Abstract
The Society of Biomolecular Imaging and Informatics (SBI2;
Day 1 of the conference opens with a series of introductory and advanced elective training sessions including an overview of hardware and image acquisition, phenotypic screening, assay types and assay development, systems biology and data analysis, kinetic imaging, single cell cytometry, CRISPR-based applications, and new technologies. During the lunch session, delegates can attend a workshop on assay quality control measures and essentials for imaging and analysis in tumor 3D organoids. This year's colloquium on Emerging Technologies in Biological Models introduces imaging mass spectrometry, live cell imaging, bioprinting, and deep learning informatics. The discussion leaders walk the group through real-life applications and there is a strong emphasis on audience participation; previous forums have been a huge success due to the interactive nature. A detailed overview of the presentations and discussions will be published in ASSAY and Drug Development Technologies in January 2018.
A full 2 days of scientific presentations, technology spotlights, and social events then begin with the scientific program. This year, the opening keynote is by Professor Brenda Andrews from the University of Toronto who will be speaking on “Automated analysis of high-content microscopy data” and Professor Richard Caprioli from Vanderbilt University closes the meeting with “Imaging mass spectrometry: molecular microscopy for biology and medicine.” The scientific sessions cover four themes: advanced and complex cell models, phenotypic drug discovery, informatics, and new technologies.
For the fourth year running, SBI2 has teamed up with ASSAY and Drug Development Technologies to publish a series of research articles on HCS and high-content analysis (HCA). This year's special issue has a tremendous diversity of research articles, starting off with a terrific perspective on HCS over the past 5 years from the team at Novartis in Basel, Switzerland. Haasen et al. walk the reader through the advances they have seen in phenotypic screening for drug discovery, changes in the way researchers have moved from standard 2D cell culture to patient-derived induced pluripotent stem cells (iPSCs) and 3D culture systems as models of disease and of course the impact of CRISPR/cas9 technology to generate gene sensors. In recent years, significant attention has been placed on making phenotypic data accessible to all those researchers who do not have high-level coding skills and importantly making the imaging data interpretable through data visualization. We are currently generating such high volumes of data that we are increasingly struggling to find the best ways to perform analysis and extract all the information possible. Omta et al. present an overview of a project whereby they generated 12 questions posed to researchers (a control and experimental group, total n = 79), relating to interactive and noninteractive data visualization based on a set of HCS data to establish whether one method improved the efficiency of the researcher to interpret data. The data set revealed that the overall time to perform the 12 different assignments was faster in the visualization group with a significantly lower error rate. Of course, using “real people” to evaluate real data sets always has its inherent issues, but this raises some thought-provoking concepts as we move forward with the deluge of data. Shun et al. present an interesting article focusing on the application of HCA of zebrafish at different larval stages and its application to high-throughput screening. They developed a GFP reporter transgene cell line that can mark kidney progenitor cell expansion to track transgene intensity on a whole well basis rather than a single animal basis, similar to the way cell culture cells are quantified. This helps to reduce the biological variability seen in whole organisms. The team showed they can overcome limitations of imaging whole organisms by using multiple wells and applying multivariate analysis on pixel-based distribution of fluorescent transgenes, leading to the possibility of drug screens being performed in a zebrafish model.
High-content screening is an extremely useful method to measure toxicity of substances in cells from a variety of tissue types and cell source. The multiparametric nature of HCS allows for a deeper understanding of mechanism of action as well as making it easier to detect false-positive and false-negative effects. There has been substantial progress in recent years to develop contextual human cell models using human iPSCs, allowing us to move away from a reliance on primary cells, which may be from a different species, in limited supply, or have limited genetic variation, or from using tumor-derived cells that might not represent the physiology of normal somatic cells. It is recognized that toxicology testing in 2D cell culture models might not fully represent the physiological response to compounds seen under 3D growth conditions, which allow for architecture more closely resembling the in vivo state.
Two articles included in this issue cover HCS approaches for in vitro toxicity, using vascular and cardiotoxicity. Iwata et al. present their work to develop screening methods for vascular endothelial cell responses to various toxic agents to develop a high-throughput screening platform. Using both HCA of endothelial cell tube formation and traditional measures for cell viability, they compare the tried and tested human umbilical vein endothelial cells (HUVEC) model of tube formation on a basement substrate (Geltrex™), compared with iPSC-derived endothelial cells. They also compare Geltrex as a growth substrate with a synthetic hydrogel. Using iPSC-derived cells and a synthetic substrate has the potential to increase the size of genetic background of toxicity screens, and also to remove batch-to-batch and donor variability. The study concludes that the HUVEC/Geltrex model was superior, but that the iPSC and hydrogel methods were still consistent and could surpass traditional methods with further development.
Sirenko et al. used iPSC-derived cardiomyocytes to develop a screening method for cardiotoxicity and calcium oscillations in beating cells. This study compares 3D cardiomyocyte spheroids with 2D cultures in multiwell plates. They utilized high-content readouts for kinetic calcium flux, as well as a separate assay for viability using Calcein-AM, MitoTracker Orange, and Hoechst 33342. In addition, fast calcium flux and beating rate measurements were made in high-throughput using a FLIPR Tetra system. To demonstrate proof-of-principle as a screening assay, 22 compounds with known cardiotoxic or cardioprotective properties were used at a range of concentrations. Comparing 2D and 3D cardiomyocyte models, there were broad similarities but no conclusions about which system is better at this point. This highlights difficulties in developing 3D models that more closely resemble tissue, organs, and animal models as increasing complexity does not necessarily deliver a model with true in vivo equivalence and can introduce artifacts, due to poor drug penetrance, or hypoxia.
The partnership between SBI2 and ASSAY and Drug Development Technologies has continued to emphasize high-content screening and analysis by engaging authors, researchers, and readers and by bringing high-impact science to the forefront. We are delighted to publish such a high-quality selection of articles and look forward to another exciting meeting in the new location of San Diego.