Molecular Markers for Granulovacuolar Degeneration Are Present in Rimmed Vacuoles

Background Rimmed vacuoles (RVs) are round-oval cytoplasmic inclusions, detected in muscle cells of patients with myopathies, such as inclusion body myositis (IBM) and distal myopathy with RVs (DMRV). Granulovacuolar degeneration (GVD) bodies are spherical vacuoles containing argentophilic and hematoxyphilic granules, and are one of the pathological hallmarks commonly found in hippocampal pyramidal neurons of patients with aging-related neurodegenerative diseases, such as Alzheimer's disease and Parkinson's disease. These diseases are common in the elderly and share some pathological features. Therefore, we hypothesized that mechanisms of vacuolar formation in RVs and GVD bodies are common despite their role in two differing pathologies. We explored the components of RVs by immunohistochemistry, using antibodies for GVD markers. Methods Subjects included one AD case, eight cases of sporadic IBM, and three cases of DMRV. We compared immunoreactivity and staining patterns for GVD markers. These markers included: (1) tau-modifying proteins (caspase 3, cyclin-dependent kinase 5 [CDK5], casein kinase 1δ [CK1δ], and c-jun N-terminal kinase [JNK]), (2) lipid raft-associated materials (annexin 2, leucine-rich repeat kinase 2 [LRRK2], and flotillin-1), and (3) other markers (charged multi-vesicular body protein 2B [CHMP2B] and phosphorylated transactive response DNA binding protein-43 [pTDP43]) in both GVD bodies and RVs. Furthermore, we performed double staining of each GVD marker with pTDP43 to verify the co-localization. Results GVD markers, including lipid raft-associated proteins and tau kinases, were detected in RVs. CHMP2B, pTDP43, caspase 3, LRRK2, annexin 2 and flotillin-1 were detected on the rim and were diffusely distributed in the cytoplasm of RV-positive fibers. CDK5, CK1δ and JNK were detected only on the rim. In double staining experiments, all GVD markers colocalized with pTDP43 in RVs. Conclusions These results suggest that RVs of muscle cells and GVD bodies of neurons share a number of molecules, such as raft-related proteins and tau-modifying proteins.

In addition to the accumulation of Ab and tau in both hippocampal neurons and muscle cells, these autophagic vacuoles, RVs and GVD bodies show immunopositivity for both CDK5 [29] and pTDP43 [12,30,31]. These findings may suggest the existence of a common pathway in the formation of autophagic vacuole in the different organs and diseases [32]. However, studies testing this hypothesis have not been performed thus far. Therefore, in the current study, we explored the compositional similarities between RVs and GVD bodies by immunohistochemistry using antibodies for known GVD markers.

Ethics Statement
The protocols for neuropathological procedures and analyses were approved by and performed under the guidelines of the    ethics committee of Hiroshima University, Graduate School of Biomedical and Health Sciences. Samples were obtained with the understanding and written informed consent of patients except for AD cases. AD samples were obtained with the understanding and written informed consent of family members. For this study, all samples were coded and personal information was dissociated from the test results. All data were analyzed anonymously, and all neuropathological procedures and analyses were conducted according to the principles expressed in the Declaration of Helsinki.

Sample collection and tissue specimens
Hippocampal tissue specimens were obtained by autopsy from one AD case. Muscle tissue specimens were retrieved by biopsy from eight cases of s-IBM and three cases of DMRV formation (Table 1). In accordance with the National Institute on Aging-Alzheimer's Association guidelines for neuropathological assessment of AD, quantitative neuropathological criteria for diagnosis of the 73-year-old male with AD were fulfilled. For a high level of AD neuropathological change, these criteria indicated: (1) A3: an Ab phase of 5, (2) B3: a neurofibrillary tangle (NFT) stage of 6, (3) C3: a score of 'frequent' in accordance with the Consortium to Establish a Registry for AD [33]. Within 24 hours of death, the hippocampus was extracted during autopsy, and fixed in 10% (v/ v) formalin for 3 weeks. Paraffin-embedded sections (thickness 7 mm) were then prepared for subsequent procedures. Eight cases of s-IBM fulfilled the clinical and histopathological diagnostic criteria as well as no family history of IBM, as previously described [34]. In all three cases of DMRV, an UDP-N-acetylglucosamine 2-epimerase/N-acetylmannosamine kinase (GNE) gene mutation was confirmed by direct sequencing. Muscle specimens were extracted through biopsy, immediately frozen in isopentane cooled with liquid nitrogen, followed by the preparation of section (thickness 7 mm).

