Modeling Midbrain and Brainstem Neuromelanins to Characterize Metal Binding and Associated MRI Contrast in Parkinson's and Alzheimer's Diseases

Abstract

Neuromelanin (NM) is a dark pigment that binds potentially toxic metal ions and is crucial for neuronal vulnerability. Magnetic resonance imaging (MRI) was proposed to measure neuromelanin in the substantia nigra or locus coeruleus, potentially providing a marker of Parkinson's disease. Here, synthetic neuromelanin analogues were prepared with iron and copper and used for characterization of metal binding and impact on proton relaxation, a prerequisite for optimizing neuromelanin-sensitive MRI. The results confirm the presence of paramagnetic mononuclear Fe(III) and antiferromagnetically coupled clusters, which enhance relaxation to variable degrees. Further complexity arises from Cu(II), which can compete for binding to mononuclear sites, aggregate in mixed-metal clusters, or bind to proteins associated with the melanin moiety. Unlike the strong relaxant Fe(III), Cu(II) only indirectly impacts relaxation by replacing iron. Overall, MRI primarily provides measures of average neuromelanin concentrations. Information on the distribution of neuromelanins with different metal compositions might be obtained with multiparametric MRI.

Keywords: Bioinorganic chemistry; Copper; Iron; Magnetic resonance; Neuromelanin.

Figure 1

Characterization of the metal content by ICP‐OES and EPR in melanin–protein conjugates. Blue symbols indicate PheoβLG and red symbols EuβLG samples. a) Total Cu content a _Cu plotted versus total Fe content a _Fe (both in µmol g⁻¹). Mixed‐metal pheomelanins contained only small amounts of Cu (0.036 µmol mg⁻¹ ≤ a _Cu ≤ 0.091 µmol mg⁻¹), whereas a _Fe varied over a large range (0.091 µmol mg⁻¹ ≤ a _Fe ≤ 0.371 µmol mg⁻¹). The eumelanin powders were classified as samples with large a _Fe variation at low a _Cu (0.088 µmol mg⁻¹ ≤ a _Cu ≤ 0.102 and 0.088 µmol mg⁻¹ ≤ a _Fe ≤ 0.417 µmol mg⁻¹; enclosed by solid line) and samples with large a _Cu variation at low a _Fe (0.102 µmol mg⁻¹ ≤ a _Cu ≤ 1.168 and 0.077 µmol mg⁻¹ ≤ a _Fe ≤ 0.091 µmol mg⁻¹; enclosed by dashed line). b) Variation of the Cu(II) EPR signal area s _Cu at g _e = 2.15 with the Fe(III) signal amplitude s _Fe at g _e = 4.3 (both measured at 10 K and in arbitrary units, a.u.). Linear regression yielded s _Cu = (172.4 ± 7.0) − (1.79 ± 0.15)s _Fe in mixed‐metal pheomelanins (solid blue line). Consistent results in eumelanins were obtained only for high Cu content, whereas samples of low Cu content (enclosed by solid line) showed a distinct deviation from the regression line. c) Fe/Cu ratio, s _Fe/s _Cu, of the EPR‐active metal plotted as a function of a _Fe/a _Cu. A fit to a straight line through the origin yielded slopes of φ = 0.5475 ± 0.0063 and φ = 1.275 ± 0.032 (in a.u.) for PheoβLG‐CuFe (solid blue line) and EuβLG‐CuFe (solid red line), respectively. The samples with the highest iron content, P9 (PheoβLG‐CuFe‐1/10) in b) as well as P9 and E11 (EuβLG‐CuFe‐1/10) in c) were considered outliers and not included in the correlation analyses.

Figure 2

¹H NMR spectra of synthetic melanins in D₂O. Spectra of PheoβLG a and b) and EuβLG conjugates c and d) are arranged according to an increasing relative amount of copper (a and c) or iron (b and d) in the reaction medium. They should be compared to those in Figure S4 of Ferrari et al.,^{[ 46 ]} which also refer to dopamine‐containing melanin‐βLG derivatives.

Figure 3

¹H NMR spectrum of human NM (isolated from substantia nigra tissue) in DMSO‐d ₆.  ¹H NMR spectra of human NM isolated from substantia nigra in the presence or absence of proteinase K were also published by Double et al.^{[ 47 ]}

Figure 4

Characterization of melanin–protein conjugates by EPR recorded at 10 K. The variation of the metal content is indicated by the line color. Scaled X‐band EPR spectra of ten PheoβLG‐CuFe a) and seven EuβLG‐CuFe powders b) are shown as well as lineshape details of the Fe(III) resonance at g = 4.3 c) and the Cu(II) resonance at g = 2.15 d) in exemplarily selected PheoβLG samples. Note that the spectra in a) and b) have been scaled according to the receiver gain and the sample mass to ensure comparability of the recording, whereas the amplitudes of the signals in (c and d) were normalized to 1 for better visibility of the lineshape differences. Main acquisition parameters are summarized in Table S2.

