HCI and Subcellular Analysis of Autophagy in 15-30 Minutes per Plate

In this application note, Araceli Endeavor® and Clairvoyance™ are used for an uncompromising end-to-end high content imaging (HCI) examination of autophagy through overnight application of autophagy inducers (Torin-1, Verapamil) or vesicle degradation blockers (chloroquine, MG-132). Sub-micron resolution with maximized well coverage and object- level analysis are used to interrogate 2 different techniques to assay autophagy: immunocytochemistry (ICC) against an autophagosome marker or a live cell dye optimized to stain autophagosomal vesicles. These methodologies are compared, arriving at largely the same EC50 values, with better single puncta resolution in ICC but a faster workflow with the live cell dye (imaging in 6 minutes 9 seconds and analysis in 8 minutes 3 seconds). Maximized well coverage is highlighted as a necessary way to avoid analysis mistakes.

Cells eliminate intracellular waste and defective components through autophagy (“self-eating”), a highly conserved pathway in protecting against, and responding to, stress. Regulating autophagy is a key therapeutic target for aging-related dysfunction (Djajadikerta et al 2020) and its dysregulation is implicated in a wide range of disorders, from neurodegeneration to cancer (Marinković et al 2018). While there are multiple strategies to detect autophagic flux, including cell lines and transfection kits tagging autophagic proteins with fluorescent markers (which may itself induce autophagy due to overexpression), this application note concentrates on two distinct, common fluorescent labeling strategies used in high content screening (HCS): immunocytochemistry (ICC) and live cell dye staining (Marx 2015). To visualize autophagosomes in IHC, we use an antibody to bind to the cleaved form of LC3 (LC3b), a protein with well characterized localization to autophagosomes and correlates well with autophagy levels (Kabeya et al 2000), which is then conjugated to a fluorescent molecule. To evaluate the performance of a live cell dye, we used a fluorescently labeled cationic amphiphilic tracer (CAT) that binds to the cellular compartment with the correct charge and lipid composition, optimized here for autophagosomic vesicles. IHC involves a more elaborate process to stain the cells, but stains more specifically with less background, whereas the CAT involves minimal preparation and can be used with live cells, but may exhibit lower signal and more background.

Many assays like autophagy rely on quantifying small puncta with subcellular precision. However, most high content systems cannot deliver this submicron-level detection with adequate sample coverage at high throughput speeds, causing a bottleneck and impeding large-scale screening. Utilizing the Araceli Endeavor® high content imager and Clairvoyance™ analysis software eliminates the bottleneck, with <10-minute imaging at sub- micron resolution with maximized well coverage, followed by subcellular object-level data quantification in <20 minutes for 96- 384- and 1536-well plates. This application note utilizes 2 distinct autophagic flux assays to image autophagic puncta (Figure 1) at high throughput speeds without sacrificing resolution or coverage, with total imaging and analysis time in as little as 15 minutes without compromising assay coverage or resolution. By measuring cytosolic autophagosomes through immunocytochemical labeling and autophagosomal aggregates with a live cell dye, this application note demonstrates high throughput quantification of autophagy in two different modalities and arrives at equivalent conclusions, showing the flexibility, resolution, and speed of high content imaging with Endeavor, paired with fast and accurate quantification in Clairvoyance.

Assay: Human bone carcinoma (U2OS) cells were plated at 12,500 cells/well in Grenier CellView 96 well plates (655891) and cultured for 48 hours in standard conditions (complete media, 5% CO, 37°C). Cells were treated with compound or DMSO in a 1:2 dilution series, 3 replicates/ plate, for 12-16 hours. All solutions and wash steps are in PBS pH 7.4 using a plate washer (Biotek ELx405).

Treatments: DMSO: vehicle control; [hydroxy] chloroquine: autophagosome degradation inhibitor; verapamil: autophagy inducer via Ca²+ inhibition; MG-132: autophagosome arrest via proteasome inhibitor; torin-1: autophagy inducer via mTOR inhibition. All compounds diluted in DMSO, with total concentration no greater than 0.4% DMSO. No effects were seen in DMSO-alone treatments (Figure 2b).

Immunocytochemistry-based detection: Cells were fixed in 10% formalin for 15 minutes, permeabilized with 0.1% Triton x-100, then stained and counterstained for 60 minutes/each in 1% BSA, with 1:1000 Rb anti-LC3b (Life Technologies L10382) and 1:1000 Gt anti-Rb Alexa 488 (Invitrogen A32731) with 1:750 Hoechst 33342 and 1:1000 Phallodin-647 (Cayman 20555). Cells were washed in PBS pH 7.4 4x between steps. 96-well plates were imaged on the Araceli Endeavor® using 3 channels, with maximized well coverage (4.8mm x 4.8mm/well, ~75% of well), full resolution (0.27 µm/pixel) data collected <10 minutes/each.

