Abstract
The PARK10 locus is associated with idiopathic Parkinson disease (PD), but the responsible gene remains to be identified. Genes associated with familial PD, as well as biochemical evidence from sporadic PD and animal models, have implicated components of the ubiquitin-proteasome system in PD pathogenesis. One attractive candidate gene at the PARK10 locus is RING-Finger Protein 11 (RNF11), the deduced amino acid sequence of which predicts a RING-H2 domain common to E3 ubiquitin ligases such as parkin. To facilitate understanding of this protein and its possible role in PD, we characterized the expression and localization of RNF11 in brain. We detected RNF11 transcript and protein and provided the first direct evidence that RNF11 is expressed in brain. Immunohistochemical analysis of RNF11 protein in rat and human brain, using 2 different antibodies, corroborated the mRNA findings. Both antibodies show that RNF11 is restricted to neurons and excluded from white matter. Moreover, RNF11 is expressed by vulnerable neurons of the substantia nigra and sequestered into Lewy bodies in brains of patients with idiopathic PD. Collectively, these findings identify RNF11 as a strong candidate gene at the PARK10 locus and highlight its potential significance in the development of the common form of PD.
Introduction
Parkinson disease (PD) is a common neurodegenerative disease characterized by progressive impairment of movement, including rigidity, bradykinesia, and resting tremor. The hallmark histopathologic features of PD are degeneration of the dopaminergic cells of the substantia nigra (SN), inclusions comprising α-synuclein, ubiquitin, and other proteins in cell bodies, Lewy bodies (LBs) and process and Lewy neurites (LNs). In addition, there is more widespread α-synuclein pathology that may contribute to a variety of nonmotor symptoms (1).
The etiology of the disease is poorly understood but may involve the interaction of genetic and environmental factors (2). The vast majority of cases of PD are sporadic; however, investigations of the rare familial cases of PD has led to the identification of 6 genes involved in PD pathogenesis: α-synuclein (SNCA), parkin, UCHL1, DJ-1, PINK1, and LRRK2 (3-9). Linkage studies provide evidence for genetic susceptibility to the more common, sporadic form of PD. One such study identified the PARK10 locus by its association with increased risk of idiopathic PD in Icelanders (10). The PARK10 locus spans a 9-megabase region on chromosome 1 containing more than 30 genes. Aberrant expression of 1 or a combination of these genes may underlie susceptibility to PD pathogenesis. The PARK10 locus has also been highlighted in other reports on PD genetics: association of the PARK10 locus with modulation of age at onset (11); single nucleotide polymorphisms associating withdisease in 2 PARK10 genes, HIVEP3 and LOC200008 (12, 13); and decreased mRNA from the PARK10 gene RING-Finger Protein 11 (RNF11) in PD brains (14). Although the genetic alteration responsible for the linkage remains unknown, these reports emphasize the potential significance of PARK10 genetics to sporadic PD. Understanding the role of the responsible gene at the PARK10 locus will depend not only on its identification but also on information about gene-product function and expression in sporadic PD.
Interestingly, 1 PARK10 candidate gene, RNF11, encodes a putative E3 ubiquitin-protein ligase with sequence similarity to that of parkin (15). Mutations in the ubiquitin-protein ligase parkin confer a rare autosomal recessive, juvenile-onset form of PD (4). Genetic evidence from this and other rare familial forms of PD and biochemical evidence from sporadic PD and animal models suggest that dysfunction of the ubiquitin-proteasome system (UPS) may be a common pathogenic insult in PD (16). Both RNF11 and parkin contain RING domains common to E3 ubiquitin ligases (Fig. 1). RNF11 is a protein of only 154 amino acids and contains 1 RING domain, whereas parkin is a larger protein of 465 amino acids and has 2 RING domains. Within the RING domain, the 8-residue motif including C-C-C-H-H/C-C-C-C is crucial for zinc coordination and functional domain structure. The RING domain is thought to be necessary for recognition of the E2 ubiquitin-conjugating enzyme and proper coordination of ubiquitin transfer to the substrate (17). The similarity between RNF11 and parkin with respect to structure and function in the UPS inspired us to study RNF11 as a biologically intriguing candidate gene at the PARK10 locus.
RNF11 contains a RING domain common to E3 ubiquitin-protein ligases such as parkin. RNF11 is a 154-amino acid protein that contains 1 RING-H2 domain. Parkin contains 2 RING domains that are similar to the RING domain of RNF11. Sequence comparison illustrates the similarity among the RING domains of RNF11 and parkin. The C-C-C-H-H/C-C-C-C motif is a defining feature of RING domains necessary for zinc ion binding. Myr, putative myristoylation site; PY, PPPY motif; Ub-like, ubiquitin-like domain; IBR, in-between RING domain.
