MitoSOX

Example 2

The next experiments asked whether inhibition of the same set of FXN-RFs would also upregulate transcription of the TRE-FXN gene in post-mitotic neurons, which is the cell type most relevant to FA. To derive post-mitotic FA neurons, FA(GM23404) iPSCs were stably transduced with lentiviral vectors over-expressing Neurogenin-1 and Neurogenin-2 to drive neuronal differentiation, according to published methods (Busskamp et al. 2014, Mol Syst Biol 10:760); for convenience, these cells are referred to herein as FA neurons. Neuronal differentiation was assessed and confirmed by staining with the neuronal marker TUJ1 (FIG. 2A). As expected, the FA neurons were post-mitotic as evidenced by the lack of the mitotic marker phosphorylated histone H3 (FIG. 2B). Treatment of FA neurons with an shRNA targeting any one of the 10 FXN-RFs upregulated TRE-FXN transcription (FIG. 2C) and increased frataxin (FIG. 2D) to levels comparable to that of normal neurons. Likewise, treatment of FA neurons with small molecule FXN-RF inhibitors also upregulated TRE-FXN transcription (FIG. 2E) and increased frataxin (FIG. 2F) to levels comparable to that of normal neurons.

It was next determined whether shRNA-mediated inhibition of FXN-RFs could ameliorate two of the characteristic mitochondrial defects of FA neurons: (1) increased levels of reactive oxygen species (ROS), and (2) decreased oxygen consumption. To assay for mitochondrial dysfunction, FA neurons an FXN-RF shRNA or treated with a small molecule FXN-RF inhibitor were stained with MitoSOX, (an indicator of mitochondrial superoxide levels, or ROS-generating mitochondria) followed by FACS analysis. FIG. 3A shows that FA neurons expressing an NS shRNA accumulated increased mitochondrial ROS production compared to EZH2- or HDAC5-knockdown FA neurons. FIG. 3B shows that FA neurons had increased levels of mitochondrial ROS production compared to normal neurons (Codazzi et al., (2016) Hum Mol Genet 25(22): 4847-485). Notably, inhibition of FXN-RFs in FA neurons restored mitochondrial ROS production to levels comparable to that observed in normal neurons. In the second set of experiments, mitochondrial oxygen consumption, which is related to ATP production, was measured using an Agilent Seahorse XF Analyzer (Divakaruni et al., (2014) Methods Enzymol 547:309-54). FIG. 3C shows that oxygen consumption in FA neurons was ˜60% of the level observed in normal neurons. Notably, inhibition of FXN-RFs in FA neurons restored oxygen consumption to levels comparable to that observed in normal neurons. Collectively, these preliminary results provide important proof-of-concept that inhibition of FXN-RFs can ameliorate the mitochondrial defects of FA post-mitotic neurons.

Mitochondrial dysfunction results in reduced levels of several mitochondrial Fe-S proteins, such as aconitase 2 (ACO2), iron-sulfur cluster assembly enzyme (ISCU) and NADH:ubiquinone oxidoreductase core subunit S3 (NDUFS3), and lipoic acid-containing proteins, such as pyruvate dehydrogenase (PDH) and 2-oxoglutarate dehydrogenase (OGDH), as well as elevated levels of mitochondria superoxide dismutase (SOD2) (Urrutia et al., (2014) Front Pharmacol 5:38). Immunoblot analysis is performed using methods known in the art to determine whether treatment with an FXN-RF shRNA or a small molecule FXN-RF inhibitor restores the normal levels of these mitochondrial proteins in FA neurons.

Example 8

Cells were treated with troglitazone, followed by incubation with the various fluorescence dyes (CALCEIN AM, MBBR, MITOSOX™ and CYTOSOX™). For each non-control well, the quadratic chi-distance distance to positive and negative control templates (for each fluorescence channel) was calculated and normalized with the scaling factor. The normalized data are reorganized into tensors, and the distance from positive and negative control templates are calculated. The similarities of the trioglitazone tensor to another compound tensor can be computed using known techniques by comparing the fiber columns of the tensors using dynamic time warping distance (see, Giorgino et al. “Computing and Visualizing Dynamic Time Warping Alignments in R: The dtw Package.” Journal of Statistical Software, 31(7), 1-24, 2009). Dissimilarities or distances to all other compounds can be calculated this way.

Results are shown in FIG. 12.