An improved method for high-throughput quantification of autophagy in mammalian cells

Autophagy is a cellular homeostatic pathway with functions ranging from cytoplasmic protein turnover to immune defense. Therapeutic modulation of autophagy has been demonstrated to positively impact the outcome of autophagy-dysregulated diseases such as cancer or microbial infections. However, currently available agents lack specificity, and new candidates for drug development or potential cellular targets need to be identified. Here, we present an improved method to robustly detect changes in autophagy in a high-throughput manner on a single cell level, allowing effective screening. This method quantifies eGFP-LC3B positive vesicles to accurately monitor autophagy. We have significantly streamlined the protocol and optimized it for rapid quantification of large numbers of cells in little time, while retaining accuracy and sensitivity. Z scores up to 0.91 without a loss of sensitivity demonstrate the robustness and aptness of this approach. Three exemplary applications outline the value of our protocols and cell lines: (I) Examining autophagy modulating compounds on four different cell types. (II) Monitoring of autophagy upon infection with e.g. measles or influenza A virus. (III) CRISPR/Cas9 screening for autophagy modulating factors in T cells. In summary, we offer ready-to-use protocols to generate sensitive autophagy reporter cells and quantify autophagy in high-throughput assays.

Flow cytometry and cell sorting. Please see the detailed protocol for the exact experimental procedures.
Single cells to create clonal cell lines were sorted into 96-wells using BD FACSAria III. To measure eGFP or allophycocyanin (APC) fluorescence, a BD FACSCanto II or Beckman-Coulter CytoFLEX with attached highthroughput samplers were used. Voltage/gain for forward and sideward scatter were set to such values that allowed clear separation of the cell population from debris. For all co-staining, we chose APC as the fluorophore to avoid compensation with the eGFP signal. eGFP mean fluorescence intensity (MFI) in mock cells was set to above 1,000 to allow for linear monitoring of changes in both directions (less/more autophagosomes) using voltage/gain settings. A minimum of 10,000 intact single cells were measured per sample. Raw fluorescenceactivated cell sorting (FACS) data were analyzed using FlowJo 10. Intact cells were gated in FSC-A vs SSC-A and single cells gated in FSC-H vs FSC-A. Stringent gating strategies to exclude debris or dead cells that exhibit autofluorescence were applied. eGFP-MFI of all samples was calculated and eGFP-MFI of the mock control subtracted (= baseline autophagic flux) from the other samples to depict the eGFP-MFI shift (= shift of autophagic flux). DNA analysis. sgRNA PCR amplicons were mixed with 6 × loading dye (Roti-Load, Carl Roth) and loaded on a 1% agarose gel (w/v, Sigma Aldrich) in 1 × TAE buffer (Carl Roth) next to a 1 kb Plus DNA ladder (Thermo Next-generation sequencing and analysis. Sequencing was performed on the Illumina Genome Analyzer in the GeneCenter sequencing facility (LAFUGA). Obtained sequences were processed with the Trim Galore! toolkit to remove adapter sequences and reads with PHRED scores below 30 as previously described 39,40 . Default settings were used to process the raw reads. Combined from three independent experiments, 6.4 million reads were obtained for the low sample and 4.2 million for the input sample. To align the reads to the GeCKO 2.0 gRNA table, quantify, and normalize them, the MAGeCK suite was used 41 . The raw sequencing data were deposited in GEO (ID: GSE147488).
Statistical and bioinformatic analysis. Data were collected at least in triplicates for all flow cytometrybased approaches and with at least 30 replicates for fluorescence confocal microscopy-based approaches. For the relative comparison of changes in autophagic flux, statistical differences were assessed by one-way ANOVA or unpaired t-test as indicated. Results were graphed by Graph Pad Prism 8, and are displayed as means ± standard error of the mean (SEM) with individual values shown. P-values ≤ 0.05 were considered statistically significant. A p-value below 0.05 is annotated with *, p-values between 0.05 and 0.01 with **, and p-values below 0.001 with ***. To evaluate the quality of the method for high-throughput screening, we calculated the Z-factor as described previously 42 . A Z-factor above 0 indicates a method suitable for high-throughput approaches, 1 represents the maximum quality reachable.

