulfate/molybdenum/putrescine/glycine-betaine; Set 4, oligosaccharide/glycerophosphate permeases. Hydropathy analysis, topological orientation, and secondary structural predictions were performed using the PredictProtein database (43).

As a starting point in our analysis of the hFbpB permease, we sought to compare the primary sequences of FbpB homologs to other permeases of the bacterial ABC transporter family. Tam and Saier demonstrated that high-affinity periplasmic binding proteins, which bind similar solutes but are derived from evolutionarily diverse bacterial sources, exhibit basic sequence similarities (48). Based on sequence alignments and solute binding characteristics, the proteins can be divided into eight general families or clusters. This classification has recently been updated to include a ninth family of manganese and zinc binding proteins (16). By such criteria, the ferric ion-binding proteins (FbpAs) were grouped into cluster 1 along with proteins that bind oligosaccharides, phosphate, and glycerol-3-phosphate, among others. By contrast, the siderophore-iron and vitamin B₁₂ binding proteins were grouped into cluster 8. This categorization is consistent with the prediction that the FbpA proteins share a common anion-binding protein predecessor with phosphate binding protein (14). The distinction between the FbpA proteins and the siderophore binding proteins is also consistent with structural and biochemical data that indicate that these proteins exhibit distinct binding mechanisms and that the substrates (Fe³⁺ versus Fe³⁺-siderophore) are chemically diverse (14, 15). A subsequent study by Saurin et al. focused on the consensus motifs of the membrane permeases, the only significantly conserved regions of these proteins (44). In this work, sequence alignments established the motifs as signatures for permeases of related function. Here, the FbpB permeases were grouped with permeases involved in the uptake of molybdenum, thiosulfate, putrescine, and glycine-betaine. In contrast, the FbpB homologs were distinguished from permeases involved in oligosaccharide, phosphate, and glycerophosphate uptake as well as those involved in siderophore-iron uptake.

In accordance with the sequence conservation between permease subfamilies and in an attempt to highlight residues of potential functional relevance, we performed multiple-sequence alignments of the conserved motifs of the FbpB permeases (Set 1) with members of the siderophore-iron (Set 2), thiosulfate, molybdenum, putrescine, and glycine-betaine (Set 3), and oligosaccharide, phosphate and glycerophosphate (Set 4) permease families (Fig. 1). The alignments demonstrate the presence of two glycine residues that are completely conserved among all the permeases analyzed and strictly conserved among all known members of the bacterial ABC transporter permease family (Fig. 1). Immediately preceding these invariant glycines are two leucines that are absolutely conserved among the FbpB permeases (Fig. 1). These leucines are well conserved among members of Set 2 and Set 3 as well; the only exceptions are FhuB (E. coli) (Val in motif 2) and ProW (E. coli) (phenylalanine in both motifs). Alternatively, in Set 4 permeases these leucines are exchanged with aspartates, which are completely conserved among members of this set (Fig. 1). The leucine-to-aspartate covariance within both motifs indicates that these residues may have a definitive role in permease function.

In addition to the aforementioned residues, the FbpBs (Set 1) possess a conserved phenylalanine in motif 1 that is shared among Set 4 and Set 3 permeases, with the exception of ProW (E. coli) (Leu; note that this residue covaries with the Phe noted above). This phenylalanine is not at all conserved among Set 2 permeases. A similar scheme is observed with a conserved proline residue found within motif 1. Motif 2 demonstrates an analogous trend regarding conservation of a key proline, whereas SfuB and YfuB of Set 1 possess arginine residues at this site.

