Development of a Phenotypic High-Content Assay to Identify Pharmacoperone Drugs for the Treatment of Primary Hyperoxaluria Type 1 by High-Throughput Screening

Abstract

Primary hyperoxaluria is a severe disease for which the best current therapy is dialysis or organ transplantation. These are risky, inconvenient, and costly procedures. In some patients, pyridoxine treatment can delay the need for these surgical procedures. The underlying cause of particular forms of this disease is the misrouting of a specific enzyme, alanine:glyoxylate aminotransferase (AGT), to the mitochondria instead of the peroxisomes. Pharmacoperones are small molecules that can rescue misfolded proteins and redirect them to their correct location, thereby restoring their function and potentially curing disease. In the present study, we miniaturized a cell-based assay to identify pharmacoperone drugs present in large chemical libraries to selectively correct AGT misrouting. This assay employs AGT-170, a mutant form of AGT that predominantly resides in the mitochondria, which we monitor for its relocation to the peroxisomes through automated image acquisition and analysis. Over the course of a pilot screen of 1,280 test compounds, we achieved an average Z′-factor of 0.72±0.02, demonstrating the suitability of this assay for HTS.

Introduction

Primary hyperoxaluria (PH) is a rare autosomal recessive hereditary disorder responsible for an excessive production of oxalate that progressively accumulates as calcium salts and forms renal and/or bladder stones.¹ As it progresses, the disease can lead to chronic kidney failure and systemic oxalosis (deposition of calcium oxalate throughout the body).^2,3 Deficiencies in the liver-specific peroxisomal enzyme, alanine:glyoxylate aminotransferase (AGT, EC 2.6.1.44), have been linked to Type 1 PH.⁴ Under normal circumstances, AGT catalyzes the transamination of glyoxylate to glycine within the peroxisomes of hepatocytes. In PH1 patients, who represent 80% of the PH population, the AGT deficiency causes the glyoxylate to diffuse from the peroxisomes into the cytosol, where it is oxidized to insoluble calcium oxalate by the lactate dehydrogenase.^1,5 Missense mutations in the AGT gene (AGXT) that lead to the mistargeting of AGT from its normal peroxisomal location to the mitochondria have been identified.^6,7 Although mistargeted, AGT is still catalytically active in the mitochondria, but metabolically ineffective in this location. The best therapies currently rely on dialysis or liver transplantation; both represent inconvenient, risky, and costly procedures. Hence, cellular rescue of AGT misrouting through small-molecule intervention is a promising and potentially safer therapeutic alternative.

Pharmacoperones are cell-permeable small molecules that serve as templates, which correctly fold and route otherwise misfolded/mistrafficked mutant proteins. In particular, G protein-coupled receptors (GPCRs) have been heavily exploited in HTS in terms of pharmacoperone discovery with promising lead scaffolds emerging for the vasopressin 2 receptor (V2R) and the gonadotropin-releasing hormone (GnRH) receptor (GnRHR). The functional rescue of these proteins by pharmacoperones has been reported by our groups and others.^8

–13 The discovery of target-specific pharmacoperones by high-throughput screening (HTS) requires the design of automation-friendly, microtiter plate-compatible assays.¹⁴ Previously reported assays measuring AGT activity are not fully compatible with HTS requirements or are effectively too far removed from the phenotypic and physiological relevance of this target.^15
–17 In this study, we describe a phenotypic microscopy-based assay that detects the colocalization of AGT with the peroxisomes in a mammalian cell-based system expressing a pathophysiologically relevant mislocated mutant form of AGT, AGT-170.^16,18 In the present work, we have successfully miniaturized the assay to a 384-well plate format and demonstrate its HTS readiness through a small-scale pilot screen that paves the way for interrogating larger small-molecule libraries in a fully automated manner.

Materials and Methods

Cell Culture

The CHO-GO (glycolate oxidase) cell lines expressing the AGT-mi and -170 variants were described elsewhere.⁶ Cells were routinely cultured in Ham's F12 medium supplemented with 10% fetal bovine serum (Hyclone), 100 units/mL penicillin, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin B (antibiotic-antimycotic mix; Gibco), as well as 400 μg/mL zeocin and 800 μg/mL G418 (Life Technologies) for selection.

