Structural characterization of the interaction of human lactoferrin with calmodulin

Abstract

Lactoferrin (Lf) is an 80 kDa, iron (Fe(3+))-binding immunoregulatory glycoprotein secreted into most exocrine fluids, found in high concentrations in colostrum and milk, and released from neutrophil secondary granules at sites of infection and inflammation. In a number of cell types, Lf is internalized through receptor-mediated endocytosis and targeted to the nucleus where it has been demonstrated to act as a transcriptional trans-activator. Here we characterize human Lf's interaction with calmodulin (CaM), a ubiquitous, 17 kDa regulatory calcium (Ca(2+))-binding protein localized in the cytoplasm and nucleus of activated cells. Due to the size of this intermolecular complex (∼100 kDa), TROSY-based NMR techniques were employed to structurally characterize Ca(2+)-CaM when bound to intact apo-Lf. Both CaM's backbone amides and the ε-methyl group of key methionine residues were used as probes in chemical shift perturbation and cross-saturation experiments to define the binding interface of apo-Lf on Ca(2+)-CaM. Unlike the collapsed conformation through which Ca(2+)-CaM binds the CaM-binding domains of its classical targets, Ca(2+)-CaM assumes an extended structure when bound to apo-Lf. Apo-Lf appears to interact predominantly with the C-terminal lobe of Ca(2+)-CaM, enabling the N-terminal lobe to potentially bind another target. Our use of intact apo-Lf has made possible the identification of a secondary interaction interface, removed from CaM's primary binding domain. Secondary interfaces play a key role in the target's response to CaM binding, highlighting the importance of studying intact complexes. This solution-based approach can be applied to study other regulatory calcium-binding EF-hand proteins in intact intermolecular complexes.

Figure 1. Surface structure of Ca²⁺-CaM, Fe³⁺-Lf, and Fe³⁺-Tf with the electrostatic surface potential on each indicated.

Negatively charged CaM (red) would be attracted to the positively charged N-terminus of Lf (blue). Unlike Lf, Tf lacks the extremely positively-charged N-terminus. PDB codes 1CLL, 1B0L, and 1H76, respectively.

Figure 2. Biophysical Characterization of the Interaction Between Ca²⁺-CaM and Lf.

A) Non-denaturing PAGE band shift analysis for Ca²⁺-CaM binding to apo- and Fe³⁺-Lf. The ratio of CaM to Lf is indicated above each lane. B) Ribbon representation of Ca²⁺-CaM with the side chain of Lys-75 highlighted (PDB code 1CLL). The dansyl fluorophore is attached to the side chain of this residue. C) Fluorescence emission spectra of Ca²⁺-dCaM in the absence and presence of apo- or Fe³⁺-Lf in a solution of 100 mM KCl. D) Effect of increasing KCl concentration on the wavelength maximum (left) and intensity (right) of the fluorescence emission of dCaM when bound to apo- or Fe³⁺-Lf. E) Titration of dCaM with apo- or Fe³⁺-Lf as monitored through dansyl fluorescence in a solution of 100 mM KCl. The relative fluorescence intensities are plotted against the ratio of the total concentration of Lf and dCaM. F) Log K_a values displayed as bars for the binding of dCaM to apo- or Fe³⁺-Lf in solutions of varying KCl concentration. Values are represented as mean ± SEM for three independent titration experiments.

Figure 3. Binding of Ca²⁺-CaM to Apo-Lf Enacts Changes in Chemical Shifts of the C-lobe of CaM.

A) CSPs in TROSY-HSQC spectra of Ca²⁺-CaM upon binding apo-Lf. A selected region of the spectra are shown. B) CSPs between bound and unbound ²H/¹⁵N -labeled Ca²⁺-CaM plotted as a function of amino acid residue. The two horizontal lines on the graph represent standard deviations in chemical shift changes. (*) Residues with missing assignments. C) [¹³C-¹H]-HMQC spectrum of Ca²⁺-bound (¹H/¹³C-methyl-Met)/²H/¹⁵N CaM either free or bound to apo-Lf. Due to the selective labeling scheme, only the ε-methyl groups of CaM’s methionines are detected. Eight of CaM’s nine methionine amino acid residues line the hydrophobic binding pockets of Ca²⁺-CaM. Significant changes in chemical shift are seen for the ε-methyls of methionine groups found in CaM’s C-terminus of CaM.

