(a) Iron deposition in substantia nigra of rotenone-treated (A) and control rats (D) (20×). Iron deposits are indicated by dark blue reaction product (arrow-heads in panel A mark iron deposits; arrow shows dendritic arborization of dopamine neurons, with moderate loss of processes after rotenone treatment). The two central panels B, C are higher magnification images (40×) showing co-localization of the iron deposits and TH+ neurons. (b) Iron deposition in the substantia nigra of rotenone-treated (A) and control (C) monkeys (5×). Neuromelanin-containing neurons have light brown appearance. The central panel B is a higher magnification detail (40×) of a rotenone-treated monkey brain, where iron deposits can be seen inside and outside the neuromelanin containing neurons. (c) Colocalization of iron in both neurons and microglia. Panel A (40×) shows iron deposition within neuronal profiles (arrows) in a rotenone-treated rat. The section was also labeled for tyrosine hydroxylase (brown)as a maker for dopamine neurons. Although the iron-laden neurons do not stain for tyrosine hydroxylase, they are approximately the same size as the nearby dopamine neurons, indicated by asterisks. Panel B shows a higher magnification (100×) of an iron-laden neuron. Note the TH associated chromogen deposits (arrows).In panel C (40×),a degenerating iron-laden blue neuronal profile is shown (arrow). The degenerating neuron is surrounded by activated microglia (brown), which were identified by OX-42 immunoreactivity. At a later stage (panel D, 60×), punctate iron (arrows) is seen within a cluster of reactive microglia. (d) Panel A shows a high magnification image (100×) of a substantia nigra monkey neuron containing neuromelanin (brown) with diffuse iron deposition (blue). In panels B–D (100×), tyrosine hydroxylase positive neurons (brown) show blue iron deposition in a mitochondrial-like punctate pattern. (e) Iron deposits in microglia in monkey tissue (A, C 20×; B, D 100×). Tissue was stained for iron (blue) and microglia (brown) using an anti-CD68 antibody. Panels A and B show low- and high-power views, respectively, of iron and microglia in a control animal; note the melanized neuron in B. Panels C and D show iron and microglia in a rotenone-treated animal. There is activation of microglia and many of the microglia contain large iron deposits.

Rotenone treatment induces thiol oxidation in substantia nigra dopamine neurons in rats (a) and monkeys (b). As neurons accumulate thiol oxidation, they lose tyrosine hydroxylase immunoreactivity. Neurons with extensive thiol oxidation and reduced tyrosine hydroxylase immunoreactivity are indicated by arrows. (c) Rotenone treatment is associated with the formation of an intermolecular disulfide crosslink between Tf and another protein. Tf from the mitochondrial fraction of substantia nigra was subjected to SDS-PAGE with or without the thiol reducing agent, dithiothreitol (+DTT or −DTT; green band); protein loading was normalized to the 39 kDa subunit of complex I (red band). Under non-reducing conditions (−DTT), specimens from rotenone-treated animals (R) had a highMWband (large arrow-head)in addition to the normal Tf band (small arrow). The upper band was not seen in control specimens (C) and disappeared under reducing conditions (+DTT). (d) Thiol oxidation of Tf is revealed by FRET. Substantia nigra sections were triple labeled for tyrosine hydroxylase (not shown), Tf and oxidized thiols. In sections from rotenone-treated rats, there was FRET between the fluorescent indicator of oxidized thiols (anilino-naphtalene sulfonate maleimide) and Tf, indicating thiol oxidation of Tf. (e) Identification and quantification of the transferrin peptide KPVDQYEDC*YLAR. The ICAT survey showed that this peptide is increased in the samples from rotenone-treated animals (inset). Because the ICAT reagents label oxidized thiols (heavy ICAT was used for rotenone, light ICAT for control samples), the data indicate higher oxidation of cys260 following rotenone administration. The MS/MS spectrum is a fully tryptic sequence with reasonable matching scores by SEQUEST algorithm (Xcorr=2.3 and ΔCn=0.34). The b and y ions are indicated in the picture. (f) In vitro oxidation of Tf through ROS production causes reductive release of ferrous iron (Fe²⁺) from Tf (left graph). Release of Fe²⁺ from purified Tf(Fe³⁺)₂ was assessed by following the absorbance of the Fe²⁺ chelator, bathophenanthroline disulfonate. Under basal conditions, Tf did not release Fe²⁺. Oxidation of thiols (by formation of a protein-glutathione mixed disulfide; Tf-GSSG) is associated with reduction (Fe³⁺ to Fe²⁺) and release of iron from Tf (right graph). This reductive release of iron is prevented if thiol groups are alkylated with NEM before incubation with GSSG; alkylation with NEM per se did not induce iron release (Supplementary Fig. 5). Values are expressed as percent of total iron capacity; 100% corresponds to the signal detected after – at the end of each experiment – total iron release from Tf was induced through acidification of the solution.

(a) The putative mitochondrial targeting sequence (MTS) of TfR2 targets red fluorescent protein (RFP) to mitochondria. Top row, HEK 293 cells were co-transfected with cyan fluorescent protein (CFP) ligated to a known mitochondrial targeting sequence (MTS-CFP) and with RFP. The MTS-CFP protein labeled mitochondria, where as RFP was seen diffusely in the cytosol. A merged image reveals no colocalization of the MTS-CFP and RFP signals. Bottom row, when RFP was ligated to the putative MTS of TfR2 (TfR2-RFP), RFP was directed to mitochondria rather than cytosol, and the merged image showed extensive colocalization with MTS-CFP. The signal of MTS-CFP is represented in the pseudo-color green to better illustrate the yellow-colored overlapping regions in the merge. (b) In dopamine neurons from control animals (top row), there was relatively little expression of Tf and TfR2, but there was substantial colocalization (arrows) – and there was FRET between the two proteins, indicating a physical interaction (molecular proximity) of the proteins. After rotenone treatment (bottom row), there was an increase in the levels of both proteins and the amount of FRET between them also increased. Images from control and rotenone-treated animals were collected using identical parameters. As a result, in the control animals (top row), the Tf and TfR2 signals appear artificially low, so as to avoid saturation of the images from rotenone-treated animals.