Compound Library

A small set of 1,280 compounds encompassing 4 compound plates from our library was used for the pilot screen. Details regarding Scripps' Drug Discovery collection can be found at http://hts.florida.scripps.edu/index.php/facilities.html#Compound-Libraries. Compounds were plated as 1 mM solutions in DMSO. The final nominal concentration during the assay cell treatment was 4 μM and the final DMSO concentration was 0.4%.

384-Well Plate Assay Protocol

A detailed stepwise protocol is presented in Table 1. Briefly, AGT-170 cells (and AGTmi cells that serve as a positive control) were seeded in black, square IQ™ 100 μ 384-well plates (Aurora; Brooks) at 500 cells/wells using a Flying Reagent Dispenser (FRD; Beckman Coulter). After an overnight incubation at 37°C, 95% relative humidity (RH), and 5% CO₂, cells were treated with 100 nL test compounds or DMSO for the positive and negative wells with a pintool transfer unit (GNF). Plates were then placed in an incubator for 3 days at 37°C, 95% RH, and 5% CO₂, immunostained, and acquired on a high-content imager as described below.

Table 1.

384-Well Plate Protocol for the AGT High-Content Assay

Step	Parameter	Value	Description
1	Plate cells	25 μL	500 cells per well
2	Incubation time	16–20 h	37°C, 5% CO₂, and 95% relative humidity
3	Controls and library compounds	100 nL	4 μM final, 0.4% DMSO final
4	Incubation time	3 days	37°C, 5% CO₂, and 95% relative humidity
5	Dispense 2× fixation solution	25 μL	8% Formaldehyde in D-PBS, 4% final
6	Incubation time	30 min	Room temperature
7	Wash	3 times	With D-PBS
8	Dispense permeabilization solution	25 μL	0.1% Triton-X100 in D-PBS
9	Incubation time	20 min	Room temperature
10	Wash	3 times	With D-PBS
11	Dispense primary antibody solution	8 μL	Anti-AGT and anti-PMP70 at 1:10,000 and1:500, respectively
12	Incubation time	1 h	Room temperature
13	Wash	3 times	With D-PBS
14	Dispense secondary antibody solution	8 μL	Hoechst at 10 mg/mL and second day antibodies at 1:10,000
15	Incubation time	1 h	Room temperature
16	Wash	3 times	With D-PBS
17	Dispense D-PBS	50 μL
18	Seal plate
19	Image wells	20× objective	Record emissions at 386, 485 and 549 nm

Step Notes

1. Uses Flying Reagent Dispenser (FRD) technology.

2. Uses PinTool compound transfer technology.

5. Uses FRD technology.

7. Plates were washed using Matrical's Squirt.

8. Uses FRD technology.

10. Plates were washed using Matrical's Squirt.

11. Uses FRD technology.

13. Plates were washed using Matrical's Squirt.

14. Uses FRD technology.

16. Plates were washed using Matrical's Squirt.

17. Uses FRD technology.

19. Acquisition on Cellomics CellInsight.

AGT, alanine:glyoxylate aminotransferase; D-PBS, Dulbecco's Phosphate-Buffered Saline.

Immunostaining

Following treatment and incubation as described above, cells were fixed by the addition of 2× fixation solution (8% paraformaldehyde) using an FRD. Plates were then washed thrice with phosphate-buffered saline (PBS) on the Squirt microplate washer using a 12° nozzle tilt and 15 PSI pressure according to the user's manual for washing normal binding cells in a 384-well plate format (Brook). Cells were then permeabilized by dispensing a solution of 0.1% Triton X-100 in PBS (PBS-TX) and incubated for 10–30 min. Next, cells were stained with a guinea pig anti-AGT A1 antibody (1:10,000; Danpure's Laboratory) and a rabbit anti-PMP70 antibody (1:500; Abcam) in PBS-TX containing 0.5% goat serum. After three PBS washes, anti-rabbit AF488 and anti-guinea pig AF546 secondary antibodies (both at 1:10,000; Molecular Probes) as well as Hoechst at 10 μg/mL were added to each well. Thirty minutes later, plates were washed thrice, filled with 50 μL/well of PBS, and sealed with black tape.

