Pink light on mitochondria in autoimmunity and Parkinson Disease

Abstract

Mitochondrial dysfunction and T cell autoimmunity have been independently implicated in Parkinson Disease pathogenesis. In a recent publication in Cell, Matheoud et al. link them by describing a new mechanism, activated in familial forms of Parkinson Disease, in which mitochondrial proteins are processed for recognition by CD8+ T cells.

Mounting evidence attributes the pathological loss of dopaminergic neurons in Parkinson Disease (PD) to the accumulation of dysfunctional mitochondria, with consequent oxidative stress and death of dopaminergic neurons. Research supporting this notion stems in part from the association of familial PD with mutations in the genes encoding PINK1 and Parkin, which target damaged mitochondria for elimination by mitophagy (Pickrell and Youle, 2015). Alternative evidence suggests that the adaptive immune system, activated by PD-associated inflammation and microglial activation, plays an important role in PD pathogenesis (Mosley et al., 2012). Accordingly, unusually high numbers of T lymphocytes are found in the substantia nigra of post-mortem PD patients (Brochard et al., 2009). Is there a link between these two pathogenic processes? Such a link would require a mechanism by which selected proteins from damaged mitochondria are presented by major histocompatibility complex (MHC) molecules to T cells. In the latest issue of Cell, Matheoud et al. (2016) uncover a pathway for mitochondrial antigen presentation (MitAP), in which mitochondrial-derived vesicles (MDVs) are targeted to endolysosomes for processing and presentation by MHC class I molecules (MHC-I; Figure 1). The pathway is antagonized by PINK1 and Parkin, evoking the provocative potential for MitAP to influence pathogenesis in familial (and perhaps sporadic) PD.

FIGURE 1. Mitochondrial antigen presentation (MitAP) and inhibition by PINK1 and Parkin.

Mitochondria derived vesicles (MDVs) arise from mitochondrial buds in a process requiring RAB9 GTPase and SNX9. RAB7 likely promotes MDV fusion with lysosomes/multivesicular bodies (MVBs) to allow processing of MDV cargo antigens and MitAP. PINK1 and Parkin normally antagonize MitAP, but their absence in familial forms of Parkinson Disease (PD) promotes MitAP, leading to the development of mitochondrial antigen-reactive CD8+ T cells. Dopaminergic neurons upregulate MHC-I molecules in inflammatory conditions, and might activate MitAP in PINK1- or Parkin-deficient PD. Reactive CD8+ T cells might cross the blood brain-barrier, bind to MHC-I:mitochondrial peptide complexes and kill these neurons, leading to PD pathology.

Cytotoxic T cells kill target cells expressing MHC-I bound to specific peptides. Most antigenic peptides are generated when proteasomes degrade newly synthesized proteins in the cytosol. Subsequently the peptides assemble with newly synthesized MHC-I upon translocation into the endoplasmic reticulum. To compare MitAP with classical antigen presentation, the authors targeted variants of herpes simplex virus 1 glycoprotein B (gB) to the mitochondrial matrix (mito-gB), nuclear membrane or cytosol of a macrophage cell line, and assayed for MHC-I presentation to a gB-specific CD8+ T cell hybridoma. Unlike cytosolic or nuclear gB, mito-gB presentation did not require new protein synthesis and was sensitive to lysosomal inhibitors. Whereas some viral antigens are processed in endolysosomes following autophagy (English et al., 2009), mito-gB presentation induced by a mild heat shock was uniquely stimulated, not inhibited, by interfering with general autophagy or by down-regulation of PINK1; conversely, induction of autophagy or mild Parkin overexpression reduced mito-gB presentation. These data indicate that MitAP exploits a novel endolysosomal mechanism that is specifically inhibited by PINK1- and Parkin-dependent mitophagy, and suggest that loss of PINK1 or Parkin function might increase MitAP in PD (Figure 1).