A strong candidate gene for regulating susceptibility to PD would not only have relevant biologic function but also be expressed in the brain and, possibly, in neurons most vulnerable to PD pathology. In this manner, a genetic alteration that changes expression or function of the gene could directly increase a cell's susceptibility to PD pathology. In addition, the strongest candidate gene might encode a protein that has altered expression and/or localization in PD. In many neurodegenerative diseases, including PD, Alzheimer disease (AD), and Huntington disease, proteins encoded by genes linked to the disease aggregate in hallmark pathologic structures (18). By extension, sequestration of a PARK10 candidate-gene product in LBs of idiopathic PD would suggest that the protein might play a direct role in LB formation and PD pathogenesis.
Here, we report on the expression and localization of RNF11 in the brain as a step toward better understanding of the neurobiology of this PARK10 candidate gene product and its possible role in PD. We detected RNF11 transcript and protein in brains of experimental animals and humans, providing the first direct evidence that RNF11 is expressed in the brain. Moreover, RNF11 was expressed in vulnerable neurons and sequestered into LBs in brains of patients with idiopathic PD. Collectively, these findings identify RNF11 as a strong candidate gene at the PARK10 locus and highlight its potential significance in the development of the common, sporadic form of PD.
Materials and Methods
Case Material
Human brain tissues used in this study were derived from 27 autopsy brains, which were pathologically evaluated for AD and PD. The neuropathologic diagnosis of definite AD was made according to criteria of the Consortium to Establish a Registry for Alzheimer's Disease (19), and National Institute on Aging (NIA)-Reagan AD likelihood was high for all AD cases except 1 case with combined AD and Lewy pathology, which was classified as intermediate. The neuropathologic diagnosis of PD was based on the presence of nigral degeneration and LBs. Control cases had no clinical history or neuropathologic diagnosis of neurologic disease. Four groups were compared: 5 subjects aged 66 to 79 years (mean = 72) with clinically and pathologically confirmed PD, 5 subjects aged 62 to 92 years (mean = 77) with clinically and pathologically confirmed AD, 8 subjects aged 60 to 86 years (mean = 76) with a clinical diagnosis of AD (n = 4) or dementia with LBs (n = 4) and neuropathologic findings of AD with concomitant Lewy pathology, and 9 control subjects aged 52 to 75 years (mean = 65). All animal tissues were obtained from male Sprague-Dawley rats, and all animals used in this study were housed and killed in accordance with guidelines of the Emory University Institutional Animal Care and Use Committee.
RNF11 Antibody Generation
Peptides RNF11-1 (KSPTSDDISLLHESQ) and RNF11-2 (SDRASFGEGTEPDQEPPPPY) were synthesized and together injected into 2 rabbits for polyclonal serum generation (Qbiogene, Illkirch, France). The sera from both rabbits were checked for antigen recognition by ELISA (QBiogene) and immunoblot. Only 1 rabbit showed positive immunoreactivity to either antigen with a strong preference for peptide 1. Serum from this rabbit was affinity purified on a column of peptide 1, and the purified polyclonal serum was named RNF11-1. Two additional rabbits were immunized with a full-length glutathione S-transferase (GST)-RNF11 fusion protein (Covance, Denver, PA). Sera from both animals were tested by immunoblot, and only 1 rabbit of this pair showed immunoreactivity for the GST-RNF11 antigen. Serum from this rabbit was affinity purified on a column of GST-RNF11 protein, and this purified polyclonal serum was named RNF11-FP. Both RNF11-1 and RNF11-FP antibodies were used for all experiments and both demonstrated similar patterns of immunoreactivity.
