Background

Since its introduction in the 1960s, the nematode Caenorhabditis elegans has become a popular model across all of biology. There are multiple reasons for its popularity: it is small, simple, inexpensive, and easy to maintain, has substantial genomic conservation with human, very powerful and ubiquitous genetic tools, and is optically transparent. These characteristics make it particularly appropriate for high-throughput screening (HTS), where a comparatively simple assay is repeated thousands of times across a library of interest (often small molecules or genome-wide RNAi). A more powerful extension of HTS is high-content HTS, where complex or multiple phenotype readouts are simultaneously ascertained (e.g., the combination of survival and expression of a particular reporter, perhaps restricted to a particular tissue or subcellular localization). Live, whole organism screens are becoming increasingly popular, and are attractive methods to test the effects of gene knockdown or small molecule treatment on physiology, to identify bioactive and antimicrobial compounds, and to study genetic interactions. Indeed, many labs looking to identify small molecules and drugs to study aging (Lucanic et al., 2018), neurological diseases (Gosai et al., 2010), and host-pathogen interactions (Moy et al., 2009; Conery et al., 2014; Kirienko et al., 2016), have used worms for their first steps, and HTS have been developed for different types and purposes of experiments, in both agar- (Burns et al., 2006; Squiban et al., 2012; Lucanic et al., 2018) and liquid-based (Moy et al., 2009; Anderson et al., 2018) systems.

As more and more of these screens are conducted in C. elegans labs, output analysis has become a major bottleneck. Fluorescence quantification of images obtained from microscopy for phenotype-based screens often requires time-consuming and expensive manual labor that is prone to inadvertent bias. Additionally, the clustering, crossing, and/or tangling that often occur with C. elegans complicate measurement of individual animals in the images. Automated microscopy platforms and complex software algorithms have been developed to resolve these issues. ArrayScan V^TI and SpotDetector BioApplication are examples of such software, and are able to recognize worms and quantify their sizes and fluorescence (Gosai et al., 2010; Maglioni et al., 2015). The Worm-align/Worm_CP workflow in CellProfiler also provides a cost-effective method to assess individual worms in images (Raviv et al., 2010; Wählby et al., 2012; Okkenhaug et al., 2020).

Unfortunately, while these pipelines are very useful for the measurement of phenotypes in individual animals, their computational demands still limit the sample sizes. Optimal numbers of worms in a well for a reliable count by using the above method is between 50 and 100, depending on the life stage of the worms (Maglioni et al., 2015). Conventional understanding is that this is low-throughput [defined as 1-500 individuals vs medium (500-10,000), high (10,000-100,000) or ultra-high (>100,000) throughput (Wildey et al., 2017)]. Many experiments may require measurements of individual animals from larger numbers of samples, such as searching for small effects or rare phenomena. For example, C. elegans at the L2 stage have a more oxidizing redox environment and increased variability between individuals compared to young adults (Bazopoulou et al., 2019), and measurement of the redox environment in a larger population provides a more accurate representation of the biological state of the organism. As another example, analysis of a large population may be necessary to identify rare crossover events between closely linked alleles in a mutant screen. Finally, experimental assays using a large number of individuals may be required to overcome substantial inter-individual variability (Angstman et al., 2015 and 2016).

Our lab routinely monitors gene activation via fluorescent transgenes and measures mitochondrial parameters using fluorescent dyes. The protocols we described (Tjahjono and Kirienko, 2017; Kang et al., 2018; Revtovich et al., 2019; Tjahjono et al., 2020) are invaluable for this purpose. Here we describe the imaging and fluorescence quantification protocols we routinely use. We commonly use a COPAS FlowPilot BioSorter (Union Biometrica) to obtain automated, unbiased, high-throughput measurements of fluorescence. This is analogous to flow cytometry, so we refer to it as flow vermimetry (from the Latin vermis for worm). This technique has been previously described for use in RNAi screens (Squiban et al., 2012). We also use the Cytation5 multimode plate reader/imager (BioTek) to obtain images and measure phenotypic changes over time. We have optimized the cellular analysis pipeline from Gen5 software for worm identification and fluorescence measurement. The Data Analysis section shows that the measurement values from these techniques demonstrate a high correlation. Advantages and limitations of each approach are also discussed.