Antibodies
The antibodies used for immunohistochemistry and immunofluorescence studies are listed in Table 2. Two independent antibodies for the identification for each protein were used whenever possible to confirm the results from both staining techniques (anti-caspase 3, anti-CK1d, anti-JNK, anti-annexin2, anti-flotillin-1 and anti-pTDP43 antibodies).

Immunohistochemistry using 3, 3-diaminobenzidine (DAB)
Hippocampal sections were deparaffinized and dehydrated. Muscle sections were subjected to antigen retrieval (microwaveheated in distilled water for 10 min) followed by washes in phosphate-buffered saline (PBS) for 3 min. Deparaffinized sections were then incubated in 3% H 2 O 2 /PBS for 60 min to eliminate endogenous peroxidase activity in tissues. Each section was incubated with primary antibodies overnight at 4uC. The sections were then washed three times in PBS and incubated with horseradish peroxidase (HRP)-conjugated anti-mouse or antirabbit antibodies for 30 min at room temperature. After sections were washed (three times in PBS), they were incubated at room temperature with DAB (Dako, Glostrup, Denmark).

Immunofluorescence staining
Double staining on sections was performed, including on those of muscle containing RVs for further characterization. The same primary antibodies as listed in Table 1 were used in combination with rabbit or mouse antibody for pTDP43. For anti-CK1d, antiflotillin-1 and anti-pTDP43 primary antibodies, the following secondary antibodies were used: Alexa Fluor 488 donkey antimouse IgG or Alexa Fluor 546 donkey anti-rabbit IgG (all 1:1000) (Molecular Probes, Eugene, OR, USA). The other primary antibodies with anti-pTDP43 were detected using the same HRP-conjugated secondary antibodies as described above, together with the TSA TM KIT #12: Alexa Fluor 488 donkey, and TSA TM KIT #4: Alexa Fluor 568 donkey (Invitrogen, Eugene, OR, USA). Slides were mounted with Vectashield (Vector Laboratories, Burlingame, CA) and observed under the LSM510 confocal laser scanning microscope (Carl Zeiss AG, Oberkochen, Germany).

Identification of RVs
Modified Gomori trichrome staining with anti-CHMP2B antibody was performed to verify RVs on serial sections prior to immunohistochemical staining.

CHMP2B-positive vacuoles corresponded to RVs in muscles from patients with s-IBM and DMRV
Modified Gomori trichrome staining, and immunohistochemical staining of serial sections using anti-CHMP2B antibody, were performed. Gomori trichrome staining revealed RVs, which was confirmed by anti-CHMP2B (Fig. 1). The anti-CHMP2B antibody in DMRV and s-IBM cases revealed RVs and sarcolemma (Fig. 2).

RVs were immunopositive for GVD markers
We next compared the immunoreactivity of RVs in the hippocampus of an AD case with those of muscle tissue, using antibodies against reported GVD markers (tau-modifying proteins, lipid raft-associated materials, CHMP2B and pTDP43). GVD bodies were immunopositive for all antibodies and mainly found in the CA1 subregion of the hippocampus (Fig. 3). In pyramidal cells, caspase 3, CDK5, JNK, annexin 2, LRRK2, flotillin-1 and pTDP43 were diffusely distributed in the cytoplasm, forming fine spherical granules, as well as GVD bodies (Fig. 3). NFTs were immunopositive for caspase 3, CDK5, LRRK2, annexin 2, flotillin-1 and pTDP43. RVs were mainly found in degenerated, atrophic muscle fibers in all s-IBM and DMRV cases (Fig. 4). Some vacuoles contained granules resembling intraluminal vesicles of GVD, both of which were immunopositive for all of markers listed in Table 2. CHMP2B, caspase 3 and pTDP43 were diffusely distributed in the cytoplasm of RV-positive fiber as well as on the rim (Fig. 4). Moreover, CHMP2B and caspase 3 were detected in sarcolemma of most muscle fibers (Fig. 4). CDK5, CK1d and JNK were only detected on the rim and scarcely present in other regions of RVpositive fibers (Fig. 4). LRRK2, annexin 2 and flotillin-1 were present on the rim and diffusely distributed in the cytoplasm of RV-positive fibers (Fig. 4).
Furthermore, the rims of some vacuole-containing nuclei were immunopositive for all of these markers, but the nuclei themselves were not immunostained. Infiltrating inflammatory cells were not stained (Fig. 4).

