Mapping of mAb epitopes
To use mAbs as probes of conformational changes, four anti-FhuA mAbs were purified to homogeneity from tissue culture supernatants by immunoaffinity chromatography. Purity (>95%) of each mAb was confirmed by SDS-PAGE with silver staining, and in all cases, a band at ∼50 kDa corresponding to the heavy chain and another at ∼25 kDa corresponding to the light chain were observed. To identify minimum FhuA epitopes, a Ph.D.-12 phage-displayed combinatorial peptide library was panned against each mAb. FhuA epitopes were determined for three mAbs using the RELIC/MATCH bioinformatics program (Table 1; Fig. 1), which performed pairwise alignments between the amino acid sequence of FhuA and each affinity-selected peptide. We previously mapped (Moeck et al. 1995) the mAb Fhu4.1 epitope to an outer surface-located region, residues 321–381. Consistent with this result, panning against this mAb significantly narrowed these boundaries, establishing residues 332–336 (LAPAD) on extracellular surface loop 4 as its cognate epitope. Peptides selected from panning against mAbs Fhu6.4 and Fhu6.6 mapped to residues 533–536 (refined from 417–550) on transmembrane strand 14 of the β-barrel.
Table Table 1.. Anti-FhuA mAb affinity-selected peptides mapped to FhuA
Figure Figure 1.. Anti-FhuA mAb epitopes mapped by phage display to FhuA (PDB code 2GRX; TonB removed for clarity; residues 1–7 were not resolved in the crystal structure). Affinity-selected peptides analyzed using RELIC and MIMOP. FhuA cork domain is shown in green; barrel domain is shown in blue. Highlighted in red circles are mAbs mapping to outer surface-exposed loops: Fhu3.1 and Fhu4.1, to the N terminus: Fhu8.3; and to β-strand 14: Fhu6.4, Fhu6.6. Ribbon representation of FhuA generated using PyMOL (http://www.pymol.org).
Download figure to PowerPoint
MAb Fhu3.1, predicted to bind to a discontinuous epitope (Moeck et al. 1995), was also panned against the Ph.D.-12 library. Thirty unique peptides were affinity selected and then analyzed using the MIMOP bioinformatics suite (Moreau et al. 2006), which includes three tools: MimAlign, MimCons, and a combined method. MimAlign and the combined method both identified the epitope for mAb Fhu3.1 that includes amino acids W246 V253 P255 P257 N258 K260 R261 P263 within loop 3 and residues Y403 N404 N416 T417 D418 within loop 5 (Fig. 1).
MAb binding kinetics altered by ferricrocin and by TonB
To investigate mAb binding to FhuA in the absence and presence of ferricrocin and of TonB, real-time binding kinetics were monitored using SPR. By SDS-PAGE followed by silver staining, purity of all preparations of FhuA and TonB (>95%) was confirmed. By immunoblotting and ELISA, we determined that mAbs were specific only for FhuA; they were not reactive toward TonB.
For SPR assays with TonB, immobilization of mAbs followed by injection of TonB alone led to nonspecific binding of TonB to reference surfaces. Hence, TonB was immobilized to sensor chips followed by injection of FhuA and then mAbs. Alternatively, FhuA was tethered by its N terminus using mAb Fhu8.3 (Moeck et al. 1995), that recognizes residues 1–7 at the N terminus of FhuA (see Supplemental material). TonB is known (Bell et al. 1990; Moeck et al. 1997; Coggshall et al. 2001; Howard et al. 2001; Moeck and Letellier 2001; Pawelek et al. 2006) to bind the Ton box of OM receptors. To further verify that TonB bound only to the periplasmic face of FhuA, mAb Fhu8.3 was injected over immobilized FhuA–TonB complexes. MAb binding was not detected, indicating that the Fhu8.3 epitope was engaged by TonB on all FhuA molecules. Thus, use of TonB or mAb Fhu8.3 to capture FhuA provided a similar orientation of FhuA to ensure that changes in mAb binding to FhuA, tethered by a physiologically relevant ligand (TonB) compared with mAb binding to FhuA, tethered by a physiologically irrelevant ligand (mAb Fhu8.3), were due to the presence of bound TonB.
