Both homozygous (L166P, M26I, deletion) and heterozygous mutations (D149A, A104T) in the DJ-1 gene have been identified in Parkinson's disease (PD) patients. The biochemical function and subcellular localization of DJ-1 protein have not been clarified. To date the localization of DJ-1 protein has largely been described in studies over-expressing tagged DJ-1 protein in vitro. It is not known whether the subcellular localization of over-expressed DJ-1 protein is identical to that of endogenously expressed DJ-1 protein both in vitro and in vivo. To clarify the subcellular localization and function of DJ-1, we generated three highly specific antibodies to DJ-1 protein and investigated the subcellular localization of endogenous DJ-1 protein in both mouse brain tissues and human neuroblastoma cells. We have found that DJ-1 is widely distributed and is highly expressed in the brain. By cell fractionation and immunogold electron microscopy, we have identified an endogenous pool of DJ-1 in mitochondrial matrix and inter-membrane space. To further investigate whether pathogenic mutations might prevent the distribution of DJ-1 to mitochondria, we generated human neuroblastoma cells stably transfected with wild-type (WT) or mutant (M26I, L166P, A104T, D149A) DJ-1 and performed mitochondrial fractionation and confocal co-localization imaging studies. When compared with WT and other mutants, L166P mutant exhibits largely reduced protein level. However, the pathogenic mutations do not alter the distribution of DJ-1 to mitochondria. Thus, DJ-1 is an integral mitochondrial protein that may have important functions in regulating mitochondrial physiology. Our findings of DJ-1's mitochondrial localization may have important implications for understanding the pathogenesis of PD.

No abstract

Aquaporins are a family of water channels found in animals, plants, and microorganisms. A subfamily of aquaporins, the aquaglyceroporins, are permeable for water as well as certain solutes such as glycerol, lactate, and urea. Here we show that the brain contains two isoforms of AQP9--an aquaglyceroporin with a particularly broad substrate specificity--and that the more prevalent of these isoforms is expressed in brain mitochondria. The mitochondrial AQP9 isoform is detected as an approximately 25 kDa band in immunoblots. This isoform is likely to correspond to a new AQP9 mRNA that is obtained by alternative splicing and has a shorter ORF than the liver isoform. Subfractionation experiments and high-resolution immunogold analyses revealed that this novel AQP9 isoform is enriched in mitochondrial inner membranes. AQP9 immunopositive mitochondria occurred in astrocytes throughout the brain and in a subpopulation of neurons in the substantia nigra, ventral tegmental area, and arcuate nucleus. In the latter structures, the AQP9 immunopositive mitochondria were located in neurons that were also immunopositive for tyrosine hydroxylase, as demonstrated by double labeling immunogold electron microscopy. Our findings suggest that mitochondrial AQP9 is a hallmark of astrocytes and midbrain dopaminergic neurons. In physiological conditions, the flux of lactate and other metabolites through AQP9 may confer an advantage by allowing the mitochondria to adjust to the metabolic status of the extramitochondrial cytoplasm. We hypothesize that the complement of mitochondrial AQP9 in dopaminergic neurons may relate to the vulnerability of these neurons in Parkinson's disease.

The cellular concentration of phosphate, the main activator of phosphate activated glutaminase (PAG) is rather constant in brain and kidney. The enzyme activity, however, is modulated by a variety of compounds affecting the binding of phosphate, such as glutamate, calcium, certain long chain fatty acids, fatty acyl CoA derivatives, members of the tricarboxylic acid cycle and protons (Kvamme et al. [2000] Neurochem. Res. 25:1407-1419). Therefore, the kinetic and allosteric properties of the enzyme are essential for regulating the enzyme activity in situ, especially because the enzymically active pool of PAG is assumed to have an external localization in the inner mitochondrial membrane, being exposed to cytosolic variation in the content of effectors. This has largely been overlooked. A hypothetical model for the allosteric interactions based on the sequential induced fit allosteric model by Koshland et al. ([1966] Biochemistry 5:365-385) is presented. Furthermore, it has been generally accepted that there exist only two isoforms of PAG, the kidney PAG that is similar to brain PAG, and the liver PAG. Therefore, the immunoreactivity of brain cells against kidney PAG antibodies has been considered a measure of PAG protein. Gomez-Fabre et al. ([2000] Biochem. J. 345:365-375) recently found, however, that a PAG mRNA from human breast cancer ZR75 cells is present in human brain and liver, but not in the kidney. We observed only traces of PAG immunoreactivity in cultured astrocytes and cultured neuroblastoma cells, regardless whether antibodies against the C- and N-termini of kidney PAG or antibodies against liver PAG were used, but considerable enzyme activity, demonstrating hitherto unknown isoforms of PAG (Torgner et al. [2001] FEBS Lett. 268(Suppl 1):PS2-031).

No abstract