Plasmids and Immunocytochemistry
RNF11 plasmids used were constructed with the Gateway System including pDEST26-RNF11 and pLenti6-RNF11V5 (Invitrogen, Carlsbad, CA). A bicistronic vector was created with pLenti6-RNF11 by adding in IRES-green fluorescent protein (GFP) following the RNF11 sequence. RNF11 was cloned into pcDNA3.1+ by polymerase chain reaction with use of Kpn1 + Not1 restriction sites. Cells were cultured, transfected, and processed as described previously (20) with only minor deviations. For our studies, HEK and COS7 cells were grown in Dulbecco's modified Eagle's medium (Cambrex, Walkersville, MD) supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY) and 1% penicillin-streptomycin (Cambrex). Cells were plated on coverslips, and 24 to 48 hours after transfection the cells were fixed, blocked, and incubated in primary antibodies overnight (Na+/K+-ATPase-1 subunit mouse monoclonal [1:500; Upstate, Millipore, Billerica, MA], mouse monoclonal anti-V5tag [1:1000; Invitrogen], and RNF11-1 or RNF11-FP at 1 μg/mL). The cells were rinsed and incubated with fluorophore-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA), rinsed, mounted with Vectashield (Vector Laboratories, Burlingame, CA) and scanned on a Zeiss (Thornwood, NY) LSM 510 laser scanning confocal microscope. All images were captured with a 1-μm optical thickness. For preadsorption control studies, primary antibodies were preincubated with 100x excess antigen (or nonspecific peptide) for 1 hour at room temperature before incubation with the cells. Additionally, preimmune serum from each rabbit was tested to confirm specificity of the antiserum immunoreactivity for RNF11.
Immunohistochemistry: Human Paraffin-Embedded Tissue
Eight-micrometer-thick paraffin-embedded sections of cingulate cortex, striatum, hippocampus, and midbrain were deparaffinized, microwaved in citrate (0.01 M, pH 6) for 10 minutes, allowed to cool, and rinsed, and endogenous peroxidase activity was blocked with 3% hydrogen peroxide. Sections were then incubated with normal goat serum followed by primary antibody (2 μg/mL RNF11-1 or 5 μg/mL RNF11-FP overnight at 4°C), biotinylated goat secondary antibody (30 minutes at 37°C), and avidin-biotin peroxidase complex (60 minutes at 37°C; ABC Elite Kit, Vector Laboratories). The chromagen used for color development was 3,3′-diaminobenzidine, and sections were counterstained with hematoxylin.
Free-Floating Rat and Human Tissue
Tissue was processed as described previously (21) for double-label fluorescence and with modifications for single-label light microscopy. RNF11-1 primary antibody was used at 0.5 μg/mL, and RNF11-FP primary antibody was used at 1 μg/mL for both fluorescent and light microscopy. Mouse monoclonal antibody to α-synuclein (kind gift from Dr. Virginia Lee, University of Pennsylvania, Philadelphia, PA) was used at 1:10,000 (22). Mouse monoclonal antibody to the dopamine transporter was used at 1:12,000 and has been extensively characterized elsewhere (23). After overnight primary incubation, sections were rinsed and incubated in biotinylated goat anti-rabbit secondary antibody (1:200, Vector Laboratories) at room temperature for 1 hour, rinsed, incubated with 3,3′-diaminobenzidine, and mounted on slides for light microscopy. For fluorescent microscopy, fluorophore-conjugated secondary antibodies were used, and for human tissue, secondary antibodies were followed by autofluorescence eliminator reagent (Chemicon, Millipore). For preadsorption studies, primary antibodies were preincubated with 100x excess antigen (or nonspecific peptide) for 1 hour at room temperature before incubation with the tissue. Images were captured using a Zeiss LSM 510 laser scanning confocal microscope, and all images were captured with a 1-μm optical thickness.
Immunoblot
Protein extracts from cultured cells and homogenates of human frontal cortex were separated by sodium dodecyl sulfate-polyacrylamide and transferred overnight to Immobilon-P membranes (Millipore). Blots were blocked at room temperature for 30 minutes (Tris-buffered saline + blocking buffer; USB, Cleveland, OH) and probed with primary antibodies (RNF11-1 or RNF11-FP used at 1-5 μg/mL or unpurified sera used at 1:1000 in Tris-buffered saline + 0.1% Tween-20 + blocking buffer) overnight at 4°C. The following day, blots were rinsed and incubated with secondary antibodies conjugated to fluorophores (1:10,000; Molecular Probes, Invitrogen) for 1 hour at room temperature. For preadsorption studies, primary antibodies were preincubated with 100× excess antigen (or nonspecific peptide) for 1 hour at room temperature before incubation with the blot. Blot images were captured using an Odyssey Image Station (LI-COR Biosciences, Lincoln, NE).
Northern Blot
mRNA from regions of a human brain was evaluated for RNF11 transcript using a Human Brain Multiple Tissue Northern Blot (Clontech, Palo Alto, CA). The blot was hybridized with RNA probes for β-actin and RNF11 as described by the manufacturer. The RNA probes were generated using a Maxiscript T7 kit as described by the manufacturer (Ambion, Austin, TX) with pDEST26-RNF11 as template DNA. Experiments were repeated twice on separate blots with similar results.
