Mitochondrial Transfer from Astrocytes to Neurons following Ischemic Insult: Guilt by Association?

Intercellular mitochondrial transfer has been shown in tumor models, following lung injury and in xenotransplants of leukemic cells, but trafficking between cells in the brain remains unexplored. A suggestion that mitochondria move from astrocytes to neurons in a model of ischemia in a recent article in Nature by Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar should be interpreted with caution. Intercellular mitochondrial transfer has been shown in tumor models, following lung injury and in xenotransplants of leukemic cells, but trafficking between cells in the brain remains unexplored. A suggestion that mitochondria move from astrocytes to neurons in a model of ischemia in a recent article in Nature by Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar should be interpreted with caution. There is widespread consensus that mitochondria of eukaryotic cells originated in an endosymbiotic event over two billion year ago and that their microbial origin is reflected in independent replication and fusion/fission processes within host cells. Less well appreciated is the retention of specialized intercellular trafficking mechanisms that facilitate transfer of cytosolic molecules and organelles between cells of higher organisms. Movement of organelles between mammalian cells in culture via membrane nanotubes or intercellular bridges was first observed by Rustom et al., 2004Rustom A. Saffrich R. Markovic I. Walther P. Gerdes H.H. Science. 2004; 303: 1007-1010Crossref PubMed Scopus (1256) Google Scholar. Subsequently, many reports of intercellular mitochondrial transfer have been documented (Berridge et al., 2016Berridge M.V. McConnell M.J. Grasso C. Bajzikova M. Kovarova J. Neuzil J. Curr. Opin. Genet. Dev. 2016; 38: 75-82Crossref PubMed Scopus (51) Google Scholar), including mitochondrial transfer from a host organism to tumor cells seen in an 11,000-year-old canine venereal tumor (Rebbeck et al., 2011Rebbeck C.A. Leroi A.M. Burt A. Science. 2011; 331: 303Crossref PubMed Scopus (85) Google Scholar). More recently, transfer of mtDNA to tumor cells lacking mtDNA was demonstrated directly in solid tumor models using genetic markers (Tan et al., 2015Tan A.S. Baty J.W. Dong L.F. Bezawork-Geleta A. Endaya B. Goodwin J. Bajzikova M. Kovarova J. Peterka M. Yan B. et al.Cell Metab. 2015; 21: 81-94Abstract Full Text Full Text PDF PubMed Scopus (452) Google Scholar) and in leukemic cells (Moschoi et al., 2016Moschoi R. Imbert V. Nebout M. Chiche J. Mary D. Prebet T. Saland E. Castellano R. Pouyet L. Collette Y. et al.Blood. 2016; 128: 253-264Crossref PubMed Scopus (238) Google Scholar). Other evidence supports mitochondrial transfer from mesenchymal stem cells to injured lung epithelial cells "in vivo" (Ahmad et al., 2014Ahmad T. Mukherjee S. Pattnaik B. Kumar M. Singh S. Kumar M. Rehman R. Tiwari B.K. Jha K.A. Barhanpurkar A.P. et al.EMBO J. 2014; 33: 994-1010PubMed Google Scholar, Islam et al., 2012Islam M.N. Das S.R. Emin M.T. Wei M. Sun L. Westphalen K. Rowlands D.J. Quadri S.K. Bhattacharya S. Bhattacharya J. Nat. Med. 2012; 18: 759-765Crossref PubMed Scopus (943) Google Scholar) and in developing oocytes (Lei and Spradling, 2016Lei L. Spradling A.C. Science. 2016; 352: 95-99Crossref PubMed Scopus (159) Google Scholar). In another study, retinal ganglion axons at the optic nerve head packaged defective mitochondria into vesicles and passed them to neighboring astrocytes for degradation, a process referred to as transmitophagy (Davis et al., 2014Davis C.H. Kim K.Y. Bushong E.A. Mills E.A. Boassa D. Shih T. Kinebuchi M. Phan S. Zhou Y. Bihlmeyer N.A. et al.Proc. Natl. Acad. Sci. USA. 2014; 111: 9633-9638Crossref PubMed Scopus (356) Google Scholar). A recent report now proposes mitochondrial transfer from astrocytes to injured neurons after stroke as a recovery mechanism (Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar). In this study, Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar found that unlike the cell contact-dependent membrane nanotubes and intercellular bridges often observed in vitro, generation of extracellular mitochondria involved vesicles or particles generated by primary cortical astrocytes in culture or in vivo. Mitochondria in vesicles from astrocyte-conditioned medium (ACM) were visualized with a membrane potential-dependent fluorescent dye, MitoTracker Red CMXRos. When mitochondria-containing ACM was added to cultured neurons injured by oxygen-glucose deprivation, fluorescent particles were observed associating with neurons. The ACM addition resulted in small increases in bioenergetic parameters such as cell-associated ATP and metabolic activity measured by NAD(P)H-dependent reduction of the water-soluble tetrazolium dye (WST) across the plasma membrane. Extracellular mitochondrial release increased following activation of CD38, an ADP-ribose cyclase enzyme involved in calcium mobilization, and CD38 knockdown using siRNA suggested that extracellular mitochondria could be important for trophic support of neurons by astrocytes. Blood vessel occlusion created an ischemic lesion in the cerebral cortex, modeling stroke. Gap43, a marker of potential neuroplasticity, was