Detailed protocol
Generating autophagy reporter cells. A detailed list of materials is provided in Table 1.

General considerations.
For the construction of the cell lines use low passage numbers. For accurate autophagy measurements, make sure that the cells never overgrew, even during the construction of the cell lines, discard any flasks which are too dense. Cell lines should be discarded once the passage number is higher than 20 (assuming a biweekly split-cycle).    1.3. Transduce 5 × 10 5 cells of the selected target cell line (e.g. HEK293T, HeLa) with 100 µl of the SN harvested in step 1.2.8. 1.4. Two days post-transduction, detach the cells using Trypsin (0.05%, GIBCO, for adherent cells), pellet at 300 g, 3 min, 4 °C. 1.5. Remove the SN and replace it with fresh medium without FBS to a concentration of 5-10 × 10 6 cells per ml in a 15 ml Falcon tube. Filter through round-bottom tubes with a cell strainer cap (5 ml, Falcon). Sort samples on a FACS sorter e.g. BD FACSAria III, using a 100-micron ceramic nozzle. The flow rate was kept below 2, and the purity setting was set to a single cell. Sort one cell per 96-well F-bottom well. Important: The nozzle size needs to be larger than the cell passing through it. Avoid crushing or stressing the cells too much. 1.6. Provide single cells with medium supplemented with 20% FBS according to the cell type. Select eGFP-LC3B cells, with moderate eGFP MFIs (Fig. 1B). Important: Some cell lines do not grow efficiently from single clones. Avoid isolating cell lines expressing too much or no eGFP-LC3B. 1.7. Approximately 3-6 weeks post sorting, clones can be transferred from half-confluent 96-wells directly into 12-well plates. Avoid overgrowing in 96-wells or at any step. 1.8. eGFP-LC3B expression and fluorescence can be monitored using a fluorescence microscope and clones with little to no detectable fluorescence are discarded. Clones with clearly visible aggregations of eGFP-LC3B in the cytoplasm are also discarded. 1.9. Grow clones up to confluent T25 flasks and then proceed to verify the clones.

Important:
The level of expressed eGFP-LC3B impacts the sensitivity of the assay. Generally, we have found that a medium amount of eGFP-LC3B seems to be optimal. Too low eGFP-LC3B expression causes a loss of detectable eGFP-LC3B fluorescence after washout. Too high eGFP-LC3B causes a high baseline of autophagosome associated eGFP-LC3B, as well as occasionally aggregates observed in confocal microscopy.
Produce frozen stocks as early as possible to preserve the generated cell line at a passage number as low as possible. Table 5.

Grow cells in either F-bottom (adherent cell types) or V-bottom (suspension cell types) 96-well plates.
Seeding of 50,000 cells per well for adherent cells and 100,000 cells per well for non-adherent cells is recommended, 18 h before a 4-h assay. Important: At the time of harvest, the cells should be approximately confluent, and kept in fully supplemented medium which is still at optimal pH. Adjust the cell number/density at the time of seeding to meet these criteria. Do not change medium/nutrient conditions before treatment/harvesting. Any stress on the cells impacts autophagy and thus may bias the assay. Consider appropriate negative and positive controls. Established drugs like Rapamycin, Chloroquine, or Bafilomycin A1 are recommended as controls. Negative controls should be treated the same way as the samples, e.g. add the carrier for a drug in the same concentration to the mock SN. Screening results have to be confirmed using orthogonal methods for assessing autophagy levels.  Important: Additional washing steps increase the signal/noise ratio. However, at a loss of cell count. We have found that washing twice is sufficient for most applications, and cell numbers stay within a reasonable range. 2.11. Spin down the cells at 500 g, 3 min, 4 °C, discard SN. 2.12. Use one of three different subsequent treatments (see Fig. 3D  2.13. Proceed to flow cytometry. Note that saponin-treated cells appear smaller than non-treated cells and have less eGFP content, as well as less autofluorescence, see Fig. 3B. 2.14. Use the voltage or gain settings of the cytometer to set the MFI of mock-treated cells to 1,000. 2.15. Living and single cells are gated using FSC and SSC, measure at least 10,000 living/single-cell events. Important: Quantification using less than 10,000 cells still is possible, however, the inherent heterogeneity of autophagy, even within cell populations derived from a single clone, may result in higher deviations between biological replicates. 2.16. Extract the MFI of eGFP-LC3B from all samples. For background correction, the MFI measured in mock conditions is subtracted from the MFI of treated samples (see Flow Cytometry and Cell Sorting).