Conformational changes of antigens based on properties of mAb binding have been previously investigated using SPR (Zhang et al. 2001; Bowlby et al. 2005). Guided by these reports, TonB (100–200 resonance units [RU]) was thiol-coupled to sensor chip surfaces at its N terminus, thereby orienting the capture of FhuA (50–200 RU) to mimic the spatial relationship of these two proteins in the bacterial cell envelope. MAbs Fhu4.1 and Fhu3.1, both with epitopes located on the extracellular surface of FhuA, were then injected over reference surfaces (TonB) and test surfaces (TonB + FhuA). For assays in the absence of TonB, 1800 RU of anti-FhuA mAb Fhu8.3 was amine-coupled to sensor chips followed by capture of FhuA (50–200 RU) at its N terminus. Having demonstrated that binding of FhuA by the capture antibody was unaltered by ferricrocin (data not shown), the following analytes were then injected over reference (mAb Fhu8.3) and over test surfaces (mAb Fhu8.3 + FhuA): mAbs Fhu4.1 and Fhu3.1. For all assays, specific and dose-dependent binding of mAbs to apo- or siderophore-bound FhuA surfaces was observed (Figs. 2, 3). Predicted KD values (Table 2) indicated a range of low nanomolar affinities (1.2–18 nM) for mAbs Fhu4.1 and Fhu3.1 binding to their FhuA epitopes.
Table Table 2.. Apparent kinetics of mAb binding to apo- and ferricrocin-loaded FhuA or FhuA–TonB complexes, as assessed by SPRa
Figure Figure 2.. SPR analysis of mAb Fhu4.1 binding to FhuA. MAb Fhu4.1 (10, 25, 50, 100 nM) binding to 200 RU apo- FhuA (A) or to ferricrocin-bound FhuA (B), both in the absence of TonB (i.e., 2000 RU amine-coupled mAb Fhu8.3 for capture). MAb Fhu4.1 (10, 25, 50, 100 nM) binding to 100 RU apo-FhuA (C) or to ferricrocin-bound FhuA (D), both in the presence of TonB (i.e., 200 RU thiol-coupled TonB for capture). Solid gray lines represent the raw data; black dashed lines represent mathematical curve-fitting according to the bivalent analyte model (BIAevaluation version 4.1). χ2 values and plots of residuals are shown below each panel.
Download figure to PowerPoint
Figure Figure 3.. SPR analysis of mAb Fhu3.1 binding to FhuA. MAb Fhu3.1 (10, 30, 50, 100 nM) binding to 50 RU apo-FhuA (A) or to ferricrocin-bound FhuA (B), both in the absence of TonB (i.e., 1800 RU amine-coupled mAb Fhu8.3 for capture). MAb Fhu3.1 (10, 30, 50, 100 nM) binding to 150 RU apo-FhuA (C) or to ferricrocin-bound FhuA (D), both in the presence of TonB (i.e., 100 RU thiol-coupled TonB for capture). Solid gray lines represent the raw data; black dashed lines represent mathematical curve-fitting according to a bivalent analyte model (BIAevaluation version 4.1). χ2 values and plots of residuals are shown below each panel.
Download figure to PowerPoint
For mAb Fhu4.1, its apparent rates of association (ka) with FhuA decreased 1.5- to twofold compared with its binding to FhuA–TonB (Table 2). Apparent rates of mAb dissociation (kd) from FhuA–TonB complexes showed 1.5-fold differences in the absence of ferricrocin compared with dissociation from the complex in the presence of ferricrocin (Table 2; Fig. 2). The most striking difference in mAb binding kinetics for Fhu4.1 was demonstrated by its six times faster rate of dissociation from FhuA–TonB complexes with bound ferricrocin (32 × 10−4/sec) compared with its rate of dissociation from FhuA plus ferricrocin (5.2 × 10−4/sec; Table 2). These changes in kinetic parameters show continued access by Fhu4.1 to its epitope upon ferricrocin and TonB binding. Ferricrocin and TonB binding likely induced changes in accessibility of epitopic residues that affected the apparent rate of association of mAb with FhuA. Induced changes would influence the structure of the mAb–FhuA binding interface and, in turn, the apparent rate of dissociation of mAb from FhuA. Significantly, our outcomes demonstrate that binding of FhuA to immobilized TonB reduces the overall affinity of Fhu4.1 for its epitope compared to mAb affinity for FhuA in the absence of TonB. We interpret these observations as conformational changes in extracellular loop 4 residues (332–336, LAPAD) that are recognized by mAb Fhu4.1.