Results
Antibody Characterization and Development
To analyze RNF11 protein expression in the brain, we developed antibodies to the predicted RNF11 sequence and rigorously assessed antibody specificity. For the RNF11-1 antibody, we chose the amino terminus of RNF11 as an antigen (amino acids 6-20) for production of polyclonal rabbit antisera and affinity purification. We also immunized rabbits with a full-length GST-fusion protein of RNF11, and the resulting affinity purified antibody was named RNF11-FP. We tested the specificity of the 2 antibodies for RNF11 in overexpressing cells. In HEK cells transfected with a cDNA construct dictating the expression of both GFP and RNF11 in a bicistronic vector, RNF11-1 labeled only the GFP-positive cells and not untransfected cells (Fig. 2A, B). This immunoreactivity could be completely blocked by preadsorption with antigen. For example, in COS7 cells transfected with V5-tagged RNF11, RNF11-1 immunoreactivity overlapped with V5 antibody staining, but when the RNF11-1 was preadsorbed with antigen, only the V5 staining remained (Fig. 2C, D). The second antibody, RNF11-FP, yielded staining that was also specific for transfected cells and could be preadsorbed with antigen (Fig. 2E, F). The similarity of the staining between the 2 antibodies further supported the specificity of the antibodies for RNF11. RNF11 antibodies recognized overexpressed protein from transfected cells by immunoblot, but neither antibody was able to detect endogenous protein by this method (Fig. 2G, H). Preadsorption of the antibodies with the antigenic peptide or full-length recombinant protein abolished immunoreactivity in all experiments (Figs. 2D, F and data not shown), confirming that the antibodies were highly specific for the RNF11 antigen. Using these antibodies, we set out to determine whether RNF11 protein could be detected in the brain by immunohistochemistry and, if so, to evaluate its distribution in a PD brain compared with control.
RNF11 antibodies were specific for RNF11. (A, B) RNF11-1 specifically recognized RNF11 in HEK293 cells transiently transfected with a bicistronic cDNA construct for expression of RNF11 and green fluorescent protein (GFP). In this field, 5 cells were each outlined in blue by Na+/K+-ATPase immunoreactivity. The transfected cell was identified by green fluorescence protein fluorescence in green (A). This cell was the only cell showing RNF11-1 immunoreactivity in red (B). Scale bar = 0.02 mm. (C, D) Cos7 cells were transiently transfected with RNF11-V5. RNF11-1 immunoreactivity in red overlaps with anti-V5-tag antibody staining in green. This overlap was shown by yellow in the merged image (C). When RNF11-1 was preincubated with antigenic peptide, RNF11-1 staining is blocked and only anti-V5-tag staining in green remained in the merged image (D). Scale bar = 0.01 mm. (E, F) HEK293 cells were transiently transfected with untagged RNF11. RNF11-FP antibody immunoreactivity (red) detected RNF11 in transfected cells (E), and this staining was completely blocked by preincubating the antibody with GST-RNF11 fusion protein (F). Scale bar = 0.005 mm. (G, H) Both RNF11-1 (1 μg/mL, G) and RNF11-FP (1:1,000 unpure sera, H) were able to detect overexpressed RNF11 by immunoblot, but neither detected endogenous RNF11 in the brain. His-tagged RNF11 had an apparent molecular mass of 25 kDa, and the molecular mass of endogenous RNF11 would be expected to be between 15 and 25 kDa.
RNF11 Expression in the Brain
To investigate the expression of RNF11 in the brain, we first sought to detect RNF11 mRNA and determine its distribution and level of expression in different brain regions. In a human brain, we detected a single transcript of just over 2.4 kilobases (kb) on Northern blots with a radiolabeled RNA probe for RNF11 (Fig. 3A). Our Northern analysis showed that RNF11 was expressed at different levels in several human brain regions. A high signal intensity was detected in the amygdala, moderate intensity in the hippocampus and thalamus, and low intensity in the caudate with respect to β-actin. The corpus callosum exhibited extremely low levels of RNF11 mRNA, which highlighted a major difference in RNF11 distribution between gray and white matter.
RNF11 was widely expressed in human and rat brains (A) Using a radiolabeled RNA probe for RNF11 we detected a single band of just >2.4 kb in a human brain. To verify equal amounts of RNA loaded in each lane we detected a β-actin band just <2.4 kb. (B, C) Coronal and sagittal sections of a rat brain immunolabeled with RNF11-1 antibody showed that RNF11 was expressed by many neuronal populations but was excluded from white mater.