Figure 4. Identification of Apo-Lf Contacts on Ca²⁺-CaM from Cross-Saturation Data.

A) Plot of residue-specific cross-saturation-induced amide proton signal intensity changes for apo-Lf-bound Ca²⁺-CaM. The unassigned residues are shown as having an intensity ratio of zero. The two horizontal lines on the graph represent the statistical significance of the change seen for an individual residue. B) Cross-saturation-induced methionine ε-methyl signal intensity changes for apo-Lf-bound Ca²⁺-CaM.

Figure 5. Comparison of the Ca²⁺-CaM:Apo-Lf Interface with Other Known Ca²⁺-CaM:Target Interfaces.

A) Summary of the CSP and saturation transfer data depicted on the primary sequence of CaM. Residues perturbed upon binding of apo-Lf are highlighted in green, cross-saturated are shown in blue, those that experience both CSP and cross-saturation in yellow, and missing residues by an *. Black circles indicate CaM residues for which amide atoms are found within 6.5 Å from the target peptide or protein in the indicated CaM complex deposited in the PDB. A red box highlights residues involved in the interaction with apo-Lf but uncommonly involved in interaction with other targets. The structures as indicated by their PDB codes are as follows: 1IWQ: MARCKs, 2HQW: NMDA receptor, 1CDM: CaMKIIα, 2WEL: CaMKIIδ, 2F3Y: CaV1.2 Ca²⁺ channel IQ domain motif, 3OXQ: CaV1.2 Ca²⁺ channel pre-IQ/IQ domain, 1QTX: smMLCK, 1MXE: CaMKI, 1YR5 and 2X0G: DAPK, 1IQ5: CaMKK, 2BCX: ryanodine receptor, 2KNE: plasma membrane Ca²⁺ pump, 2YGG: Na⁺/H⁺ exchanger NHE1, 1CFF: plasma membrane Ca²⁺ pump, 1K93: B. anthracis oedema factor. For each, whether or not the CBD is represented by a peptide (pept), extended sequence (ext), or the intact protein (int) is indicated. The classification of the CaM-binding motif is also shown. The secondary structure of Ca²⁺-CaM is indicated above: cylinders and arrows indicate α-helix and β-strand, respectively. CaM amino acid residues experiencing a significant change in chemical shift or cross-saturation upon binding apo-Lf are mapped on the structure of B) free Ca²⁺-CaM (PDB code 1CLL) or C) Ca²⁺-CaM bound to the CBD of CaMKK (PDB code: 1IQ5). In both, the same coloring scheme as in A is employed. Highlighted side chains reflect amino acid residues commonly found at the interaction interface, except those colored in red which reflect residues rarely involved in the binding interface.

Figure 6. Disrupting the Secondary Interface.

A) Residues that make up a secondary interface (red) as identified from CSP and cross-saturation data are indicated on the structure of Ca²⁺-CaM (grey, surface representation) bound to the CaMKK peptide (green, helix). The position of the mutant studied, Glu-139 is indicated (deep pink). B) Non-denaturing PAGE band shift analysis of the interaction of CaM mutants E139R and E139Q with apo/Fe³⁺-Lf. The ratio of CaM to Lf is indicated above each lane. C) Steady-state fluorescence titration of wild-type and mutant dCaMs with apo- or Fe³⁺-Lf. The relative fluorescence intensities are plotted against the ratio of total Lf and dCaM concentrations. D) Fluorescence parameters and derived binding constants for apo- or Fe³⁺-Lf binding to wild-type and mutant dCaMs in 100 mM KCl. Error bars represent SEM for three independent titrations.

Figure 7. Comparison of the Secondary Interface of Apo-Lf to that of Other Complexes.

From X-ray structures, ribbon representation of CaM bound to: A) intact CaMKIIδ (PDB code: 2WEL) or B) intact B. anthracis oedema factor (PDB code: 1K93). In each, the N- and C-lobes of CaM are colored purple and light blue, respectively. CaMKIIδ and oedema factor are colored orange and green, respectively. C) Secondary interfaces seen for apo-Lf or in other intact complexes (including DAPK, PDB code: 2X0G) are mapped onto the surface of free Ca²⁺-CaM and delineated by color. Glu-139 is highlighted in deep pink. The secondary interface of apo-Lf overlaps with that of the oedema factor.