The authors then dissected the mechanism underlying MitAP. Using immunofluorescence microscopy and subcellular fractionation, they showed that heat shock induces mito-gB segregation from mitochondria, partially into low-density MDVs detected in close proximity to late endosomes/lysosomes and partially to the lysosomes themselves. MDVs lacked other mitochondrial cargo and seemed to target mito-gB for MitAP, and likely derived from small double membrane buds seen emerging from mitochondria by electron microscopy. MDV formation, release of mitochondrial buds, and mito-gB presentation all required two known endolysosomal effectors: RAB9 (a small GTPase that regulates late endosome/lysosome function and unconventional autophagy (Nishida et al., 2009)) and sorting nexin 9 (SNX9; a BAR domain-containing protein that regulates clathrin-dependent endocytosis), both of which were recruited to mitochondria upon heat shock (Figure 1). Interestingly, Parkin overexpression resulted in SNX9 degradation, providing a potential mechanism by which PINK1/Parkin-dependent mitophagy down-regulates MitAP and distinguishing the MDVs described here from PINK1/Parkin-dependent MDVs implicated in quality control (McLelland et al., 2014). RAB7 (another small late endosome-associated GTPase) was required later in mito-gB presentation, likely during fusion of MDVs with late endosomes/lysosomes.

Key findings based on the model antigen mito-gB were confirmed in dendritic cells (DCs), macrophages and fibroblasts using an endogenous mitochondrial matrix protein, OGDH, that is targeted by T cells in an autoimmune disease. Importantly, the authors found that splenic DCs isolated from Pink1^−/− and Parkin^−/− mice presented OGDH more readily to specific T cells than wild-type DCs, suggesting a potential role for MitAP in stimulating immune responses in familial PD. Mito-gB or OGDH presentation and mito-gB-containing MDV formation in wild-type cells were stimulated by proinflammatory stimuli in vitro and in vivo, and this effect was exacerbated in Pink1^−/− and Parkin^−/− DCs. Thus, inflammation might activate MitAP in both familial and sporadic forms of PD.

The findings by Matheoud et al. clearly elucidate important new mechanisms for presentation of mitochondrial antigens and for cross-talk between mitochondria and the endolysosomal system. Given that CD4+ T cells mediate PD pathology in animal models (Brochard et al., 2009), it will be important to determine whether this pathway also provides mitochondrial antigens for MHC class II – for which processing occurs primarily within endolysosomes. The observation that Parkin and PINK1 expression antagonizes MitAP suggests that in contrast to macroautophagy, mitophagy represses antigen presentation, and supports the intriguing idea that MitAP is normally restrained to prevent autoimmunity (Baum, 1995). Breaching this restraint thus might unleash autoimmunity, potentially inducing PD. Identifying the mitochondrial peptides presented upon inflammation and the mitochondrial antigen specific T cell repertoire might provide novel insights into how cargo is selected for MitAP and into the autoimmune response, and will provide tools to test whether T cells specific for mitochondrial antigens contribute to PD or other neurodegenerative disorders.

While clearly establishing a precedent for a new antigen processing mechanism, the study raises a number of important questions for future investigation. Most important is whether MitAP functions in activated microglia and/or neurons of the substantia nigra. MHC-I is expressed at low levels by neurons, but is upregulated – particularly in dopaminergic neurons – by proinflammatory factors secreted by activated microglia (Cebrián et al., 2014). Additionally, protein aggregates such as beta-amyloid in Alzheimer disease can activate inflammasomes and induce IL-1β secretion (Halle et al., 2008). Therefore, it will be critical to assess whether MitAP is influenced by sterile stimuli, such as α-synuclein aggregates in PD, and whether IL-1β is secreted by activated microglia and regulates MitAP or modulates mitochondrial antigen-specific T cell responses. Equally important is testing whether mitochondrial antigen-specific T cells infiltrate the CNS. T cell infiltration would require increased permeability of the blood-brain barrier, as observed upon aging or inflammatory conditions, and is proposed by the authors to be at least in part responsible for the death of mitochondrial antigen-presenting dopaminergic neurons. This hypothesis is tantalizing, and might be addressed using mouse models of PD and adoptive transfer of T cells specific for mitochondrial antigens like OGDH or mito-gB.

The findings of Matheoud et al. undoubtedly shed new light into the etiology of mitochondria-specific autoimmune disorders, provide a mechanism that might link autoinflammation and autoimmunity with the onset and/or progression of neurodegenerative disease, and certainly open new venues for exciting future research.