Dye-based detection: After treatment, cells were incubated at 37°C with 100 µL/well of 1:500 dilution of CYTO-ID® Autophagy detection kit 2.0 (Enzo KIT175) in complete media. The plate was washed then fixed with 10% formalin for 15 minutes, washed again, then stained with 1:750 Hoechst 33342 for 60 minutes before a final wash. The plate was then imaged on Endeavor in 2 channels, with focus set 1.5 micron above default and 70% well coverage, full resolution (0.27 µm/pixel) data collected for the whole 96-well plate in 6 minutes 9 seconds.

Analysis: Clairvoyance™ software was used for all analysis, analyzing all fields of view within a well, excepting the single FOV experiment in Figure 4. For both assays, template matching (a machine vision technique for objection detection based on similarity) was used to identify and quantify nuclei and autophagic vesicles, with a 3×3 µm template used for ICC and a 4.6×4.6 µm used to quantify CytoID staining, as staining appeared more aggregate-like. For ICC data, thresholded actin staining was used to define cells within a defined region around the center of each identified nucleus, then a cytoplasmic mask made by subtracting the nuclear mask from the cellular mask. Within the area defined by the cytoplasm, vesicles were then counted per cell using machine vision (16 minutes 52 seconds and 18 minutes 56 seconds for the 2 plates). Unlike ICC, live stain data (Cyto-ID) was analyzed segmentation-free, with aggregates counted per image using machine vision and normalized to nuclear count (8 minutes 7 seconds). Graphing was done in GraphPad Prism, with a 5-parameter linear regression used to generate curves and calculate EC50.

Autophagy was examined with five different modulators, with robust increases in autophagic flux detected across these mechanistically distinct compounds using both the CAT dye and ICC (Figures 1-4). Maximized well images at submicron resolution (Figure 1), representing 70-75% of well area at 0.27 µm/pixel, were acquired in 6 minutes 9 seconds and <10 minutes, respectively. Differences in imaging speed here are largely due to improvements in Araceli Endeavor® software in the months between when the ICC and CAT plates were scanned. Submicron-level resolution is required to accurately resolve puncta in these assays, as mammalian autophagosomes are 0.5-1.5 µm in diameter (Parzych and Klionsky 2014). Analysis was performed in 8 minutes for the CAT plate and 16-19 minutes for the two ICC plates using Clairvoyance™ software. Both methodologies yielded similar data quality with consistent changes, Z’>0.5 for all compounds tested (Table 1), demonstrating assay, imaging, and analysis reliability.

Using cationic amphiphilic live cell tracer CytoID, we first looked at a direct inducer of autophagy (Figure 2), mTor inhibitor Torin-1 (Andersson et al 2016). Aggregates/cell were quantified using a machine-vision approach. We observed minimal baseline aggregation that consistently increased with dosage before flattening, yielding a robust effect (Z’=0.82), accurately represented by a sigmodal curve (R²=0.99) with an EC50 of 0.1 µM (Figure 2). The corresponding dosage of DMSO alone (vehicle control) produced no effect.

This application note also uses a more traditional methodology to measure autophagy: ICC against autophagosome protein LC3b. Since LC3b accumulates in the nucleus in the absence of autophagy (Kraft et al 2016), the analysis looked at autophagosomes exclusively in the cytosol, using actin staining to define the cell boundaries, excluding nuclear signal defined by Hoechst 3342. After treatment with an inducer of autophagy through Ca²+ inhibition (verapamil) or a proteasome inhibitor that prevents autophagosome degradation (MG132), LC3b ICC yielded distinct, readily quantifiable cytosolic puncta (Figure 3) in a dose-dependent manner, though MG132 exhibited cytotoxicity at the highest dose. While nuclear staining was abundant in controls, low baseline levels of autophagosomes were detected in the cytosol, yielding a 30-60x increase in puncta in treated cells compared to control, with Z’=0.63 for verapamil and Z’=0.83 for MG132.

These data also highlight the importance of maximized well coverage on this assay: due to over washing, cells in the middle of this plate unadhered (Figure 4a and b). If only this center FOV was analyzed (Figure 4c), non- representative data would have been gleaned from this experiment, with an entirely different concentration curve leading to potentially misleading conclusions.