Applications of high-throughput quantification of autophagy (Examples).
A detailed list of materials is provided in Table 6. Important: All hits in screening approaches have to be verified using orthogonal methods to assess autophagy (for a comprehensive overview of available methods see 24 ).
3.1. High-throughput screening for novel autophagy modulating drugs. Optionally: Cells can be treated longer or shorter, we have observed that a time frame between 2 and 4 h is optimal for most drug-based applications. However, drugs that are expected to block autophagic flux may be kept on the cells for longer to measure the accumulation of vesicles from basal autophagy. Please note that especially blocking autophagy decreases the viability of the cells after 4-6 h.  Important: Transfection induces autophagy. We have tested a few transfection reagents and found that PEI and calcium phosphate transfection induce the lowest amount of autophagy as opposed to commercially available transfection reagents. We generally use PEI, as it results in higher transfection rates/expression rates than calcium phosphate transfection. Thus, pre-test your transfection method. Alternatively, transduction can be used, however, gene expression is lower and an autophagic response is still induced albeit to a lesser degree.

Construction of eGFP-LC3B expressing cell lines. Highly robust quantification of autophagy can
be achieved by measuring the amount of LC3B-positive vesicles, a hallmark of autophagy, using eGFP-LC3B expressing reporter cells 24,33 . To this end, we constructed autophagy reporter cell lines stably expressing eGFP-LC3B from a genomically integrated, CMV-promoter controlled, expression cassette. Third generation lentiviral particles harboring the expression cassette were generated and target cell lines transduced. The cell lines include cell lines (HeLa) classically used for autophagy research, easy-to-transfect cell lines (HEK293T) and cells of the immune system like Monocyte-like and T cells (THP-1 and Jurkat) (Fig. 1A). Following transduction, single-cell clones were sorted from a pool of medium level eGFP-LC3B expressing cells (exemplarily shown for HeLa eGFP-LC3B cells in Fig. 1B). These single-cell clones were grown into clonal cell lines and eGFP-LC3B fluorescence monitored by fluorescence microscopy (Fig. 1C). Unchanged p62 levels between parental and stable cell lines suggest that autophagic flux was not significantly altered by the expression of eGFP-LC3B (Fig. S1A). Finally, expression and size of the fusion protein eGFP-LC3B (ca. 45 kDa) was confirmed by immunoblotting (Fig. 1D) Table 7. Pipetting scheme for sgRNA amplification PCR.

Cycle # Temperature (°C) Duration
Initial denaturization 1 98 3 min  Fig. 2A, B) displayed the characteristic puncta of eGFP-LC3B-positive autophagosomes in addition to a diffuse cytoplasmic signal. To confirm that the reporter cell lines respond to stimulation and blockage of autophagy, they were treated with different drugs that induce (Rapamyicin) or block autophagic flux (Chloroquine, Bafilomycin A1). Whereas mock-treated cells show a low number of autophagosomes (= eGFP-LC3B puncta in the cytoplasm), the number of puncta visibly increased upon Rapamycin treatment. Chloroquine and Bafilomycin A1 treatment led to an accumulation of perinuclear autophagosomes (Fig. 2B). The pixel area of eGFP-LC3B, which correlates with autophagy levels, was quantified using semi-automated analysis www.nature.com/scientificreports/ (Fig. 2C). Confocal images were taken and randomly selected single cells extracted. Aided by an ImageJ macro 50 , automatic thresholding, and particle counting resulted in the pixel area of autophagosomes ( Fig. 2A, C). Western blotting confirmed that p62 accumulates in the reporter cell lines upon Bafilomycin A1 treatment, demon- www.nature.com/scientificreports/ strating functional autophagy (Fig. S1B). Taken together, the generation of eGFP-LC3B expressing autophagy reporter cell lines derived from HeLa, HEK293T, Jurkat, and THP-1 cells was successful.