Image Acquisition and Analysis

Plates were read and images acquired with the CellInsight high-content reader (Thermo Fisher Scientific, Pittsburgh, PA) using a 20× objective. A detailed and illustrated description of the image acquisition and analysis is presented in Figure 1. A maximum of nine fields of view per well were acquired, yielding an average of ≈700 cells that were detected. The nuclear stain channel (Hoechst, 386 nm, 9.8 ms exposure time) was used to focus on the cell layer. Puncta labeled with the anti-PMP70 antibody (labeling peroxisomes) and the anti-AGT antibody were acquired in the green (485 nm, 40 ms exposure) and red (549 nm, 15 ms exposure) channels, respectively. Nuclei were detected using the fixed threshold method with an intensity threshold of 300. Next, a cytoplasmic region of interest was created around each nucleus by creating a ring with a distance of one pixel from the nucleus and a width of eight pixels. Peroxisomes and AGT proteins were identified in the cytoplasmic region of interest using a fixed intensity threshold of 200 and 141, respectively. Colocalization of the two populations of puncta was determined using the Cellomics' Colocalization BioApplication; the output feature providing the Pearson's correlation coefficient between the two targets is called ROI_A(B)_Correlation Coef. This feature, which shows values ranging from −1 to 1, describes the degree of overlap of the two targets with each other independent of the intensity differences of the two targets, and is calculated using the following equation: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} r_p = \frac {\Sigma (x - \dot {x}) (y - \dot {y})} {\sqrt {\Sigma (x - \dot {x}) ^2 \Sigma (y - \dot {y} ) ^2}} \end{align*} \end{document}

Fig. 1.

Automated image acquisition and analysis to evaluate the colocalization of AGT with the peroxisomes. (A) Hoechst-stained nuclei, (B) anti-PMP/AF488-labeled peroxisomes, and (C) anti-AGT/AF546 immunostaining of AGT were imaged on a CellInsight high-content reader with a 20× objective. (D) The Red/Green/Blue (RGB) composite of the three aforementioned channels. Upon image processing by the Scan software, segmented nuclei (E), peroxisomes (F), and AGT (G) are recognized as such by thresholding. The nuclei outlined in orange in (E) correspond to those that were located on the edge of the field of view and were discarded from the analysis because they were only partially acquired. Upon dilatation of a cytoplasmic area of interest around each validated nucleus (H, in yellow ), the overlap between the immunostaining of the peroxisomes and AGT was calculated and reported as a Pearson's colocalization coefficient. Under our experimental conditions, AGT localization was predominantly peroxisomal with a calculated Pearson's colocalization coefficient of 0.88±0.04 (n=144 cells). Note that these images correspond to ∼1/9th of a single field of view. AGT, alanine:glyoxylate aminotransferase.

where r_p is the Pearson's correlation coefficient, x and y are pixel intensities of each pixel detected in the AGT and peroxisome channels, respectively, and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\dot{x}$$ \end{document} and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\dot{y}$$ \end{document} are average pixel intensities of the puncta identified as AGT and peroxisomes, respectively.

Images and data were automatically spooled to the STORE database (Thermo/Cellomics) hosted on Scripps's HTS server.

Data Management and HTS Data Analysis

The relevant well feature generated by the detection algorithm, named the MEAN_ROI_A_Correlation Coef, was exported from the STORE database as a tabulated file using a custom Simple Object Access protocol (SOAP) web service utilizing Thermo's high-content screening connect API. Reader files were then uploaded into Scripps' Drug Discovery database (Symyx Assay Explorer). Plate Z-factor, Z′-factor, and sample-to-background ratio (S/B) were automatically calculated as previously described.^19,20 Assay results were normalized on a per plate basis using the following equation: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \%rescue = 100 \times \frac { Test \ Well - Median \ Low \ Control } { Median \ High \ Control - Median \ Low \ Control } \end{align*} \end{document}

where High Control represents the well containing AGTmi cells treated with DMSO (n=32) and Low Control represents the well containing AGT-170 cells also treated with DMSO (n=16).