The distribution of RNF11 protein as visualized by immunohistochemistry in a rat brain corresponded well to the regional distribution of mRNA detected by Northern blot. For example, RNF11 immunoreactivity was found widely throughout many regions, and like the transcript, it was largely excluded from white matter (Fig. 3B, C). Areas with some of the most intense RNF11 immunoreactivity included the hypothalamus, cortex, amygdala, CA2 region of the hippocampus, habenula, globus pallidus, and superior colliculus. Thus, RNF11 was expressed in a brain with discrete regions of enrichment and low expression in white matter.
Further analysis of RNF11 expression indicated that RNF11 appeared restricted to neurons, but expression levels varied among neuronal populations. Figure 4 illustrates RNF11 distribution in several regions of interest from a rat brain including cingulate cortex, striatum (caudate-putamen), hippocampus, and amygdala. In cortex, RNF11 was expressed by many neurons, with most striking immunoreactivity in the apical dendrites of pyramidal cells (Fig. 4A). In contrast, RNF11 was expressed at low levels in striatum (Fig. 4B). This finding of low levels of RNF11 in striatum was consistent with our mRNA analysis (Fig. 3A). There was an obvious contrast between RNF11 expression in the striatum and globus pallidus (Fig. 4B). Although RNF11 immunoreactivity was almost absent in the striatum, it was very intense in cells and processes in the globus pallidus. In the hippocampus, RNF11 immunoreactivity was widely expressed by many neurons including intense expression in the pyramidal cells. We found noticeable enrichment in the dendrites of CA2 pyramidal cells (Fig. 4C). Also in agreement with the mRNA analysis, intense RNF11 immunoreactivity was found in amygdaloid neurons (Fig. 4D). In the amygdala, RNF11 immunoreactivity was distributed more uniformly throughout the neurons compared with the pyramidal cells of the cortex, which had increased RNF11 signal in the apical dendrites (Fig. 4A inset; D). The RNF11 immunoreactivity was abolished by preadsorption of the RNF11 antibodies with the antigenic peptide or full-length recombinant RNF11 (not shown). Taken together, these studies show that RNF11 was expressed by a wide variety of neuronal populations and within these neurons, RNF11 protein was enriched in the somatodendritic compartments.
RNF11 localized to neurons in a rat brain. (A) Cingulate cortex of rat showed intense neuronal RNF11 immunoreactivity that was very pronounced in somatodendritic compartments of pyramidal cells (inset). Scale bar = 0.2 mm; (inset) 0.05 mm. (B) Basal ganglia structures globus pallidus (GP) and striatum (caudate-putamen) in rat had very different RNF11 expression patterns. Scale bar = 0.2 mm. (C) Rat hippocampal neurons expressed RNF11; especially noticeable were the pyramidal cell layer and the dendritic label in CA2. Scale bar = 0.2 mm. (D) Rat central nucleus of the amygdala expressed high levels of RNF11. Scale bar = 0.5 mm.
To test whether RNF11 was expressed by the dopaminergic cells in the SN, we first used double-label immunohistochemistry in a rat brain. We identified the dopaminergic cells in the rat SN by dopamine transporter (DAT) immunoreactivity (Fig. 5B). In rat SN there was extensive overlap of RNF11 and DAT immunoreactivity. Low-power images illustrated that RNF11 was present in both dopaminergic and nondopaminergic cells of the rat SN (Fig. 5A-C). At higher power, the yellow coloring indicated the overlapping expression of RNF11 and DAT in individual cells and processes (Fig. 5D). There were many more RNF11-positive neurons and processes in the SN than DAT-positive; however, the vast majority of DAT-containing cells had easily detectable levels of RNF11. Therefore, the dopaminergic cells of the SN were among the neuronal populations that expressed RNF11 in rat.
RNF11 was expressed by dopaminergic cells of substantia nigra (SN) in rat. Double-label immunofluorescence in SN of a rat brain showed that RNF11 in green (A) was expressed by dopaminergic cells as identified by an antibody to the dopamine transporter (DAT) in red (B). The merged image (C) showed that RNF11 was in both dopaminergic and nondopaminergic cells. A higher-power image (D) showed RNF11 and DAT clearly expressed in the same cell bodies and in some processes in the rat SN. Processes containing both DAT and RNF11 are indicated with arrowheads. Scale bars = 0.02 mm.
