Metalloprotein-based MRI probes

Abstract

Metalloproteins have long been recognized as key determinants of endogenous contrast in magnetic resonance imaging (MRI) of biological subjects. More recently, both natural and engineered metalloproteins have been harnessed as biotechnological tools to probe gene expression, enzyme activity, and analyte concentrations by MRI. Metalloprotein MRI probes are paramagnetic and function by analogous mechanisms to conventional gadolinium or iron oxide-based MRI contrast agents. Compared with synthetic agents, metalloproteins typically offer worse sensitivity, but the possibilities of using protein engineering and targeted gene expression approaches in conjunction with metalloprotein contrast agents are powerful and sometimes definitive strengths. This review summarizes theoretical and practical aspects of metalloprotein-based contrast agents, and discusses progress in the exploitation of these proteins for molecular imaging applications.

Figure 1. Mechanisms of MRI contrast enhancement by paramagnetic metalloproteins

(A) Protein-based T₁ contrast agents operate largely through an inner sphere relaxation mechanism dependent on metal-coordinated water molecules in fast exchange with bulk solvent. In the example shown (top), the heme iron (brown) of an engineered P450 BM3 variant (gray ribbon structure) interacts with an axial inner sphere water ligand (blue ball, arrowhead) [24]. Bound water protons (not shown) experience dipole coupling with the metal ion and undergo relaxation. T₁ relaxivity of the complex may also be promoted by the presence of second sphere water molecules (additional blue balls) that participate in weak dipole coupling with the iron atom and exchange with water in the inner sphere position. The T₁ contrast affects the saturation of MRI signals during repeated application of a pulse sequence (schematics at bottom; pulses in gray and raw MRI signal in black). In the absence of the contrast agent, the MRI signal tends to decline due to saturation over time (top trace). The T₁ agent relieves this effect (bottom trace), resulting in a enhanced signal and relative hyperintensity in images. (B) The best-characterized protein T₂ contrast agent is ferritin (Ft, ribbon structure), a shell-shaped oligomer of 24 subunits (one shown in purple) that contains a paramagnetic hydrated iron oxide core. The core induces a dipolar field perturbation (field lines shown) over a length scale comparable to the core diameter, and water molecules (blue balls) undergo transverse relaxation by diffusing through the dipole field. The amplitude of an MRI signal obtained using a T₂-weighted pulse sequence (schematics at bottom) depends on the amount of T₂ relaxation that has occurred prior to acquisition of the signal with each repetition of the pulse sequence. A T₂ contrast agent like Ft promotes T₂ relaxation and leads to attenuation of the signal (bottom trace) and relative hypointensity in MRI scans. Ft structural model from reference [125].

Figure 2. Engineered metalloprotein-based MRI sensors

(A) Metalloprotein-based T₁ contrast agents can sense analytes via a so-called “q-modulation” mechanism. In this mechanism, exemplified by BM3h dopamine sensors, inner sphere water molecules bound to the paramagnetic center in the ligand free structure (left) are displaced by analyte binding (right). For the BM3h-based sensors, neurotransmitter binding reduces q of the heme iron atom from one to zero. (B) The ligand dependent change in q induces a sharp drop in the r₁ of BM3h variants, from roughly 1 to 0.2 mM⁻¹s⁻¹ for the best two dopamine-binding variants, BM3h-B7 and -8C8, identified in reference [25]. The relaxivity decrease upon dopamine binding also leads to a reduction in the corresponding T₁-weighted MRI signal intensities (inset). (C) Metalloprotein-based T₂ sensors can be constructed by modifying protein domains in Ft to include analyte-sensitive moieties. In the example of reference [99], a kinase-sensitive Ft-based probe was constructed by genetically fusing the KID domain of the protein CREB and the KIX domain of the protein CBP to Ft light chain to make chimeric KID-Ft (blue) and KIX-Ft (magenta) variants. In the presence of protien kinase A (PKA), KID domains are phosphorylated and tend to bind KIX domains, leading to clustering of the multivalent KID-Ft and KIX-Ft proteins. (D) Kinase-dependent Ft clustering leads to a change in per-particle r₂ values. Relaxivities measured from prephosphorylated KID-Ft (pKID) mixed with KIX-Ft or from KID-Ft mixed with KIX-Ft in the presence of PKA (middle two bars) are approximately twice the r₂ values measured from KID-Ft/KIX-Ft in the absence of phosphorylation (left bar), or in the presence of the ATP phosphate source but not the kinase (right bar). Corresponding MRI image intensities are shown in the inset.

Figure 3. Nuclear magnetic relaxation dispersion of BM3h variants

(A) Mutant low spin (S = 1/2) BM3h proteins show differing T₁ relaxivities as a function of magnetic field strength [24]. NMRD curves were fit to a Solomon-Bloembergen model equation with an iron proton distance r, water exchange time constant τ_M, and field-independent electronic relaxation time τ_S. Best fit values for five protein variants are shown in the inset, along with corresponding measured relaxivity values (circles) and fitted curves (solid lines), color coded by mutant. Substantial relaxivity variation among mutants is apparent, showing that relaxivity changes are accessible by mutagenesis of the metalloprotein, even without changing the nature of the bound metal complex. (B) Location of amino acid substitutions in the BM3h variants of panel A. Cα positions of mutated residues are denoted by blue balls in the protein backbone trace (gray) [126], with the heme shown in pink and the native fatty acid ligand shown in black. Color-coded dots denote which residues are mutated with respect to the wild type protein in each of the variants listed in panel A. Mutations were selected to alter ligand binding near the heme site and most are clustered in the ligand binding region of the protein.

Figure 4. Structures of lanthanide binding sites

(A) Structure of europium-DOTA, a compound isomorphous to the canonical contrast agent Gd-DOTA [127]. The structure shows coordination of the lanthanide (magenta ball) by carboxylate oxygens and amine nitrogens on the chelator, with a single water molecule (cyan) bound at a Eu–O distance of 2.5 Å. (B) The structure of a polypeptide lanthanide binding tag fused to ubiquitin, in complex with gadolinium (magenta), is surprising similar to the Eu-DOTA structure. The lanthanide is again coordinated by a mixture of oxygen and nitrogen ligands and exhibits a q of 1, with a single inner sphere water molecule bound at a Gd–O distance of 2.9 Å and a second sphere bound water, which may also contribute to r₁, at a distance of 5.4 Å from the lanthanide.