Results

Development of an HTS-Compatible Microscopy-Based Assay to Monitor AGT-Peroxisome Colocalization

Previously, we described the use of stably transformed CHO cells to study the subcellular distribution of normal and various PH1 mutant AGT constructs by immunofluorescence microscopy.⁶ Among these mutants, the AGT-170 variant presented the most severe peroxisome-to-mitochondrion mistargeting phenotype. Accordingly, we incorporated a cell line expressing this specific mutation to design a high-content high-throughput assay to identify potential pharmacoperones able to rescue AGT mistargeting by rerouting it to the peroxisomes. To be compliant with HTS requirements, the original assay had to be simplified, adapted to microtiter plates, and automated.

We first investigated our ability to detect and measure AGT/peroxisome colocalization with our HTS-compatible, wide-field high-content imager by labeling CHO cells expressing either the AGT-170 PH1 mutant or the AGT minor allele for nuclei, peroxisomes, and AGT in a 96-well plate format. Acquisition was made using a 20× objective. A dedicated detection algorithm was defined that automatically calculates and reports the Pearson's colocalization coefficient between the AGT- and peroxisome-labeled areas for each well (Fig. 1). Figure 2 shows images from a representative well of each of the two cell lines as acquired with the high-content reader. Upon automated analysis, the Pearson's colocalization coefficient was determined at 0.85±0.08 for AGTmi cells and 0.56±0.15 for AGT-170 cells, values that are in accordance with those reported in our previous work using a confocal laser-scanning fluorescence microscope, a 60×objective, cells fixed on glass coverslips, and the ImageJ JACop plugin.⁶ This indicated that a lower magnification wide-field instrument geared toward higher throughput was also able to properly measure the biology of interest.

Fig. 2.

Differential localization of the AGTmi and AGT-170 proteins with regard to the peroxisomes. RGB composite images of CHO-K1 cells stably expressing the AGTmi (A) or AGT-170 (B) protein upon acquisition on the CellInsight high-content reader using a 20× objective. Nuclei are shown in blue, AGT in red, and the peroxisomes in green. Yellow denotes the area where both AGT and peroxisome immunostaining are present, that is, colocalization. (C) The individual calculated Pearson's correlation coefficient for all cells identified in a single field of view of CHO-K1 expressing either AGTmi (n=110 cells, Pearson's=0.85±0.08) or AGT-170 (n=103 cells, Pearson's=0.56±15). A two-tailed t-test comparison between the two cell populations yielded a P value <0.0001, indicative of an extremely significant difference between both.

To assess our ability to induce and detect an increase in the Pearson's colocalization coefficient, we treated both AGTmi and AGT-170 cells with glycerol, a viscous polyol compound known to enhance the stability of proteins in solution.²¹ This cosolvent shifts the native protein to more compact states and inhibits protein–protein aggregation while proteins are being refolded during biological synthesis. Accordingly, treatment with increasing concentrations of glycerol resulted in a marked increase of Pearson's colocalization coefficients for the AGT-170 cells changing from 0.41±0.01 to 0.80±0.04 (n=4), that is, a 1.91-fold increase as shown in Figure 3A. Notably, AGTmi cells were also showing an increase in Pearson's values upon glycerol treatment, although this increase was marginal (∼1.10-fold increase). Concentrations of glycerol higher than 5% affected cell viability prevented accurate determination of the Pearson's colocalization coefficient; for these reasons, they were avoided. To probe the robustness of the assay, we repeated the aforementioned glycerol titration assay on two separate plates. The plot presented in Figure 3B shows the Pearson's coefficient determined from multiple wells treated with varying glycerol concentrations ranging from ≈0.4 to ≈0.9 and indicates that data generated from the two separate plates were virtually identical, yielding an R ² greater than 0.99. The Z′-factor, a statistical measurement indicative of HTS readiness,²⁰ was calculated between the AGT-170 cells in the absence and presence of 5% glycerol and found to be 0.58 and at 0.81 between untreated AGT-170 and AGTmi cells.