Rapid quantification of autophagosomes using eGFP-LC3B expressing cells.
For high-throughput applications, an efficient system to quantify LC3B-positive autophagosomes is desirable. To measure autophagy using flow cytometry, eGFP-LC3B-II decorated autophagosomes have to be separated from cytosolic eGFP-LC3B-I. Thus, eGFP-LC3B expressing cells were permeabilized with 0.05% saponin containing PBS, the cytoplasm subsequently washed out (Fig. 3A) and only autophagosome-bound eGFP-LC3B-II retained inside the cells 24,33 . As determined by flow cytometry, the cells decreased in size and granularity (Fig. 3B, upper  panel). Complete permeabilization as indicated by cell size is reached after 10 min (Fig. S1C). Furthermore, after successful removal of cytoplasmic eGFP-LC3B-I, the mean eGFP fluorescence levels are drastically reduced (Fig. S1D), representing only fluorescence of autophagosome-bound eGFP-LC3B-II (Fig. 3B, lower panel). Therefore, treatment with saponin allows quantification of changes in autophagosome numbers, which are not visible in non-permeabilized cells (Fig. 3C). The washout procedure is very robust, and only minor variations are observed in the treatment (Fig. S1E). It is possible to fix the cells to preserve the eGFP-LC3B signal after saponin permeabilization for longer storage using two different methods: PFA and MeOH. Compared to non-fixed cells, the signal in PFA-fixed cells was well preserved, MeOH fixation, however, caused a slight drop of the absolute eGFP-LC3B signal (Fig. 3D). However, as the differences between differently treated cells were preserved, both fixation methods were suitable. Thus, isolation of autophagosome-bound eGFP-LC3B and its detection using flow cytometry was successful and can be used to reveal changes in autophagy levels.
High-throughput discovery of compounds that modulate autophagy. Drug screenings, as exemplified for the known autophagy manipulating drugs Rapamycin, Chloroquine, and Bafilomycin A1, can be readily performed (Fig. 4A-D) using the autophagy reporter cell lines (HeLa-, HEK293T-, Jurkat-and THP-1 eGFP-LC3B). Dotted red lines represent twice the standard deviation of the mock control to illustrate the sensitivity of the approach. Induction of autophagic flux by Rapamycin was visible even at low nanomolar (< 15 nM) concentrations. Accumulation of autophagosomes induced by treatment with Chloroquine or Bafilomycin A1 was detected significantly above background at very low drug concentrations (< 1.25 µM for Chloroquine, < 78.1 nM for Bafilomycin A1), illustrating the sensitivity of the approach. For all treatments, Z-factors ranging between 0.29 and 0.91 were calculated. A Z-factor above 0 indicates high robustness of the high-throughput approach 42 , suggesting that the method is robust enough for high-throughput applications. As a proof-of-principle whether our approach can be used to detect novel autophagy modulating compounds, we assessed the impact of 18 different human amino acids on eGFP-LC3B levels. Our results indicate that whereas amino acids like cysteine, isoleucine, asparagine, serine, valine or threonine may induce autophagy, others like arginine, tyrosine or glycine slightly reduce autophagic flux (Fig. 4E). This is in accordance with previous reports that indicate a role of amino acids in the modulation of autophagy [51][52][53] . Taken together, all cell lines that were generated responded accurately, robustly, and sensitively to drug treatment. Thus, this approach is suitable for high-throughput quantification of autophagy to discover novel compounds that modulate autophagy (see 3.1.).

