While we only implemented ArrayEdit

While we only implemented ArrayEdit for fluorescence loss and growth rate differences, the method employing HCA is extendable in principle to any image-based phenotype. Phenotypes could be defined by changes in uptake of cytoxicity dyes, live immunocytochemistry for cell surface or extracellular matrix markers, calcium flux dyes, or mitochondrial functional dyes (Taylor and Haskins, 2007). A distinguishable phenotype in edited kinesin inhibitor may not be readily apparent by HCA in the pluripotent stem cell state. Hence, differentiation on ArrayEdit may be required to distinguish edited phenotypes. These capabilities on ArrayEdit seem possible, as arrayed neural organoid culture has already been achieved on μCP plates (Knight et al., 2015). More sophisticated computational methods could be easily implemented in our analysis pipeline in CellProfiler to prospectively identify imaging phenotypes that connect to proper or abnormal biological and epigenomic characteristics of edited cells (Singh et al., 2014). Because all edited clones share the same culture medium, the intra-well benchmarking of phenotypes on ArrayEdit could enable identification of phenotypes that may be lost due to noise or fluctuations in medium composition among culture wells or plates. Variations in signaling factors in the medium could also lead to hESC clones that become more or less lineage committed (Nazareth et al., 2013), opening a window to study the biological variability within clones that is difficult to ascertain during standard genotyping of clones. The stringent definition of phenotypes enabled by HCA with ArrayEdit will likely permit a more thorough characterization of the biological and functional consequences of various gene-editing protocols. Current limitations of ArrayEdit arise from setting up the platform and screening for phenotypes in pluripotent cells. Although μCP is a straightforward technique, ArrayEdit is not a turn-key ready platform for many traditional biology or industrial laboratories who may need access to laser cutting or automated microscopy units. However, many commercial HCA instruments are available on the market, and our HCA pipeline can be readily employed using standard cloud-based computing or even a personal computer. The simple, versatile, and well-characterized μCP chemistry requires only standard laboratory equipment. The chemistry could also be flexibly modified to create various hydrophilic and hydrophobic areas on a single surface (McNulty et al., 2014; Sha et al., 2013), even well-of-the-well, water-in-oil culture platforms that are routinely used in pre-implantation embryo culture. Such chemically defined surfaces may be particularly attractive for clinical application in future work. One-pot transcription of sgRNAs from PCR amplicons generated by oligonucleotide DNA primers produced clean and functional sgRNAs. One-pot transcribed sgRNAs can be rapidly designed and made from commercial vendors overnight with costs scaling with 20–60 base pair synthesis; the costs are anticipated to decrease over time (currently <$1 USD per sgRNA per experiment; see Table S4). In contrast to other methods that require the purchase of multiple oligonucleotides, our method requires only one unique oligonucleotide that can be synthesized in a multiwell plate format by commercial vendors, decreasing the setup time and the possibility of pipetting error. Errors in long (>60 nt), chemically synthesized oligonucleotides (up to 10%) have been observed (Liang et al., 2015), and our method notably avoids the use of long oligonucleotides by using a sequence-verified, synthesized, double-stranded DNA for the long universal region of the sgRNA. Our modular design also permits facile incorporation of additional RNA elements and devices (Nissim et al., 2014; Shechner et al., 2015). Furthermore, our method performed better than several previously described methods (González et al., 2014; Mali et al., 2013) (Figures 1D and E) and generated sgRNAs in less than 2 days. When analyzed via deep sequencing, one-pot sgRNAs had an efficiency of editing between 20% and 92% of mRNA transcripts after EB differentiation. Interestingly, many of the transcripts analyzed (5 of 7) had a higher percentage of in-frame mutations than would be caused by random chance (33%) (Shi et al., 2015). This may suggest that there were selection pressures in the EB cultures that modify the mutation spectrum observed.