Moeck et al. (1997) determined that mAbs Fhu6.4 and Fhu6.6 were not surface reactive; they gave positive reactions only by Western blotting versus denatured FhuA. Our phage panning results (Table 1) now show that these mAbs recognize a transmembrane epitope located on β-strand 14. As negative controls in our SPR assays, mAbs Fhu6.4 and Fhu6.6 (each at 100 nM) were injected over FhuA surfaces; low binding responses (0–5 RU) were observed, consistent with the location of epitopes for these mAbs.
In the absence of ferricrocin, we observed faster mAb Fhu3.1 dissociation from FhuA–TonB (11 × 10−4/sec; Table 2) than from FhuA alone (4 × 10−4 /sec; Table 2). In the presence of ferricrocin, ka for mAb Fhu3.1 binding to FhuA (23 × 104/Msec) (Table 2) versus its binding to FhuA–TonB complexes (45 × 104/Msec; Table 2) illustrated its faster rate of binding following FhuA–TonB complex formation. Apparent kd showed twofold faster mAb dissociation from FhuA–TonB complexes in the absence of ferricrocin than in the presence of ferricrocin (11 × 10−4 /sec vs. 5.5 × 10−4 /sec, respectively; Table 2; Fig. 3). These outcomes show that FhuA–TonB binding promotes conformational changes in outer surface-exposed loops 3 and 5 of FhuA. Additionally, our data indicate that the presence of ferricrocin may alter the properties of FhuA–TonB binding, thereby modifying the conformational changes that occur upon complex formation.
Repetition of the highest analyte concentration at the start and end of each titration series validated reproducibility of our assays. Binding responses for replicate injections agreed within 2 RU on average. All data were analyzed globally using the bivalent analyte model (DiGiacomo et al. 2004). χ2 values were all <1. These values along with residual plots establish the goodness of fit of the data, and reported standard errors validate their statistical significance.
Residue 336 in loop 4 undergoes conformational changes in response to binding of ferricrocin and of TonB
To extend these studies on conformational changes in FhuA, we measured changes in extrinsic fluorescence emission of FhuA(D336C) labeled with the thiol-specific fluorophore MDCC. Fluorescence emission by MDCC is affected by changes in polarity of its immediate environment. We avoided use of a reducing agent in the labeling protocol to prevent reduction of the two native disulfide linkages in FhuA, thereby ensuring that only the free cysteine at residue 336 in outer surface-exposed loop 4 was amenable to labeling by MDCC. In studies reported by Eisenhauer et al. (2005), residue 336 on FhuA(D336C) was labeled by thiol-specific fluorescent probe Oregon green maleimide. In the present work, we conjugated MDCC to the free cysteine residue in FhuA(D336C) and achieved a coupling efficiency of 1 mole of dye for each mole of protein. To ensure loading of FhuA with ferricrocin, a 10-fold molar excess of ferricrocin was added to FhuA(D336C). This addition resulted in a considerable decrease in emission maxima relative to fluorescence emissions of FhuA(D336C-MDCC) alone (Fig. 4): 33% fluorescence quenching was observed. Such an outcome is consistent with results from previous studies (Bos et al. 1998) showing that addition of ferrichrome to bacterial cells expressing FhuA labeled with fluorescein maleimide led to a decrease in fluorescence emission. Our present study has further exploited this approach to investigate the effects of FhuA–TonB complex formation. Addition of TonB to FhuA(D336C-MDCC) in a 1:1 molar ratio resulted in a marked 48% fluorescence quenching (Fig. 4). Titration of TonB (0.5 to 10 μM) into FhuA(D336C-MDCC) identified a concentration-dependent decrease in emission of extrinsic fluorescence from 5% to 56%. As negative controls, addition of 2 μM BSA or of buffer alone to FhuA(D336C-MDCC) produced emission spectra that were almost coincident with those of FhuA(D336C-MDCC) alone.