Monitoring modulation of autophagy by viral infections. Induction/reduction of autophago-
somes upon viral infection was resolved in a time-dependent manner using infected HeLa eGFP-LC3B cells (Fig. 5A) 56,57 . In monocyte-like cells, both IAV and MeV infection induced autophagy at early time points (Fig. S2A). Infection with encephalomyocarditis virus (EMCV) rapidly induced high levels of autophagosomes 10,58 (Fig. 5A). In summary, time-dependent changes in autophagy induced by viral infection can be accurately monitored on a small scale (96-well) using our system. Furthermore, samples with higher biosafety levels (above BSL1) can be easily processed and fixed, and then safely analyzed in BSL1 conditions (see 3.2.).

Identification of cellular factors modulating autophagy. Autophagy induction by protein trans-
fection into HEK293T eGFP-LC3B cells can be rapidly assessed as exemplified by inducing autophagy with TRIM32 overexpression 59 (Fig. 5B in HEK293T reporter cells or Fig. S2B in HeLa reporter cells). Thus, it is possible to screen whole libraries of proteins for autophagy induction (see 3.3.). Similarly, knockdown or knockout libraries using either siRNA or CRISPR/Cas9 can be approached. CRISPR/Cas9 mediated knockout or siRNA mediated knockdown of core components of the autophagic machinery (ATG proteins) led to a clear decrease of the eGFP-LC3B signal ( Fig. S2C and D). CRISPR/Cas9 screens require isolation of cell populations that display altered autophagy levels. Subsequently the integrated sgRNA cassettes are amplified and analyzed by next-generation sequencing (NGS) to compare the abundance of sgRNA sequences in the 'sample' to the control population ('input') and identify target enrichment. We successfully extracted genomic DNA from Jurkat eGFP-LC3B cells after processing and MeOH fixation as described in our basic protocol. Subsequent PCR analysis revealed an amplicon at the correct size (~ 270 nt) that was only present in the sample transduced with a CRISPR Lentivirus library (Human CRISPR Knockout Pooled Library (GeCKO v2) 37,49 , Fig. 5C). As a proofof-principle for CRISPR/Cas9 screens to identify novel key factors in autophagy in T cells, we transduced Jurkat eGFP-LC3B cells with the Human CRISPR Knockout Pooled Library (GeCKO v2). The samples were processed and fixed with MeOH according to our basic protocol. FACS sorting isolated high, low and medium eGFP-LC3B (= autophagosome) containing cells. 8% of the CRISPR treated cells displayed higher autophagosome content than non-targeting sgRNA infected cells and 10% showed lower levels of eGFP-LC3B (Fig. 5D). Genomic DNA Scientific RepoRtS | (2020) 10:12241 | https://doi.org/10.1038/s41598-020-68607-w www.nature.com/scientificreports/ was extracted from 'low' and 'high' fractions as well as from the control population, 'input' . The sgRNA cassette was amplified by PCR (Fig. S2E). The amplicons of 'low' and 'input' from three independent experiments were pooled and sequenced using NGS to identify the contained sgRNAs. From a total 119,461 individual sgRNAs in the original GeCKO library, we could obtain sequences for 97.92% (Fig. S2F), demonstrating that complexity was retained. Knockout of components of the autophagic machinery should lower autophagosome levels. In line with this, sgRNAs targeting ATG genes (6 per gene) were significantly enriched in the 'low' population on average (Fig. S2G). Analysis of the aggregated sgRNA counts showed that the counts for large majority of ATG genes were higher in the 'low' fraction compared to the 'input' (Fig. 5E). This confirms that the method is able to identify components of the autophagic machinery. Taken together, these assays revealed that our method is suitable for high-throughput overexpression or CRISPR/Cas9 mediated KO approaches to discover novel key factors in autophagy (see 3.4.).