A human brain has regional and cellular distributions of RNF11 immunoreactivity that are nearly identical to those of a rat brain. RNF11 was restricted to neurons, enriched in somatodendritic compartments, and excluded from white matter. Cortical pyramidal neurons exemplified this pattern of staining, with high expression levels and somatodendritic localization of RNF11 immunoreactivity (Fig. 6A). As predicted by regional mRNA levels (Fig. 3A) and much like a rat brain, human caudate, putamen, and corpus callosum have very little RNF11 immunoreactivity, whereas amygdala, hippocampus, and thalamus have strong RNF11 labeling (not shown). In human SN, the dopaminergic cells can be identified by the presence of neuromelanin. RNF11 was present in cell bodies and processes of dopaminergic and nondopaminergic cells of the SN (Fig. 6C). Control experiments using preimmune serum or preadsorbing the antibodies with antigen highlight the specificity of the antibodies for RNF11 by abolishing all immunoreactivity (Fig. 6B, D). These data show that RNF11 was expressed by a wide population of neurons in normal rat and human brains, including the neurons most vulnerable to degeneration and Lewy pathology in PD.
RNF11 was expressed by vulnerable neurons in a human brain. In a human brain, RNF11 was present in cortical pyramidal neurons (A), and this immunoreactivity was specific for RNF11 antigen as no staining was found when preimmune serum was substituted for RNF11 antibody (B). In human substantia nigra, dopaminergic cells were indicated by the brownish-black pigment, neuromelanin. RNF11 immunoreactivity was indicated by reddish-brown deposits in the cells (C). All immunoreactivity could be blocked by preadsorption with antigenic peptide; thus, only the brownish-black neuromelanin remained (D). Scale bars = 0.05 mm.
RNF11 Distribution in a PD Brain
Because many proteins genetically linked to PD, such as α-synuclein and parkin, have altered expression and/or localization in a PD brain, we tested the hypothesis that this might also be true of the PARK10 candidate, RNF11. We used RNF11 immunohistochemistry to characterize RNF11 expression in 13 postmortem human brains with PD pathology (5 PD only and 8 AD + Lewy pathology) as well as 5 brains with AD pathology (AD only) and 9 pathologically normal controls. Analysis of the overall expression pattern indicated that in most cell populations, the general distribution and level of RNF11 expression was not changed in an AD or a PD brain. We detected no cell populations with visibly increased or decreased expression in diseased brains and the general pattern of neuronal, somatodendritic RNF11 expression was not different among the cases. However, in regions with Lewy pathology, such as cingulate cortex and SN, we found that RNF11 was sequestered in LBs (Figs. 7, 8). RNF11 immunoreactivity was not restricted to any part of the inclusions, and we detected RNF11 in the LBs of cells with single (Fig. 7A, C, H-J) as well as multiple inclusions (Figs. 7B, E-G, 8D-F). To verify that these inclusions were α-synuclein-positive LBs, we performed double-label immunofluorescence on SN from cases with PD pathology and control cases. The inclusions labeled with RNF11 antibodies also labeled with α-synuclein antibodies and RNF11 showed nearly complete overlap with α-synuclein immunoreactivity within the LB (Fig. 7E-J). In addition to the hallmark nigral LB, we detected RNF11 staining in other α-synuclein pathology. As shown in Figure 8, RNF11 staining in PD cortex was quite similar to the typical, neuronal, somatodendritic staining found in a control brain; however, we found that RNF11 immunoreactivity also strongly labeled cortical LBs. In SN, RNF11 labels LNs identified by accumulation of α-synuclein immunoreactivity in neuronal processes (Fig. 8D-F). We detected RNF11-positive LBs in the cingulate cortex or SN of 4 of 5 cases of PD and 6 of 8 cases with combined AD and Lewy pathology. Although nearly every case with Lewy pathology had RNF11-positive LBs and LNs, α-synuclein identified much more pathology than did RNF11. The LB staining was due to specific interaction of the RNF11 antibodies, because both antibodies localized RNF11 to the LBs, all LB immunoreactivity was completely abolished by preadsorption (Fig. 7D), and no inclusions were found in AD or control brains. Taken together, our findings suggest that RNF11 is sequestered in α-synuclein-positive LBs and LNs in a PD brain.
RNF11 was sequestered in Lewy bodies (LBs). RNF11 immunoreactivity localized to a subset of LBs in a Parkinson disease (PD) brain. (A-C) Representative RNF11-positive LBs in the substantia nigra (SN) of 3 PD brains are marked by arrows. (D) Shows that all RNF11 immunoreactivity, including the LB label, was completely blocked by preadsorption with the antigenic peptide. The remaining unlabeled LB was marked by an arrowhead and the remaining dark color in the cell was due to neuromelanin. (E-J) RNF11 colocalized with α-synuclein in LBs of substantia nigra. Red indicates α-synuclein, green RNF11, and (G) and (J) are merged where red and green overlap in yellow. Scale bar = 0.05 mm.
