To monitor changes in cellular mitochondrial content, we developed a fluorescent image-based assay () in human umbilical vein endothelial cells (HUVECs). These primary human cells grow as a monolayer and have a flat morphology, which is ideal for image analysis. Hoechst nuclear stain was used to identify each nucleus in the images, the F-actin stain phalloidin was used to define the boundary of the cell, and MitoTracker Deep Red was used to monitor changes in mitochondrial content. For each cell, total MitoTracker intensity as well as that cell's cytoplasmic area were measured. For a given image, median values were calculated from the population of cells.
An image-based assay for mitochondrial content and cell size.
We screened in biological duplicate 28,786 small molecules treated for a three-day period (). Surprisingly, we found that hundreds of compounds increased MitoTracker intensity () on a per cell level. On closer inspection, however, most of these compounds also increased cell size in proportion to the fold increase in mitochondrial content (). These compounds included many of the known bioactives including previously identified enhancers of mitochondrial content 
such as microtubule modulators paclitaxel () and deoxysappanone. The screening results support and extend the original observations of Posakony et al.
that cell size and mitochondrial content are strongly coupled and suggest that mechanisms regulating cell size also influence mitochondrial content.
The relationship between mitochondrial content and cell size revealed through 28,786 chemical perturbations.
Hence, we became particularly interested in molecules capable of dissociating the relationship between cell size and mitochondrial content, as they could potentially reveal novel mechanisms controlling mitochondrial content (). The 160 highest scoring compounds that increased MitoTracker intensity after normalizing for cell size were selected for secondary assays. The assays consisted of immunofluorescence assays for mitochondrial protein content and qPCR assays for mitochondrial DNA (mtDNA) copy number. Ten highest scoring compounds from each secondary assay were promoted for oxygen consumption assays to ensure that elevations in mitochondrial content corresponded to an increase in functional respiratory capacity. Of the 20 compounds tested, three compounds (BRD6897, BRD6445, BRD1108) showed statistically significant increases in uncoupled respiration (Figure S1
We chose to focus on one of these three compounds, namely BRD6897 (), as it imparted the greatest, dose-dependent increase in uncoupled respiration (Figure S2
). BRD6897 comes from a kinase inhibitor-biased library of compounds that are predicted to bind the ATP ligand site of kinases, though its specific target is not known. The binding ability of BRD6897 to a panel of 442 kinases 
was tested using 10 µM compound concentration but failed to reveal any significant binding. The result suggests that BRD6897 does not have strong affinity for a range of kinases often observed with known kinase inhibitors 
. In our cells, a 10 µM dose of BRD6897 induces a 1.55 fold increase in uncoupled respiration (). Additional screening of 36 structural analogs of BRD6897 not present in the initial screening collection failed to reveal more potent inducers of mitochondrial content (Figure S3
). An increase in uncoupled respiration with BRD6897 is detectable by two days of treatment (Figure S4
A), plateaus after three days of treatment (Figure S4
B), and is partially reversible after removal of the compound for three days (Figure S4
C). The increase in respiration is robust to the normalization scheme used. BRD6897 increases respiration when normalized to total cell number [1.35-fold], total cell volume [1.46-fold], or total cellular protein [1.57-fold]. The compound is toxic in dividing cells but is capable of enhancing uncoupled respiration in other non-dividing cells. For example, it induces a 1.60-fold elevation in uncoupled respiration in confluent 3T3-L1 preadipocytes at a dose of 10 µM (Figure S5
Effects of BRD6897 on mitochondrial content, gene expression, and physiology.
We next sought to identify the cellular pathways through which BRD6897 might be acting. If it is acting via the known transcriptional pathways of mitochondrial biogenesis 
, we would expect concomitant elevations in both nuclear DNA and mitochondrial DNA (mtDNA) encoded OXPHOS genes. In addition, mtDNA copy number itself as well as OXPHOS protein level should increase. Contrary to this expectation, BRD6897 shows no significant elevation in nuclear DNA encoded OXPHOS gene expression while mtDNA encoded OXPHOS gene expression is increased (). Moreover, BRD6897 actually reduces mtDNA copy number () while elevating OXPHOS protein content (). These findings indicate that BRD6897 is not acting through the canonical transcriptional programs and suggests that other mechanisms are at play.
To examine the mechanisms responsible for the increase in uncoupled respiration, we studied the morphology of mitochondria by electron microscopy. The proportion of cytoplasm area occupied by mitochondria in BRD6897-treated HUVEC did not differ significantly from DMSO-treated HUVEC (). This is in contrast to the increase in mitochondrial area observed during PGC-1α mediated mitochondrial biogenesis 
. Instead, mitochondria in BRD6897-treated HUVEC appeared more electron dense compared to mitochondria in DMSO-treated HUVEC () suggesting increased protein density within mitochondria. We next examined whether OXPHOS protein synthesis or degradation was altered upon BRD6897 treatment using pulse-chase labeling. BRD6897 did not increase the rate of cytochrome C synthesis (Figure S6
, left). However, BRD6897 slowed the rate of cytochrome C degradation (Figure S6
, right) suggesting that mitochondrial protein turnover is partially inhibited by BRD6897. Therefore, BRD6897 does not appear to induce mitochondrial biogenesis via transcriptional programs, but perhaps through a mechanism involving decreased mitochondrial protein turnover.
Changes in mitochondrial ultrastructure upon 3-day 10 µM BRD6897 treatment.
In summary we have performed a high-throughput, high-content microscopy assay that reveals a stereotyped, coordinated control of mitochondrial content and cell size preserved across hundreds of small molecule perturbations. Given that growth factors 
such as VEGF and downstream mitogenic signals such as myc 
are known to induce mitochondrial biogenesis as well as stimulate cell growth, these signals may be responsible for ensuring the strong coupling between mitochondrial content and cell size. Importantly, we identified a handful of molecules that are unique in their ability to disrupt the strong relationship between mitochondrial content and cell size. One such compound, BRD6897, increases cellular mitochondrial content on the basis of microscopy, protein content, and respiration in two different non-dividing cell types. It does not appear to activate the known transcriptional programs of mitochondrial biogenesis, but rather, appears to influence mitochondrial content perhaps by modulating protein turnover. At present the precise molecular target of BRD6897 remains to be elucidated, but if identified, could reveal an important new pathway, that together with transcriptional programs of mitochondrial biogenesis, would serve to regulate mitochondrial content.