Discussion
High-throughput quantification of autophagy. A wide variety of methods to reliably quantify autophagy is currently available (for a comprehensive review see 24 ). These methods include visualization of autophagosomes by electron microscopy, monitoring of degradation of targets of autophagy such as p62 using western blotting, processing of endogenous LC3B, and visualization of LC3B puncta using eGFP-LC3B and confocal fluorescence microscopy. Advanced imaging methods using automated image processing reduce the manual labor required for confocal image acquisition and analysis 50 . Several currently used methods to quantify autophagy rely on monitoring a hallmark of autophagy induction, the lipidation and translocation of (eGFP-LC3B) to autophagosomal membranes. However, most of these methods are not applicable for high-throughput approaches. Here, we describe a detailed protocol for an easily accessible method to robustly and rapidly quantify autophagosomes for high-throughput applications based on flow cytometry-mediated quantification of membrane (= autophagosome)-bound eGFP-LC3B. Compared to classical methods for monitoring autophagy, like western blotting, this system is less labor-intensive, faster, and allows the quantification of large numbers of cells (10,000 vs 50-100) at once, allowing extraction of robust means of autophagy levels but also visualization of the heterogeneity of cell systems. Therefore, this system is well fitted for approaches that measure the mean induction of autophagy by e.g. drugs or peptides, but also approaches relying on single-cell autophagy levels like CRISPR/Cas9 screens (Figs. 4 and 5). Proof-of-principle assays revealed that novel autophagy modulating compounds like e.g. human amino acids can be readily identified. Alteration in autophagy levels due to viral infections can be assessed over time. Our data further reveal that overexpression approaches monitoring the autophagy response of individual cells are possible. Proteins above a size threshold of approximately 50 kDa can be easily co-stained, allowing e.g. overexpression screenings. To avoid washout of the co-stained protein, it may be anchored via a tag (e.g. GPI anchor) to a membrane and thus be retained in the cell 60 . Lastly, the system is ready for CRISPR/Cas9 KO screens to identify novel factors modulating autophagy in different tissues. We could isolate populations with different autophagy levels in reporter T cells that were transduced with the GeCKO v2 CRISPR/Cas9 pooled library. sgRNA amplicons were extracted and analyzed by NGS. Taken together, we demonstrate that our protocol is suitable for a wide range of high-throughput approaches that can be applied to answer various scientific problems, ranging from pathway analysis and key factor identification to drug discovery.
Limitations of the method: autophagic flux vs. accumulation of autophagosomes. Autophagy is a highly dynamic process, relying on the complex turnover and activation of signaling cascades. While the reporter cell lines allow fast processing of samples and quantification of autophagy, its use is limited to genetically modified cell lines. The consistent presence of the reporter is mandatory and transiently transfected cell lines do not provide adequate stability of the signal. Thus, the screen should be complemented with monitoring the endogenous LC3B status in primary cells to support conclusions indicated by initial screening methods 24,61 .
One major issue of all systems relying on the processing of LC3B is that both de novo induction of autophagy and blockage of autophagic flux increases the amount of autophagosomes (or processed LC3B) at a given time point. Thus, a rigorous secondary assessment of hits obtained in the primary screen has to take place (for a comprehensive review of methods see 24 ). Complementary assays to monitor autophagy, that do not rely on LC3B are mandatory to properly assess the status of autophagy in a cell. For example, the cellular levels of SQSTM1/p62 are decreased upon induction of autophagic flux but p62 accumulates if autophagy is blocked 62,63 . Monitoring p62 levels will provide further hints whether a compound/molecule induces or blocks autophagy 62 . Alternatively, other cellular proteins targeted by autophagy like NBR1 64 or cGAS 65 can be used as indicators of autophagic degradation. Autophagy-like cellular processes such as LC3-associated phagocytosis 66 or viruses may use LC3B and redirect it to membranes, thus giving rise to false-positive signals in our flow cytometry assay and other methods that measure autophagy using LC3B processing or localization 3,54,66 . In general, for the majority of noncanonical