induced in injured cortex and induction was lost with CD38 knockdown, suggesting a potential role for CD38, and perhaps mitochondrial particles, in neuroplasticity. While CD38-dependent improvements in neural performance and recovery from stroke are shown, it remains to be determined whether these effects are related to mitochondrial transfer into relevant neurons. To directly examine this, purified CMXRos-labeled particles were infused into the ischemic lesion and 24 hr later some could be found in close association with neurons. While Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar clearly demonstrate that CD38 activity is important for astrocytic support of neuronal functions, the mitochondrial transfer from astrocytes to neurons will need corroboration from future studies. When used at high concentrations, the fluorescent dye has a tendency to damage mitochondrial networks and to leak out of labeled cells, forming fluorescent particles in suspension. Future long-term studies in mice using genetic approaches to track and confirm mitochondrial transfer in vivo will circumvent the toxicity issues of mitochondrial dyes. Another important caveat is the extent to which neuron-associated ACM vesicles containing mitochondria are merely cell associated rather than intracellular. In both cell culture and in vivo approaches, the results could equally well be explained by ACM vesicles associating with the surface of injured neurons, potentially through β-integrin or CD63, both of which are expressed on the majority of ACM particles. The confocal microscopy data presented by Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar are supported by our own experience with neonatal astrocyte-neuron co-culture, both of which clearly demonstrate that neuron-associated fluorescent particles are most often on the outside of the cell. If ACM particles adhere to the surface without directly transferring mitochondria into neurons, functional effects of metabolic parameters could be explained, since ACM particles themselves are metabolically active and could release or transfer small molecules (Figures 1C and 1E ). To our knowledge, vesicles containing mitochondria have not been shown to directly release mitochondria into the cytoplasm of adjacent cells (Figure 1D), but gap junction formation between vesicle and cell would allow metabolite and small molecule entry (Figure 1E). If particles containing mitochondria undergo conventional endocytosis by neurons, endocytic pathways of vesicle destruction would likely be invoked (Figure 1B), so release of functional mitochondria from endosomes would be unlikely. Reverse transfer of damaged mitochondria packaged in vesicles, from neurons to astrocytes for degradation via the endocytic pathway, is also shown in Figure 1A (Davis et al., 2014Davis C.H. Kim K.Y. Bushong E.A. Mills E.A. Boassa D. Shih T. Kinebuchi M. Phan S. Zhou Y. Bihlmeyer N.A. et al.Proc. Natl. Acad. Sci. USA. 2014; 111: 9633-9638Crossref PubMed Scopus (356) Google Scholar). Whether transfer of functional mitochondria from astrocytes to neurons occurs following ischemic injury remains debatable. If astrocytic mitochondria can be shown unequivocally to be intracellular within damaged neurons, it is possible that bioenergetic support to the damaged neuron could be provided. However, mechanistic explanations of how these mitochondria are released into the cytoplasm, or used to provide bioenergetic and metabolic support for damaged neurons, remains a challenging issue. It is possible that transfer of replication-competent functional mitochondria into recipient cells with damaged mitochondria in vivo requires membrane nanotube connections or intercellular bridges with cytoskeletal structures to direct movement, as observed in cell culture systems (Rustom et al., 2004Rustom A. Saffrich R. Markovic I. Walther P. Gerdes H.H. Science. 2004; 303: 1007-1010Crossref PubMed Scopus (1256) Google Scholar), whereas transfer of discrete membrane-bound particles is likely to result in mitochondrial degradation or recycling. In this context, we have established an orthotopic brain tumor model in which tumor growth occurs only after acquisition of mitochondria from adjacent host cells in the brain, similar to the solid tumor models mentioned above (Tan et al., 2015Tan A.S. Baty J.W. Dong L.F. Bezawork-Geleta A. Endaya B. Goodwin J. Bajzikova M. Kovarova J. Peterka M. Yan B. et al.Cell Metab. 2015; 21: 81-94Abstract Full Text Full Text PDF PubMed Scopus (452) Google Scholar). In summary, while Hayakawa et al., 2016Hayakawa K. Esposito E. Wang X. Terasaki Y. Liu Y. Xing C. Ji X. Lo E.H. Nature. 2016; 535: 551-555Crossref PubMed Scopus (629) Google Scholar have demonstrated that vesicles containing mitochondria are produced by astrocytes and that these vesicles become associated with damaged neurons both in vitro and in vivo, whether these neuron-associated mitochondria are intracellular and functional remains contentious. Thus, while donor cells can transfer and establish functional mitochondria into recipient cells with damaged mitochondrial DNA in vivo, debate will continue over the location of these mitochondria following brain injury, the cellular source of these mitochondria in the brain, and the mechanism of transfer.