Objective

Although the intention of high content screening is to extract as much relevant information as possible, a large percentage of high content screening assays only analyze a small number of image based properties. This can result in the loss of valuable information. Nearly all screening approaches use a nuclear counterstain like Hoechst to aid in segmentation. Sometimes Hoechst total sum intensity distribution is also used to analyze cell cycle distribution, in particular G0/G1, S and G2/M populations. However, besides the cell cycle analysis, there is still more information that can be retrieved from the nuclear “counterstain”. Here we show that it is possible to distinguish cell lines based on their nuclear morphology only. Morphology not only refers to outer shape parameters but also to the distribution of intensity and texture properties inside the nucleus. We seeded the hepatocyte cell line HepG2 and the fibroblast cell line NIH3T3 either alone or as a coculture in a 384 well plate and stained the nuclei with Hoechst. The plate was then imaged and analyzed using an Opera Phenix™ or Operetta CLS™ High Content Screening System and Harmony™ High Content Imaging and Analysis Software. Using the advanced texture and morphology analysis algorithms in combination with the PhenoLOGIC machine learning option built into the Harmony software we are able to distinguish the two cell types and to identify them in the cocultures. To control the classification result, HepG2 cells were stained with CellTracker™ Green and NIH3T3 with CellTracker™ Red. For all cells from the output populations of the PhenoLOGIC classification, the intensity of the CellTracker™ dyes was determined and the rate of falsely assigned cells determined. The accuracy of the classification was 98%.To push the analysis even further, we seeded seven commonly used cell lines (A549, HeLa, HepG2, MDCK, MCF7, NIH3T3, HT1080) individually, stained them with Hoechst, imaged and analyzed them. Using Principle Component Analysis and unsupervised machine learning we demonstrate that the nuclear counterstain provides sufficient information to separate the seven cell lines. Additionally, we treated the cells with inhibitors of either histone acetyltransferases or histone deacetylases and analyzed their effect on changes of the nuclear morphology. These studies clearly show that combining high resolution imaging capabilities with advanced image analysis tools opens up the door to more extensive high content analysis.