autophagic processes, degradation of p62 is not observed. Furthermore, the activity of the compound/ factor should be dependent on multiple essential factors of the core machinery of autophagy such as ATG5, ATG4, and ATG16L. For example, while LC3B associated phagocytosis is dependent on most core machinery factors, it is independent of e.g. ATG14L, which is required for canonical autophagy 3,66 . Finally, visual analysis of double-layered membrane structures by electron microscopy will support the notion of whether a compound/ regulatory factor modulates classical autophagy 67,68 . Comparison to other high-throughput approaches to measure autophagy. Besides relying on eGFP-LC3B, other high-throughput methods to quantify autophagy are available. Cell lines expressing the double fluorescence reporter fusion to LC3B (mRFP-eGFP-LC3B) allow rapid quantification of autophagy without Scientific RepoRtS | (2020) 10:12241 | https://doi.org/10.1038/s41598-020-68607-w www.nature.com/scientificreports/ cytoplasmic washout 69 . Upon induction of autophagy, eGFP fluorescence is lost in the acidic environment of autophagolysosomes, thus the ratio between eGFP and mRFP fluorescence decreases. While this method is elegant, it was less sensitive in our hands and required complicated compensation during flow cytometry, which may increase rates of false-positive/false-negative results. Occasionally, eGFP fluorescence is not completely quenched by the acidic pH, resulting in remaining signal, thus alternative fluorophores have been proposed 70 . Still, recently novel autophagy regulating factors like TMEM41B 30 were discovered using mRFP-eGFP-LC3B reporter constructs 24 .
Recently, another sophisticated approach to quantify autophagic flux using flow cytometry was described 26 . Using a reporter construct expressing a fusion protein of GFP, LC3, RFP, and a non-cleavable variant of LC3B, LC3BΔG, this system allows, similarly to the double-labeled LC3B, monitoring of autophagy based on the ratio between GFP and RFP fluorescence. Upon autophagy induction, the protease ATG4 is activated and cleaves the fusion protein into GFP-LC3and RFP-LC3BΔG. Eventually, only GFP-LC3B can be incorporated into autophagosomes and is degraded. Thus, upon induction of autophagy, the ratio between GFP and RFP fluorescence decreases. This system has allowed the identification of several novel autophagy modulators 26 . However, recombination between the two LC3B ORFs, especially during lentiviral driven applications like CRISPR screens, may render the system inactive.
Instead of assessing the number of autophagosomes to quantify autophagy, the consequences of the induction of autophagic flux can also be monitored by examining p62 levels. Based on this approach several genetic screens have identified novel factors involved in autophagy regulation, such as 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 29 , the ufmylation pathway 28 , or the acetyltransferase EP300 27 . However, upon induction of autophagy, transcriptional upregulation of p62 has been observed in some instances. Furthermore, longer assay times are needed to allow the degradation to proceed enough to detect decreased levels of p62.
Lysosomal dyes like lysotracker may be used to stain lysosomes or autophagolysosomes and even to monitor changes in pH 24,[71][72][73] . However, as acidification of lysosomes is not a process exclusive to autophagy induction, such approaches may lead to high false positive/false negative rates. Thus, systems based on lysosomal dyes are rarely used to monitor autophagy.
Taken together, using eGFP-LC3B as a reporter for quantification of autophagy is currently still the most established and advantageous system for quantifying autophagy. It avoids convoluted reporter systems and thus directly quantifies autophagosomes. Our method applies this established tool for high-throughput approaches.

Concluding remarks
Accurate and robust quantification of autophagosomes by measuring the mean fluorescence intensity of membrane-bound eGFP-LC3B is a powerful high-throughput tool to study autophagy. Our protocol is designed to easily approach e.g. drug discovery, key factor identification, pathway analysis, and virus/host response monitoring. The setup is easy to adopt and provides a robust, flexible, and rapid readout. However, screening results have to be confirmed using orthogonal methods for assessing autophagy levels. We strongly believe our highthroughput approaches may pave the way for the discovery of novel compounds modulating autophagy and provide an immediately accessible and thoroughly tested system for labs to test their compounds/proteins/ viruses for autophagy modulation.