(a) As in the rat brain, TfR2 is localized to dopamine neurons in the human brain and it colocalizes, in part, with mitochondria (arrow-heads). (b) There is a marked increase in Tf in substantia nigra dopamine neurons – which contain neuromelanin (brightfield) – and in the surrounding neuropil in Parkinson's disease compared to controls. (c) Quantification of TH and Tf fluorescence intensity in PD cases (n=6) and controls (n=4) reveals that a statistically significant decrease in TH immunoreactivity is paralleled by a significant increase in Tf levels. (d) In Parkinson's disease, there is thiol oxidation of Tf, as indicated by FRET between oxidized thiols (anilino-naphtalene sulfonate maleimide) and Tf (see Fig. 3). (e) In the dopamine neurons, some of the Tf is colocalized with mitochondria. Human substantia nigra sections were triple labeled for tyrosine hydroxylase (not shown), mitochondria and Tf (arrow-heads).

(a) Rotenone treatment induces a selective increase in Tf in rat dopaminergic neurons. Those neurons with the greatest amount of Tf accumulation (arrows) tend to have the greatest loss of tyrosine hydroxylase (TH) immunoreactivity. (b) Immunoblotting of Tf in striatal and substantia nigra cytosolic and mitochondrial fractions from control and rotenone-treated rats. With rotenone treatment, there was a marked increase in Tf, particularly in mitochondria from substantia nigra (mitochondrial marker=39 kDa subunit of the respiratory complex I). (c) Immuno-electron microscopy shows that both Tf (top row) and TfR2 (bottom row) are associated with substantia nigra mitochondria. (d) Subfractionation of mitochondria reveals Tf and TfR2 associated with the inner mitochondrial membrane (IM) along with the 39 kDa subunit of complex I. As expected, cytochrome c (cyt C) is associated with both the IM and the intermembrane space (IMS); Mn-superoxide dismutase (MnSOD) is found in matrix (MTX) and, to a lesser extent, in the IMS. (e) Progressive trypsinization of isolated mitochondria shows that Tf and TfR2 are located in a trypsin resistant compartment, like the inner membrane protein, the 39 kDa subunit of complex I. In contrast, an outer membrane protein, Tom20, is digested by moderate concentrations of trypsin. (f) In human SK-N-MC neural cells, rotenone (50 nM for 3 days) induced an increase in Tf and colocalization with a mitochondrial marker. (g) Similar to rotenone-treated rats, monkeys treated with rotenone selectively accumulated Tf, and there was a tendency for TH immunoreactivity in neurons with heavy Tf accumulation to be reduced. (h) In monkeys, rotenone treatment induced a Tf increase in mitochondria of dopaminergic neurons (panels a and b). Heavy mitochondrial localization of Tf was detectable only in those neurons with high Tf levels (arrow-heads; mitochondrial marker=mitochondrial Heat Shock Protein 60). As noted above, TH levels decreased in Tf-accumulating neurons (panel b).

(a) TfR2 is colocalized with tyrosine hydroxylase (TH) in dopamine neurons of the substantia nigra, as assessed by laser scanning confocal microscopy. (b) Dopaminergic neurons of the substantia nigra pars compacta (SNc) stain intensely for TfR2, while neurons in the adjacent substantia nigra pars reticulata (SNr) contain little, if any, TfR2. Note the clear cytoplasm (arrows) surrounding neuronal nuclei in SNr, which is in marked contrast to the pattern in SNc. There is also substantial TfR2 immunoreactivity in the neuropil of SNr.

In cells expressing TfR2, extracellular Tf and iron are taken up and targeted to mitochondria. (a) HEK 293 cells were co-transfected with TfR2 and MTS-CFP (see Fig. 4)and loaded with the mitochondrial potentiometric dye, TMRM. All cells were visualized with TMRM, and only transfected cells were identified by MTS-CFP expression. When fluorescently labeled Tf was applied extracellularly to the cells, it was rapidly taken up and targeted to mitochondria only in cells expressing TfR2 (arrows). Only small amount of Tf, which did not colocalize with mitochondria, were detectable in non transfected cells (arrow-heads). (b) When cells were transfected with TfR2 (at 50% efficiency), there was a marked, significant increase in mitochondrial ⁵⁹[Fe³⁺] uptake, whereas the increase in cytosolic iron was less pronounced. In this experiment, transfected cells were exposed to ⁵⁹[Fe³⁺]₂-Tf for 1 h, washed, then incubated for 2 h before fractionation. (c) A portion of the ⁵⁹[Fe³⁺] that enters mitochondria is incorporated into complex I, and this is significantly higher inTfR2-transfected cells. In contrast to the experiment shown in b, these dividing cells were incubated for 16 h prior to fractionation and immunoprecipitation. (d) Over-expression of TfR2 in SK-N-MC neuroblastoma cells reduced the toxicity of rotenone, but not H₂O₂ or staurosporine (STS). A shRNA, which effectively reduces TfR2 protein expression (not shown), blocked this protective effect. The results represented in the graphs are the average of n=3 independent replicates.