Site-directed mutagenesis was performed to probe the functional relevance of conserved hFbpB residues. The invariant hFbpB glycine residues, Gly155 of motif 1 and Gly418 of motif 2, were initially replaced with the small aliphatic residue alanine, and effects on transporter function were measured using the radiolabeled Fe³⁺ transport assay as described in Materials and Methods (7). The H-1443/pBR322 (vector-only control) cells demonstrated a minimal level of iron uptake, consistent with the inability of this strain to transport iron under defined assay conditions, with Fe(NTA)₂ as a supplement (Fig. 2). The H-1443/pAHIΔC control cells demonstrated a low-level increase in signal over time (4.96 ± 0.78 pmol Fe/10⁹ cells at 7 min [23.7% of pAHIO results]) (Fig. 2). This result is not due to functional hFbpABC transport; rather, it is due to binding of labeled ⁵⁵Fe(NTA)₂ to hFbpA within the periplasm of H-1443/pAHIΔC cells (Fig. 2B), as detailed in a previous study (7). Nonetheless, this control served as the baseline for productive transport in these experiments. Wild-type control H-1443/pAHIO cells demonstrated a high-level time-dependent increase in signal (20.95 ± 2.69 pmol Fe/10⁹ cells at 7 min) (Fig. 2). Transport activities of the pAHIO and pAHIΔC controls were similar to those reported previously, and hFbpA protein expression levels were indistinguishable (data not shown) (7).

The G155A mutant demonstrated transport levels only slightly lower than those of wild-type control cells (17.08 ± 0.65 pmol Fe/10⁹ cells at 7 min) (Fig. 3A; Table 2). This result was unexpected, considering this glycine is completely conserved among all permeases studied to date. Furthermore, investigations in other transport systems have demonstrated that mutation of the invariant glycine on either conserved motif leads to altered transport activity (32). To verify this result, several additional mutations were tested, including replacement with a larger aliphatic residue (G155V) and a charged residue (G155E). The G155V mutation resulted in transport activity that was similar to that of the wild type (19.09 ± 0.83 pmol Fe/10⁹ cells at 7 min), while the G155E mutation resulted in slightly higher levels of activity than that of the wild type (22.68 ± 1.40 pmol Fe/10⁹ cells at 7 min) (Fig. 3A; Table 2); both levels were within standard errors of the wild-type control (Fig. 3A). These observations demonstrate that mutation of Gly155 has no significant effect on activity, and the identity of the residue has little consequence on proper permease function according to this assay. By contrast, the G418A mutation in motif 2 led to a significant reduction of iron transport (12.49 ± 2.97 pmol Fe/10⁹ cells at 7 min) compared to that of wild-type H-1443/pAHIO cells (23.33 ± 2.17 pmol Fe/10⁹ cells at 7 min) (Fig. 3B; Table 2). Thus, this conserved glycine in motif 2 is required for proper permease function.

To probe the significance of the conserved leucine residues immediately preceding the glycines, the L154D and L417D mutations were constructed and tested for transport. Aspartate substitutions were made at these sites to directly address the observation that the conserved leucines among sets 1, 2, and 3 are exchanged for conserved aspartates among Set 4 permeases. The L154D mutation in motif 1 demonstrated significantly lower activity levels than the wild type (10.58 ± 1.36 pmol Fe/10⁹ cells at 7 min) (Fig. 4A; Table 2), indicative of attenuated iron transport. The L417D mutation in motif 2 exhibited dramatically lower activity levels than the wild type (6.63 ± 2.04 pmol Fe/10⁹ cells at 7 min) (Fig. 4B; Table 2). This latter activity is similar in magnitude to that of the H-1443/pAHIΔC control (5.84 ± 1.44 pmol Fe/10⁹ cells at 7 min) and is indicative of null iron transport.

Several other conserved residues were targeted for mutation and had minimal effects on transport activity: E149Q (20.11 ± 1.42 pmol Fe/10⁹ cells at 7 min), S153A (22.92 ± 2.52 pmol Fe/10⁹ cells at 7 min), F162L (24.36 ± 1.45 pmol Fe/10⁹ cells at 7 min), and P168A (26.2 ± 3.59 pmol Fe/10⁹ cells at 7 min). All values were within standard errors of H-1443/pAHIO cells (data not shown).

To substantiate the results of the radiolabeled iron transport assay, further qualitative and semiquantitative assays measuring growth on complex media in the presence of Fe³⁺ chelators were performed. The first such assay measured growth or lack of growth on NB supplemented with 200 μM Dip, an avid iron chelator. This assay was used in the molecular cloning of the sfuABC, hitABC, and fbpABC operons in E. coli strain H-1443 and has been used successfully in recent complementation experiments (1, 2, 7, 8). The second assay, developed specifically for the current study, measured growth rescue by supplemental Fe(NTA)₂ on media containing the Fe³⁺-specific chelator CDTA. This assay is a more specific measure of Fe³⁺ uptake as, unlike Dip, which is predominantly an Fe²⁺ chelator, CDTA coordinates Fe³⁺ with high affinity (FeCDTA¹⁻ log β = 28; FeOHCDTA²⁻ log β = 19) (42). The results of these assays along with results of the radiolabeled transport assay are summarized in Table 2. The combined data indicate that the L154D mutation results in attenuated iron transport, L417D results in null transport, G418A results in attenuated transport, and all others result in wild-type iron transport activity.