RNF11 was found in cortical Lewy bodies (LBs) and Lewy neurites (LNs). RNF11 antibodies labeled a subset of cortical LBs. (A-C) For example in cingulate cortex from a Parkinson disease (PD) brain, a cortical LB was labeled with α-synuclein antibody (red, A), and, in addition to labeling neuronal cell bodies and processes, RNF11 antibodies labeled the LB (green, B). The merged image shows overlap between the α-synuclein and RNF11 antibodies in yellow (C). Scale bars = 0.02 mm. (D-F) In substantia nigra (SN), RNF11 labeled LBs and LNs. Antibodies for α-synuclein labeled LBs and LNs in SN (red). RNF11 labeled cell bodies and processes including LBs and LNs (green). The merged images shows overlap between the 2 antibodies (yellow). LBs and LNs labeled with both α-synuclein and RNF11 are indicated with arrows. Scale bars = 0.05 mm.
Discussion
Antibody Characterization and Development
We were successful in generating 2 polyclonal antibodies against distinct epitopes of RNF11. Antibody specificity was rigorously tested by: 1) identification of overexpressing cells by immunocytochemistry (Fig. 2); 2) preadsorption of the antibodies with antigenic peptides or full-length fusion proteins or substitution of preimmune serum, abolished immunoreactivity in all assays (Figs. 2, 6, 7); and 3) identification of bands of predicted size by immunoblot of proteins from cells overexpressing RNF11 (Fig. 2). Further confirming the specificity, both antibodies show identical staining patterns despite recognizing distinct RNF11 epitopes. The first antibody, RNF11-1, was raised to a unique, 15-amino acid sequence near the amino terminus of the protein, whereas the second antibody, RNF11-FP, was raised to a full-length GST fusion protein. Although purified recombinant GST-RNF11 abolished staining when preincubated with either antibody, the 15-amino acid peptide (RNF11-1) only abolished immunoreactivity of the RNF11-1 antibody, indicating that the RNF11-FP antibody recognized a distinct epitope (data not shown).
The RNF11 antibodies are an excellent tool for identifying RNF11 by immunohistochemistry and immunocytochemistry; however, despite our extensive experience generating antibodies to low-abundance nervous system proteins, neither antibody detects endogenous RNF11 in a brain by immunoblot (23-28). This result is most likely explained by the low sensitivity of the antibodies to denatured RNF11 protein or the low abundance of RNF11 in brain homogenates. The lack of detection of endogenous RNF11 by immunoblot could make interpretation of immunohistochemistry difficult; however, RNF11 expression in a brain is supported by the shared immunohistochemical results with 2 different RNF11 antibodies and the concordance of protein localization with the regional mRNA analysis. Taken together, the specificity and convergence of these 2 antibodies indicates that the immunoreactivity found in cells and brain is specific for the RNF11 protein.
RNF11 Expression in a Normal Brain
Our analysis of RNF11 is the first characterization of this novel PARK10 candidate in a brain. We show that RNF11 mRNA is detectable in several regions of a human brain as a band just >2.4 kb (Fig. 3A). This result is consistent with the report by Seki et al (15), in which they first cloned mouse RNF11 and reported a 2.4-kb transcript in several mouse tissues. This regional mRNA analysis suggests that RNF11 is variably expressed in different brain regions with high expression in limbic and cortical areas and low expression in striatum and corpus callosum. These findings support our immunohistochemical analysis of RNF11, which shows similar areas of enrichment (e.g. amygdala and cortex) and relative exclusion (e.g. striatum and white matter). Within these regions, RNF11 expression is limited to neurons and restricted to somatodendritic compartments (Figs. 4–6).
A strong candidate gene for the PARK10 linkage might be expressed by the most vulnerable dopaminergic neurons of the SN. Our analysis indicates that RNF11 is widely expressed by many neuronal populations including both the dopaminergic and nondopaminergic cells of the SN. This distribution is consistent with other genes currently linked to PD. Parkin, α-synuclein (SNCA), UCHL1, synphilin, and DJ-1 are all expressed in many cell types throughout the brain including, but not restricted to, the vulnerable neurons of SN (29-32). Despite their ubiquitous distribution, these PD-associated proteins could affect the dopaminergic neurons of the SN selectively, whether as neurotoxicants or as neuroprotectants (33-35). Similarly, we hypothesize that alterations in RNF11, although universally expressed, could confer susceptibility to PD pathogenesis in the SN dopaminergic neurons. These neurons are known to be selectively vulnerable to systemic toxicants, and similarly, could be particularly sensitive to systemic alterations in protein expression (35-38). Therefore, although widely expressed, a particular protein could be more important for SN dopaminergic cell survival than other cell types because of the specific sensitivities of these cells.