Testing the well-characterized autophagosome recycling blocker chloroquine, ICC and CAT assays yielded equivalent results when acquired on Araceli Endeavor® and analyzed in Clairvoyance™ (Figure 5). Notably, these approaches were visually distinct. While maintaining similar staining patterns and localizing to the same subcellular compartment, anti-LC3b staining yielded bright, distinct somatic puncta (Figure 5a, Figure 1c) with diffuse nuclear staining, while the live cell dye yielded dimmer, more diffuse aggregates (Figure 5c, Figure 1d). Consequently, the measured output of these assays is subtly different: aggregated groups of autophagosomes in CAT rather than the individual autophagosomes in ICC.

For object detection, larger templates were used for CAT aggregates versus ICC puncta, with almost an order of magnitude difference in number of objects detected. Despite these differences, the two assays showed the same effective responses to the drug, with similar curves Araceli Biosciences Inc. 2024© and EC50 values of 3.5 µM for CAT and 5 µM for ICC. Both assay workflows proved reliable, with Z’=0.63 and 0.85, respectively. Despite differences between staining, Endeavor and Clairvoyance delivered robust, equivalent results between assays.

By pairing resolution and speed with rapid image analysis, Endeavor high content imager and Clairvoyance high content imaging analysis software demonstrate that assaying for autophagy can truly be a high throughput screen. With this system, end-to-end imaging and analysis took <15- 30 minutes/96-well plate, with robust, consistent results (average Z’=0.79, Table 1). This readily scales up: since total imaging area is the same, maximized well imaging of 1536-well plates only takes an additional 1-2 minutes/ plate longer than the plates used here, with equivalent analysis times. This enables over 100,000 compounds to be screened in a single day without compromising assay quality: truly high throughput speeds in HCS.

Here, the combination of full well coverage and submicron resolution is shown to be essential for minimizing variability and quantifying micron-diameter autophagosomes, yielding Z’>0.5 for all four mechanistically distinct modulators examined under 2 different assays (Figure 2 and 3). By analyzing full well data, artifacts and misleading results can be avoided (Figure 4). Arriving at equivalent results with two distinct strategies for autophagic detection with different staining characteristics (Figure 5), Endeavor and Clairvoyance were shown to be resilient and flexible at the assay level. This speaks well to generalizability of Araceli’s imaging and analysis platform to other quantitative spot- and aggregate- based HCS assays such as subcellular protein localization, plaque formation, aggregation, FISH, and phagocytosis.

Andersson AM, Andersson B, Lorell C, Raffetseder J, Larsson M, Blomgran R. Autophagy induction targeting mTORC1 enhances Mycobacterium tuberculosis replication in HIV co-infected human macrophages. Sci Rep. 2016 Jun 15;6:28171. doi: 10.1038/srep28171.

Djajadikerta A, Keshri S, Pavel M, Prestil R, Ryan L, Rubinsztein DC. Autophagy Induction as a Therapeutic Strategy for Neurodegenerative Diseases. J Mol Biol. 2020 Apr 3;432(8):2799-2821. doi: 10.1016/j.jmb.2019.12.035.

Kabeya Y, Mizushima N, Ueno T, Yamamoto A, Kirisako T, Noda T, Kominami E, Ohsumi Y, Yoshimori T. LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing. EMBO J. 2000 Nov 1;19(21):5720-8. doi: 10.1093/emboj/19.21.5720.

Kraft LJ, Manral P, Dowler J, Kenworthy AK. Nuclear LC3 Associates with Slowly Diffusing Complexes that Survey the Nucleolus. Traffic. 2016 Apr;17(4):369-99. doi: 10.1111/tra.12372.

Marinković M, Šprung M, Buljubašić M, Novak I. Autophagy Modulation in Cancer: Current Knowledge on Action and Therapy. Oxid Med Cell Longev. 2018 Jan 31;2018:8023821. doi: 10.1155/2018/8023821

Marx V. Autophagy: eat thyself, sustain thyself. Nat Methods. 2015 Dec 1;12(12):1121-5. doi: 10.1038/nmeth.3661. PMID: 26624796.

Panda PK, Fahrner A, Vats S, Seranova E, Sharma V, Chipara M, Desai P, Torresi J, Rosenstock T, Kumar D, Sarkar S. Chemical Screening Approaches Enabling Drug Discovery of Autophagy Modulators for Biomedical Applications in Human Diseases. Front Cell Dev Biol. 2019 Mar 19;7:38. doi: 10.3389/fcell.2019.00038. PMID: 30949479; PMCID: PMC6436197.

Parzych KR, Klionsky DJ. An overview of autophagy: morphology, mechanism, and regulation. Antioxid Redox Signal. 2014 Jan 20;20(3):460-73. doi: 10.1089/ ars.2013.5371.