Fig. 3.

Effect of glycerol on the colocalization of AGT with the peroxisomes. CHO-K1 cells expressing either AGTmi (gray bars) or AGT-170 (white bars) were cultivated for 4 days in 96-well plates in the presence of different concentrations of glycerol ranging from 0 to 5%. Images were then acquired and the Pearson's correlation coefficient calculated (A, n=4 wells for each condition). Verification of the fidelity of the AGT assay is shown by repeating the glycerol treatment over two successive plates, noted A and B, and comparing the obtained values in the scatter plot (B). The calculated R ² is 0.9955, indicative of an excellent correlation between the results generated for the two successive assay plates.

Miniaturizing and Automating the AGT-170 Cell-Based Assay to the 384-Well Plate Format

With Z′-factor values greater than 0.5, the assay warranted further miniaturization and automation to the 384-well plate format. Whereas the majority of HTS assays that are homogeneous in format typically rely on three dispense steps (cell addition, compound delivery, detection reagents dispense), and immunostaining-based assay protocols usually require a minimum of eight dispenses, including aspiration and/or wash steps. A primary concern was that automated liquid handling devices will disrupt the cell layer during this procedure. To prevent this from occurring, we chose to fix the cells as early as possible in the protocol by replacing the initial media aspiration step with a dispense of a 2× concentrated fixation solution. In addition, we implemented a noncontact plate washer that does not rely on aspiration to empty the wells, but instead utilizes an air blade that ejects liquids from the well with little to no residual volume (Squirt; Brooks). To verify that cells remained in the wells and were not being dislodged during the plate preparation process, Hoechst staining was added during the fixation step of a mock immunostaining protocol and nuclei were counted at each step. The results presented in Figure 4 indicate that the optimized dispensing and washing conditions assured little to no cell detachment over the course of the plate preparation process. These results also indicated that cells seeded at densities higher than 600 cells per well reached total confluence; a cell seeding density of 500 cells per well was considered optimal and used for the rest of this study. Volumes, concentrations, and incubation times were optimized to offer the best balance between cost per well, time, and assay performance. The resulting miniaturized assay protocol is presented in Table 1.

Fig. 4.

The effect of multiple dispense/wash steps using noncontact, automated instruments on the cell layer integrity of CHO-K1 cells cultivated in a 384-well plate. CHO-K1 cells were plated in 384-well plates at different cell densities ranging from 100 to 1,000 cells per well (64 wells per condition) and incubated for 4 days. In this case, cells were immediately fixed and their nuclei labeled by the addition of 2× fixation solution (8% paraformaldehyde) containing 20 μg/mL Hoechst using an FRD. After each step, as indicated in the stepwise protocol in Table 1, the cell number was determined using a CellInsight high-content reader after imaging a single field of view with a 5× objective. FRD, Flying Reagent Dispenser.

Validation for HTS Purposes

To verify that the assay was compatible with HTS requirements, we first investigated the potential for position effects by running plates treated with DMSO only. A heat map of a representative plate treated with DMSO is shown in Figure 5; this plate did not show any edge effect or position effect within the sample field, indicating that despite an extended incubation time (4 days), all variables that can lead to well-to-well variability (such as temperature gradient, evaporation, and contamination) were tightly controlled. The coefficients of variation of 6.12%, 1.17%, and 7.47% for the sample field, high control, and low control, respectively, were well under an empirically accepted maximum of 10%. The Z′-factor was 0.62, demonstrating the miniaturized AGT assay's robustness and readiness for drug screening purposes. In addition, the lack of observable outliers within the sample field in general indicates this assay should be fairly devoid of false positives or negatives. The assay does have at least some indication of DMSO sensitivity as observed in column 24 of Figure 5 with the normalized data displaying a negative % rescue for nontreated wells.

Fig. 5.