Rather than undertake a large-scale site-directed mutagenesis effort on a protein lacking significant sequence conservation or hot spots of interest, we sought alternative approaches in pinpointing further functionally significant residues. In a previous study, we demonstrated that several metals inhibited the growth of H-1443 E. coli cells (7). Interestingly, gallium (Ga³⁺) toxicity was specific to H-1443/pAHIO cells (H-1443/pBR322 and H-1443/pAHIΔC cells were unaffected), thus providing a correlate between hFbpABC transport and gallium-induced toxicity. Using a radiolabeled Ga⁶⁷(NTA)₂ transport assay, we subsequently demonstrated direct Ga³⁺ uptake by the hFbpABC transporter (7).

During the course of these metal competition experiments, we made an intriguing observation that warranted further investigation. Upon plating on Ga³⁺-containing medium (NBamp₁₀₀dip₇₅Ga₁₀₀), H-1443/pAHIO cells exhibited a growth-suppressed microcolony phenotype, indicative of Ga³⁺-induced toxicity. However, approximately 1 × 10⁻⁸ colonies (1 to 10 CFU/plate) exhibited a larger macrocolony phenotype among a lawn of microcolony growth. We hypothesized that these macrocolonies had developed a mutation(s), perhaps within the hFbpABC transporter, that rendered the cells insensitive to Ga³⁺ toxicity. As detailed in Materials and Methods, an initial putative mutant was taken through the screening protocol, and the following criteria were satisfied: the gallium-resistant phenotype remained following passage on NBamp₁₀₀dip₇₅Ga₁₀₀ plates, the mutation did not result in major insertions or deletions within the pAHIO plasmid, and the mutation did not cause alterations in hFbpA expression or secretion. Upon sequencing the hitB and hitC genes, a single-nucleotide missense mutation was identified within hitB, which translated into a single-amino-acid point mutation (V497I). No other mutations were found, and site-directed mutagenesis back to the wild type (I497V) resulted in complete reversion to the wild-type gallium-sensitive phenotype (Fig. 5A). Results of the radiolabeled iron transport assay demonstrated that the V497I mutation led to a significant effect on iron uptake (10.44 ± 2.39 pmol Fe/10⁹ cells at 7 min) (Fig. 5B; Table 3). The I497V reverse mutation led to transport activity similar in magnitude to that of wild-type control cells (23.3 ± 2.5 pmol Fe/10⁹ cells at 7 min) (Fig. 5B; Table 3).

With proof of concept in hand, we developed a subcloning strategy to generate a series of further mutations that were directed to the hitB gene. To limit the extensive screening process involved in authentication of the initial mutation, we implemented a PCR-based random mutagenesis method which created an enriched starting pool of putative mutants. This enrichment, coupled with the subcloning and gallium selection protocols described in Materials and Methods, led to the identification of further mutations localized within the hFbpB permease. The mutations arose from numerous locations throughout the hitB gene and represented a diverse sampling of amino acid alterations. The mutants were subjected to the transport assays as shown in Table 3. Subsequently, each mutation was genetically reversed to the wild type, which in turn restored wild-type transport activity (data not shown).

The initial single-site mutation, I174F, demonstrated null transport activity, while a second mutation, S146Y, demonstrated attenuated activity (Table 3). A third mutant possessed a double mutation, I174N/S475I. Individual single mutations were constructed to discern whether individual alterations contributed to the null iron transport activity or whether one of the mutations was actually a false positive. Indeed, the single I174N mutation resulted in wild-type iron transport activity, while the S475I mutation led to a null iron transport phenotype (Table 3). Admittedly, this result was unexpected, as the previously identified mutation, I174F, resulted in null iron transport activity. This inconsistency is likely due to the difference in substituted residues; phenylalanine introduces a large aromatic moiety which may impart a significant steric or electrostatic effect, while the polar uncharged asparagine may be well tolerated at this site.