All GVD markers colocalized with pTDP43 in RVs
Because pTDP43 is present in RVs [12,30,31] and GVD bodies [35], we therefore performed double immunofluorescence staining using anti-pTDP43 antibody together with antibodies for CHMP2B, caspase 3, CDK5, CK1d, JNK, LRRK2, annexin 2 and flotillin-1 in s-IBM cases. Results showed that pTDP43 colocalized with all of these markers in RVs (Fig. 5). These results suggested that numerous GVD markers were present in RVs.

Discussion
The present study reveals a number of GVD markers present in RVs, namely raft-related proteins and tau-modifying proteins. These findings suggest that RVs are similar in nature to GVD bodies, and that GVD bodies and RVs partly exhibit a common origin.
The lipid raft is a subdomain of the plasma membrane containing high concentrations of cholesterol and glycosphingolipids, thus providing a platform for cell signaling [36]. Phosphatidylinositol 4,5-bisphosphate (PtdIns [4,5]P2) is another lipid component of the lipid raft [37], and colocalizes with proteins, such as flotillin-1 and annexin 2 (both lipid raft markers) [38][39][40]. Recently, lipid rafts have been reported to play a role in endocytic pathways independent of the clathrin pathway [41]. Furthermore, lipid raft vacuoles have been shown to be transported to the autophagosome via the ESCRT pathway [42,43]. The raft-dependent pathway is also involved in the formation of intracellular Ab deposits [44][45][46]. Several lines of evidence show that intracellular Ab is significantly more toxic than extracellular Ab in disrupting synaptic activity and facilitating the hyperphosphorylation of tau in the AD brain [46]. We have previously reported that GVD bodies are immunopositive for PtdIns(4,5)P2, flotillin-1, annexin2 and LRRK2 [40], indicating that GVD bodies are autophagic vacuoles with a raft component. In the current study, we demonstrated that these raft markers were also evident in RVs. In addition to the presence of raft-related proteins in GVD bodies, immunoreactivity of RVs for raft markers suggest that RVs are raft-dependent structures like GVD bodies. MAP tau is phosphorylated by a number of kinases, such as CK1 [19], GSK3b [20], JNK [21], and cleaved by caspase 3 [47]. These tau-modifying proteins are present in GVD bodies. In contrast, CDK5, another tau kinase, is present in RVs [9]. Recently, we have reported that CDK5 is a novel marker for GVD bodies [22]. In the present study, a number of tau-modifying proteins (i.e. GVD markers) were present in RVs. In addition to the presence of CDK5 in GVD bodies, immunoreactivity of RVs for other tau-modifying proteins suggests RVs are structures related to tau modification like GVD bodies.
We and others have previously shown that CHMP2B is present in GVD bodies [27,28], thus suggesting that GVD bodies may be generated as multivesicular bodies in the endocytic pathway. Morphological observations via electron microscopy suggest that GVD bodies are an age-related and unique form of autophagic vacuole, and that intraluminal vesicles are residues of the autophagic process [17]. RVs are also reported to be derived from autophagic vacuoles [13]. For example, use of LC3 antibodies indicated that RVs were involved in autophagosome formation [8]. Because both GVD bodies and RVs are vacuoles likely to be derived from autophagic mechanisms, they may thus share similar structural components. In the current study, we revealed that markers of GVD bodies: CHMP2B, lipid raftassociated proteins and tau-modifying proteins, were present in RVs. Some of these markers were only localized in RVs, and others were diffusely distributed in the cytoplasm or sarcolemma as well as in RVs. Although the immunoreactivity of GVD markers for RVs and GVD bodies in this study did not differ significantly, differences between RVs and GVD bodies have been previously reported. RV formation may be related to early-stage autophagic organelles as shown in their immunoreactivity for LC3 at this time-point [8]. Conversely, GVD bodies may be predominantly related to late-stage autophagic organelles, evidenced by their immunoreactivity for LAMP1 and cathepsin D, and weak signals for early-stage autophagic markers [28]. Further study is required to determine the materials associated with the autophagy and endosome pathways, as well as the tau phosphorylation pathway, to ascertain the differences in components between RVs and GVD bodies.
Immunohistochemistry for selected proteins may have been a limitation of the present study. An attempt at a comprehensive analysis on the composition of both structures using imaging mass spectrometry was unsuccessful because of limited resolution. Further advances in technology would therefore enable us to perform extensive analyses of RVs and GVD bodies in the future.

Conclusions
Our results suggest that RVs of muscle cells and GVD bodies of neurons have a number of common molecules, such as raft-related proteins and tau-modifying proteins. Further investigations will elucidate more precisely the mechanisms by which vacuolar pathological structures are generated, including aging-related changes.