RNF11 Expression in a PD Brain
In a PD brain, we find the overall pattern of RNF11 expression to be similar to that of a control brain. However, our method focused on localization, rather than quantitative levels of expression, and it is possible that there was a general alteration in RNF11 levels in PD. Evidence for such dysregulation of RNF11 has come from work by Noureddine et al (14) that identified a 4-fold decrease in RNF11 mRNA in PD SN compared with controls. However, our data show that RNF11 expression is restricted to neurons, and, therefore, because of their choice of regional analysis, it is difficult to interpret whether the decrease in RNF11 mRNA is due to loss of SN neurons in the PD sample compared with the control. In our analysis, the most striking difference in RNF11 localization in a PD brain is its sequestration into cytoplasmic LB inclusions, the pathologic hallmark of PD. Many proteins linked to PD have been found in LBs, including α-synuclein, parkin, synphilin, and UCHL1 (39-44). However, there is some disagreement in the literature with some reports of these proteins labeling many LBs, and others reporting few or no LB labels (45). In general, it seems that many of the proteins linked to PD are present in at least some LB inclusions, with α-synuclein remaining as the most sensitive label of LB (46). This altered localization suggests, but does not clearly indicate, a role for RNF11 in the formation of LBs or in altering the susceptibility of SN DA neurons to insults. Further experiments focused on the roles of RNF11 in protein processing and aggregation are needed to establish connections between RNF11 function, development of LB pathology, and cell vulnerability. Taken together, the report of decreased RNF11 mRNA along with our localization data support a role of RNF11 dysregulation in PD pathogenesis.
RNF11 as a PARK10 Candidate
In this study, we have investigated RNF11 as a candidate gene for conferring susceptibility to idiopathic PD as identified by the PARK10 locus. Given that recent evidence suggests that UPS dysfunction may be a common mechanism underlying PD pathogenesis, we chose to examine whether RNF11 expression and localization in normal and PD brains support its identification as a PARK10 candidate gene. Our studies provide the first evidence that RNF11 mRNA and protein are expressed in neurons, that RNF11 is highly enriched in the cell bodies and dendrites of select neurons, including those vulnerable in PD, and that RNF11 is localized to LBs and LNs-the major pathologic lesions seen in a PD brain. Taken together, these findings indicate that RNF11 stands out as an intriguing candidate gene that has a plausible biologic function in PD pathogenesis.
Many cellular pathways have been implicated in PD pathogenesis including apoptosis, oxidative stress, growth factor signaling, and, most recently, the UPS. Much current research has focused on the UPS with the identification of mutations in UPS members UCHL1 and parkin that cause familial PD. Also, recent studies suggest that proteasome inhibition can model PD symptoms in cells and rodents, leading to an accumulation of intracellular proteins resembling the hallmark LBs of PD (38, 47). RNF11 may also be a member of the UPS system that, like parkin, plays a role in UPS function. Although little is known about RNF11 function, recent data suggest that it facilitates ubiquitination of substrates involved in mediating glial-derived neurotrophic factor signaling (48). Previous work indicates that the overexpression of RNF11 may lead to potentiation of growth factor signals and tumorigenesis (49, 50). Our localization of RNF11 to somatodendritic compartments is consistent with a proposed role in postsynaptic growth factor receptor signaling. It is plausible that, in neurons, dysregulation of a gene whose product functions in the UPS system and mediates growth factor signaling may leave vulnerable cells susceptible to PD pathogenesis. Although our data, which localize RNF11 to vulnerable cells, support the possibility of such a role for RNF11, further characterization of RNF11 cellular function is needed to strengthen this hypothesis, together with definitive identification of the responsible gene(s) for PD susceptibility at the PARK10 locus.
Acknowledgments
The authors thank Robert Baul, Veronica Walker, Craig Heilman, and Dr. Howard Rees for excellent technical assistance.
References
Author notes
This work was supported by the National Institutes of Health Collaborative Center For Parkinson's Disease Environmental Research (U54 ES012068 to AIL) and National Institute on Aging Grant P50AG025688.