A fourth point mutation, I383N, exhibited curious activity. Although this mutation led to gallium resistance under selection conditions, iron transport activity was apparently unaffected (Table 3). As both Fe³⁺ and Ga³⁺ are trivalent metals that share similar ionic radii (Ga³⁺, 0.62 Å; Fe³⁺, 0.65 Å), gallium-induced toxicity is normally thought to arise by competitive inhibition of Fe³⁺ transport. The I383N result signifies that there may actually be subtle differences between the mechanisms of Ga³⁺ and Fe³⁺ hFbpABC transport and that modest mutations within the permease may create a transporter that is selective for one metal over the other. Preliminary experiments suggest that although both metals form stable complexes with hFbpA, the metal-protein interactions are significantly different. UV difference spectra suggest weaker binding for Ga³⁺ to nFbpA in the presence of phosphate (26), and SUPREX analysis of matrix-assisted laser desorption ionization-time of flight mass spectra shows that the Ga³⁺- and Fe³⁺-bound proteins operate in different folding regimens (K. D. Weaver, P. L. Roulhac, M. C. Heymann, M. C. Fitzgerald, T. A. Mietzner, and A. L. Crumbliss, unpublished data). Further experiments will address whether the Ga³⁺-bound conformation of hFbpA is indeed different from the Fe³⁺-bound form and whether I383N can discriminate between these potentially distinct conformations.

Clearly, further investigation of the described residues, through saturation mutagenesis and detailed biochemical assays, is required to delineate their specific roles in the mechanism of transport. Before advancing these experiments further, however, we thought it was essential to develop an assay with which to measure permease protein expression levels. As mentioned previously, the permease contains an inordinate number of hydrophobic residues (∼65%) and has proven to be particularly difficult to track. Recombinant overexpression results in cell toxicity preventing isolation by traditional means, and low expression coupled with the propensity to adhere to cellular membranes and to aggregate has precluded reliable visualization on SDS-PAGE and Western blots. In response to these limitations, we developed an assay which allowed us to track functional expression through the incorporation of a β-lactamase (blaM) fusion within the hitB gene.

Using a transposon-based strategy, we generated a β-lactamase (blaM) gene fusion within pACYCHIO, specifically at nucleotide 1098 within the hitB gene of the hitABC locus. Following the transposon insertion reaction and sequencing to verify the site of insertion, the β-lactamase stop codon was replaced with a glutamine codon and the noncoding downstream transposon sequence was deleted. This created an in-frame “sandwich” fusion, with β-lactamase inserted between hFbpB amino acids 366 and 374. Transposon-mediated insertion of the blaM gene included the insertion of upstream mosaic sequence, which translated as an 11-residue hydrophobic linker (LSLIHISTIID) between hFbpB residue 366 and BlaM residue 1. To accommodate this linker and to negate the effects of redundant hydrophobic segments, hFbpB residues 367 to 373 were removed. The hFbpB-BlaM fusion (expressed from the pB366BLAM plasmid) demonstrated resistance to ampicillin up to a concentration of 100 μg/ml on LB plates. This result signified that both expression and proper membrane orientation was retained by the hFbpB-BlaM fusion protein, with β-lactamase localized to the periplasmic side of the inner membrane. Mutations in hFbpB that resulted in null or attenuated iron transport phenotypes (as described above) were introduced into pACYCHIO (negative control) and pB366BLAM, and growth on LBamp₁₀₀ was assessed (summarized in Table 1). Results demonstrated that each of the mutant strains tested grew on LBamp₁₀₀ in a fashion similar to that of wild-type pB366-BLAM, while mutations in the pACYCHIO control background demonstrated no such growth. This indicated that the mutations (L154D, L417D, G418A, and Y497I) did not significantly alter FbpB permease expression levels compared to those of the wild type (Table 1). Although this β-lactamase fusion assay demonstrated that the point mutations did not significantly alter expression or membrane incorporation of the hFbpB-BlaM proteins, this does not rule out the possibility that wild-type hFbpB-BlaM expression differs from that of wild-type hFbpB.