Figure Figure 4.. Fluorescence quenching of MDCC-labeled FhuA(D336C). Fluorescence emission maxima of 1 μM FhuA(D336C) (♦); FhuA(D336C) plus 1 μM TonB (+); FhuA(D336C) plus 10 μM ferricrocin (+); and FhuA(D336C) plus 10 μM ferricrocin plus 1 μM TonB (+). Plots are representative of three independent assays.
Download figure to PowerPoint
Taken together, our SPR and fluorescence studies indicate that FhuA outer surface-exposed loop 4 undergoes TonB-dependent changes in conformation in addition to changes induced by ferricrocin. Although at this time we cannot exclude the possibility that conformational changes in nearby loops affect accessibility of mAb Fhu4.1 for its epitope, in vivo and in vitro studies have also advocated the importance of loop 4 in ferrichrome binding and transport (Killmann et al. 1993; Braun et al. 1994); the LAPAD sequence was previously (Killmann et al. 1995) shown to play some role in TonB-dependent uptake of phage φ-80. By demonstrating measurable changes in mAb Fhu4.1 binding after FhuA and TonB form a complex, our data substantiate the importance of this sequence in outer surface-exposed loop 4.
Our previous measurements of FhuA–TonB binding kinetics by SPR (Khursigara et al. 2004, 2005) demonstrated an interaction best described by a two-state conformational rearrangement model. Assays using both TonB 32–239 and a C-terminal construct TonB 155–239 indicated formation of an initial lower affinity complex with FhuA and a subsequent higher affinity complex that was enhanced by the presence of ferricrocin. These studies identified a twofold increase in stability of a FhuA–TonB complex in the presence of ferricrocin compared with stability of the complex in the absence of ferricrocin. From our current SPR and fluorescence data, we conclude that binding of both ferricrocin and TonB leads to conformational changes of residue 336 of outer surface-exposed loop 4 of FhuA. Although enhancement of quenching by the addition of TonB to FhuA plus ferricrocin, beyond that observed in the absence of ferricrocin was not evident, we propose that the presence of ferricrocin leads to conformational changes in other residues of FhuA that promote binding by TonB.
Binding of ferricrocin to FhuA was shown to trigger unwinding of its switch helix (Ferguson et al. 1998b; Locher et al. 1998), an event that may signal receptor occupancy to TonB. Recently, site-directed spin labeling studies (Kim et al. 2007) indicated that the switch helix of FhuA and of FecA is unwound even in the absence of siderophore, implying that there may be a different recruitment signal for TonB. The cocrystal complex of Shultis et al. (2006) demonstrated that binding of TonB induces conformational change in an apical loop of the cork domain of BtuB. Changes in outer surface-exposed loops of the barrel domain were attributed to the crystallization process. The present study provides new details regarding conformational changes in FhuA outer surface-exposed loops 3, 4, and 5 of the barrel domain by identifying changes in apparent mAb binding kinetics for apo-FhuA versus siderophore-bound FhuA. In our model of siderophore transport, extension of the FhuA N terminus into the periplasm not only facilitates interaction at multiple sites between FhuA and TonB (Killmann et al. 2002; Khursigara et al. 2005; Pawelek et al. 2006) but also induces changes in extracellular regions of FhuA. More strikingly, conformational changes of epitopic residues in loops 3, 4, and 5 as measured by real-time apparent kinetics of mAb binding to apo- and siderophore-bound FhuA and by fluorescence spectroscopy were promoted by the presence of bound TonB. The differences in apparent kinetic parameters measured by SPR are consistent with structural changes that were not evident in the FhuA–TonB cocrystal structure (Pawelek et al. 2006), a static representation. Loops 4 and 5, the longest loops on FhuA, may be involved in a gating mechanism as identified for loops 7 and 8 in FecA (Ferguson et al. 2002), and as suggested for loop 8 in FhuA (Faraldo-Gomez et al. 2003), to prevent diffusion of substrate into the extracellular environment. Movement of the epitopic residues signals a change in the molecular architecture of the mAb binding regions that decreases affinity of ferricrocin for its initial binding site on FhuA, thus promoting siderophore import. Hence, conformational changes in extracellular loops play a measurable and demonstrable role in the siderophore transport cycle of this OM receptor. Binding kinetics of loop residues of other OM receptors at different stages of the cycle will elucidate commonalities in their transport mechanisms.