Figure 5

Lineweaver–Burk plots describing the dependencies of the EPR signals measured at 10 K on the metal content. Blue and red symbols indicate pheomelanin and eumelanin conjugates, respectively, as in Figure 1. a) Variation of 1/s _Fe obtained from the EPR signal of Fe(III) at g = 4.3 as a function of a _Cu/a _Fe and results from fits to Eq. S9 for the limiting case of c ₀ a _Fe + c ₂ a _Cu ≫ c ₁ (i.e., conditions of competition and high metal content) shown as solid lines. b) Variation of 1/s _Cu obtained from the EPR signal of Cu(II) at g = 2.15 as a function of a _Fe/a _Cu. A fit to Eq. S11 for the limiting case of c ₀ a _Fe + c ₂ a _Cu ≫ c ₁ describes the experimental result for PheoβLG‐CuFe (solid blue line; sample P9 regarded as outlier) but not in EuβLG‐CuFe samples of low Cu content (enclosed by solid red line). c) Variation of 1/s _Cu as a function of 1/a _Cu. A fit to Eq. S11 for the limiting case of c ₁ + c ₂ a _Cu ≫ c ₀ a _Fe (i.e., negligible competition) approximately describes the experimental result for EuβLG‐CuFe (solid line) but not for PheoβLG‐CuFe.

Figure 6

Characterization of melanin‐protein conjugates by XAS at 10 K. a) Fe K‐edge XANES spectra and b) magnitude of k ²‐weighted Fourier transforms (FT) of the EXAFS for PheoβLG‐Fe‐10%, PheoβLG‐CuFe‐1/10, and Fe₂O₃. c) Cu K‐edge XANES spectra and d) magnitude of the k ²‐weighted FT of the EXAFS for PheoβLG‐Cu‐5%, PheoβLG‐CuFe‐1/10, CuO, Cu₂O, and CuS. The black arrow indicates a shoulder at ∼8.994 keV. Room‐temperature XANES spectra for FeO, FeS₂, FeS, and Fe₃N in a) and for Cu₂S, Cu(NO₃)₂· 3H₂O [here referred to as h‐Cu(NO₃)₂] and Cu₃N in c) were taken from the MDR XAFS database (see Supporting Information). Different temperatures do not affect the XANES but preclude comparing the EXAFS.

Figure 7

Proton relaxation rates and bulk magnetic susceptibility at room temperature in dependence of the metal content. Blue and red symbols indicate pheomelanin and eumelanin conjugates, respectively, as in Figure 1. Variation of a) R ₁ and b) R obtained by monoexponential fitting with the amount of EPR‐active Fe(III) (measured at 10 K). Linear regression in mixed‐metal conjugates (solid lines) yielded R ₁/s⁻¹ = (0.685 ± 0.028) + (0.00201 ± 0.00061)s _Fe and R ₁/s⁻¹ = (0.796 ± 0.042) + (0.00441 ± 0.00085)s _Fe in PheoβLG‐CuFe and EuβLG‐CuFe, respectively, as well as R ₂/s⁻¹ = (2.285 ± 0.066) + (0.0027 ± 0.0012)s _Fe for the combined data from both melanins. Blue and red arrows show the R ₁ or R difference between the metal‐free conjugate (see Table S4) and the extrapolation of the regression line toward s _Fe = 0. c) Negative correlation of the fraction of the slow‐relaxing compartment with the total metal content (here expressed as magneton concentration $\mathcal{p}$). Estimates of f_a in the gray shaded area were considered less reliable and excluded in the analysis. d) Variation of the bulk magnetic susceptibility with the total metal content. The Δχ values reflect the susceptibility difference between the melanin/gel suspensions and the agarose gel doped with (paramagnetic) gadopentetate dimeglumine surrounding the samples (see Figure S6). The resulting negative values, hence, do not mean that the conjugates are diamagnetic.

Figure 8

Model of the variation of R ₁ (at room temperature) with metal content. R ₁ values estimated with Eq. 7 are shown as a function of a _Fe and the Cu/Fe ratio for a) PheoβLG‐CuFe and b) EuβLG‐CuFe as well as c) in comparison with measured relaxation rates. Dashed blue and red lines indicate R _1,0 in the metal‐free preparations. Dotted blue and red lines correspond to R _1,0 + R , where the combined contribution from homonuclear (Cu only or Fe only) and mixed‐metal (Cu‐Fe) clusters, R , was assigned to the extrapolated R ₁ for $s_{\mathrm{Fe}}\to 0$ (Figure 7a) and assumed to be invariant (areas shaded in dark blue or orange. Areas shaded in light blue or orange indicate the PRE due to mononuclear Fe(III). Solid blue and red lines labeled as “0” show the R ₁ variation for negligible copper content. ${\kappa }_{\mathrm{Fe}}{r}_{1,\mathrm{Fe}}^{\mathrm{m}}$ was taken from the slope of the linear‐regression result (Figure 7a) and c ₀ and c ₂ from the Langmuir model (Figure 5a). Further solid lines show R ₁ in mixed‐metal samples (“iso‐Cu contour lines”) for different Cu/Fe ratios of the conjugates. Arrows indicate the decreasing PRE from Fe(III) with increasing copper content. Open diamonds and circles indicate previously obtained metal concentrations in NM extracted from human substantia nigra and locus coeruleus, respectively (Table 2).

Figure 9

Illustration of the assumed metal binding sites in melanin–protein conjugates. a) Schematic representation of the various types of binding sites and metal environments in the melanic (red) and protein (blue) parts of PheoβLG‐CuFe and EuβLG‐CuFe conjugates. b) Structural model of melanin–βLG conjugates showing the pheomelanic and eumelanic components surrounding the protein core and the distribution of various mononuclear and multinuclear metal sites in the melanic and protein components. For obvious reasons the size of the clusters is kept at a minimum.