Example of raw and normalized data from a representative DMSO plate. Five hundred CHO-K1 cells expressing either AGTmi or AGT-170 were plated in a 384-well plate and incubated overnight. The following day, all wells, but the last column, were treated with DMSO (0.9% final concentration) using a pintool transfer unit. After 3 days, plates were prepared for automated high-content imaging. (A) The heat map of the calculated Pearson's correlation coefficient for each well of the plate as well as the plate map. (B) A scatter plot that shows the results upon normalization, using column 1 and 2 as the high control (▵, DMSO-treated AGTmi cells, 100% rescue) and column 23 (▿, DMSO-treated AGT-170 cells, 0% rescue). In addition, DMSO-treated wells within the sample field (○) and wells that were not treated with DMSO (⋄) are shown. Horizontal bars represent the mean percent rescue for each value. Note that this value is always comprised within the±10% rescue interval delineated in gray. The %CV of the sample field was 6.12% and the one of the high and low controls was 1.17% and 7.47%, respectively. The calculated S/B was 1.76 and the Z′ value was 0.62. CV, coefficient of variation; S/B, sample-to-background ratio.

Pilot Screen

To assess the performance of the AGT assay under HTS conditions, a pilot screen against a set of 1,280 diversified molecules was conducted. The final concentration at which the compounds were tested was 4 μM, which corresponds to a final DMSO concentration of 0.4%. A total of six plates were used for this assay, comprising four separate compounds plates and two DMSO plates, one at the beginning and at the end of the run. The S/B, Z-, and Z′-factors of each plate are reported in Figure 6A. With an average Z′-factor of 0.72±0.02 and a steady S/B at 1.81±0.09, the pilot screen exhibited satisfactory assay statistics and the control wells excellent plate-to-plate reproducibility, which warranted further analysis. To identify compounds, which caused a significant relocation of AGT-170 to the peroxisomes (i.e., significant percent rescue), two values were calculated: (1) the average percent rescue of all test compounds and (2) three times the standard deviation of the percent rescue of all test compounds. The sum of these two values, (1) and (2), was used as a cutoff parameter and calculated at 29.43% rescue [with (1)=3.75% and (2)=3×8.59%=25.68%]. Compounds showing a percent rescue that was higher than the hit cutoff were considered as hits (4 compounds were identified out of 1,280, that is, a 0.31% hit rate).

Fig. 6.

AGT assay pilot screen results. (A) The calculated Z′-factor and S/B of the 6 plates part of the pilot screen. The average Z′-factor and S/B were 0.72±0.02 and 1.81±0.09, respectively. (B) Scatter plot of the normalized results of the pilot screen showing the % rescue calculated for the high and low control wells (▵ and ▿, respectively), as well as for the sample field wells (•). The hit cutoff was calculated at 29.43% inhibition and is represented by the dotted line on the graph.

Discussion

The PH1-triggering AGT-170 mutation has been shown to be responsible for the misrouting of the enzyme from the peroxisomes to the mitochondria. Although functional in the mitochondria, the enzyme must reside in the peroxisomes to properly convert toxic glyoxylate into glycine. Failure to do so results in the accumulation of glyoxylate in the peroxisomes, which is then transported to the cytoplasm where it is converted to oxalate by the lactate dehydrogenase, potentially leading to kidney failure. Pharmacoperones are small molecules that serve as molecular templates to help mutant misfolded proteins to properly fold and route correctly within the cell. Since the enzymatic activity of AGT-170 is maintained despite the mutation, its correct rerouting to the peroxisome is considered a promising therapeutic option. The use of small-molecule pharmacoperones would be preferred to more controversial strategies such as gene therapies aiming at providing a nonmutated version of AGXT ²² or those aiming at pharmacologically modulating the activity of protein chaperones such as the HSPs²³ that are involved in the proper folding of hundreds of other proteins.

In this report, we describe the design, miniaturization, and validation of a cell-based HTS assay able to monitor the rerouting of the AGT-170 mutant protein to its correct location, the peroxisomes. Automated high-content microscopy was considered the technology of choice to accurately measure AGT rerouting; it was indeed favored over other technologies relying on reporter systems that usually require creating fusion or tagged proteins and can potentially lead to assay artifacts.²⁴ In contrast, using immunodetection allowed us to use AGT proteins that are devoid of any modifications and hence represent exact carbon copies of those found in patients.

Surprisingly, it appears that the use of a wide-field microscope yields enough spatial resolution to be able to clearly map the AGT location with regard to the peroxisomes. Even though our previous work relied on four acquisition channels (nuclear stain, anti-AGT and peroxisome immunostaining, and MitoTracker Red), we reasoned that for HTS purposes, the MitoTracker channel could be removed since it is now well documented that the AGT-170 mutant protein is predominantly located in the mitochondria. The goal is indeed to confirm its rerouting to the peroxisome upon treatment with a hit compound rather than its departure from the mitochondria.

In addition to the biology of interest and without any additional plate preparation, high-content imaging offers the opportunity to capture additional cell features, such as cell count, nuclear intensity, morphology, and condensation, which can help identify undesirable cytotoxic compounds early during the screening process.

As noted, DMSO itself was shown to have an effect on the % rescue in our assay (Fig. 5A). The fact that all test compounds were prepared in DMSO and that the low control 0% rescue wells were also receiving an equivalent DMSO concentration demonstrates that DMSO should not interfere with our ability to detect compounds able to reroute AGT to the peroxisomes.

An explanation for the low hit rate as well as the relatively low % rescue of the four hits identified is that the exposure time to compounds was potentially too long. We may reasonably assume that potentially active compounds will degrade over such an extended period of time, masking their effect. Another important consideration is that AGT already synthesized and mistargeted to mitochondria may not be affected by pharmacoperones; the latter can indeed only affect newly synthesized AGT before it has been imported. Therefore % rescue values are most likely underestimated, and the effect of potential pharmacoperones might be more effective in a real-life situation than they first appear. Regardless, time-course experiments with DECA, a mitochondria translocation blocker we recently identified during a chemical genetic screen on yeast,¹⁷ should help us determine in an unbiased way the optimal incubation time that is relevant to the biology we are interrogating. In addition, we plan on further miniaturizing the assay to the 1,536-well plate format, as routinely performed in our laboratory and by others,²⁵ to reduce reagent and consumable costs by roughly four-fold.

In conclusion, we demonstrated the implementation of a novel, cost-effective, and robust, miniaturized high-content assay for the discovery of pharmacoperones that can rescue an enzyme-trafficking defect involved in PH1. The protocol we designed yielded satisfactory assay statistics and demonstrated its compatibility with HTS requirements during a pilot screen. Future efforts will be focused on the integration of additional well features that can help annotate the potential deleterious effect of test compounds and further miniaturization to the 1,536-well plate format. The assay presented herein represents an important stepping stone to further screen large libraries on robotic platforms to identify pharmacoperones able to rescue AGT-170 mistrafficking. The assay can also be easily adapted to other mistrafficked proteins regardless of their enzymatic activity. Finally, in addition to screening drug libraries, this assay can also be used to uncover genes and proteins involved in AGT trafficking regulation by interrogating cDNA or siRNA libraries.

Footnotes

Acknowledgment

This work was supported by the Oxalosis and Hyperoxaluria Foundation (OHF) grant, Development of HTS Assays for Drugs for Hyperoxaluria, to P.M.C.

Disclosure Statement

No competing financial interests exist.

Abbreviations Used

References

Danpure

: Primary hyperoxaluria type 1: AGT mistargeting highlights the fundamental differences between the peroxisomal and mitochondrial protein import pathways. Biochim Biophys Acta, 2006; 1763:1776–1784.

Fargue

, Rumsby

, Danpure

: Multiple mechanisms of action of pyridoxine in primary hyperoxaluria type 1. Biochim Biophys Acta, 2013; 1832:1776–1783.

Hoppe

, Beck

, Milliner

: The primary hyperoxalurias. Kidney Int, 2009; 75:1264–1271.

Danpure

: Molecular etiology of primary hyperoxaluria type 1: new directions for treatment. Am J Nephrol, 2005; 25:303–310.

Lorenzo

, Torres

, Salido

: Primary hyperoxaluria. Nefrologia, 2014; 34:398–412.

Fargue

, Lewin

, Rumsby

, Danpure

: Four of the most common mutations in primary hyperoxaluria type 1 unmask the cryptic mitochondrial targeting sequence of alanine:glyoxylate aminotransferase encoded by the polymorphic minor allele. J Biol Chem, 2013; 288:2475–2484.

Danpure

, Purdue

, Fryer

, et al.: Enzymological and mutational analysis of a complex primary hyperoxaluria type 1 phenotype involving alanine:glyoxylate aminotransferase peroxisome-to-mitochondrion mistargeting and intraperoxisomal aggregation. Am J Hum Genet, 1993; 53:417–432.

Conn

, Smith

, Hodder

, Janovick

, Smithson

: High-throughput screen for pharmacoperones of the vasopressin type 2 receptor. J Biomol Screen, 2013; 18:930–937.

Conn

, Smith

, Spicer

, Chase

, Scampavia

, Janovick

: A phenotypic high throughput screening assay for the identification of pharmacoperones for the gonadotropin releasing hormone receptor. Assay Drug Dev Technol, 2014; 12:238–246.

10.

Conn

, Janovick

: Pharmacoperone identification for therapeutic rescue of misfolded mutant proteins. Front Endocrinol (Lausanne), 2011; 2.

11.

Janovick

, Park

, Conn

: Therapeutic rescue of misfolded mutants: validation of primary high throughput screens for identification of pharmacoperone drugs. PLoS One, 2011; 6:e22784.

12.

Wang

, Wang

, Zhao

, Yu

, Fan

: Rescue of defective MC4R cell-surface expression and signaling by a novel pharmacoperone Ipsen 17. J Mol Endocrinol, 2014; 53:17–29.

13.

Bernier

, Lagace

, Lonergan

, Arthus

, Bichet

, Bouvier

: Functional rescue of the constitutively internalized V2 vasopressin receptor mutant R137H by the pharmacological chaperone action of SR49059. Mol Endocrinol, 2004; 18:2074–2084.

14.

Smithson

, Janovick

, Conn

: Therapeutic rescue of misfolded/mistrafficked mutants: automation-friendly high-throughput assays for identification of pharmacoperone drugs of GPCRs. Methods Enzymol, 2013; 521:3–16.

15.

Rumsby

, Weir

, Samuell

: A semiautomated alanine:glyoxylate aminotransferase assay for the tissue diagnosis of primary hyperoxaluria type 1. Ann Clin Biochem, 1997; 34 (Pt 4):400–404.

16.

Behnam

, Williams

, Brink

, Rumsby

, Danpure

: Reconstruction of human hepatocyte glyoxylate metabolic pathways in stably transformed Chinese-hamster ovary cells. Biochem J, 2006; 394(Pt 2):409–416.

17.

Miyata

, Steffen

, Johnson

, Fargue

, Danpure

, Koehler

: Pharmacologic rescue of an enzyme-trafficking defect in primary hyperoxaluria 1. Proc Natl Acad Sci U S A, 2014; 111:14406–14411.

18.

Lumb

, Birdsey

, Danpure

: Correction of an enzyme trafficking defect in hereditary kidney stone disease in vitro . Biochem J, 2003; 374(Pt 1):79–87.

19.

Madoux

, Simanski

, Chase

, et al.: An ultra-high throughput cell-based screen for wee1 degradation inhibitors. J Biomol Screen, 2010; 15:907–917.

20.

Zhang

, Chung

, Oldenburg

: A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J Biomol Screen, 1999; 4:67–73.

21.

Vagenende

, Yap

, Trout

: Mechanisms of protein stabilization and prevention of protein aggregation by glycerol. Biochemistry, 2009; 48:11084–11096.

22.

Salido

, Rodriguez-Pena

, Santana

, Beattie

, Petry

, Torres

: Phenotypic correction of a mouse model for primary hyperoxaluria with adeno-associated virus gene transfer. Mol Ther, 2011; 19:870–875.

23.

Mesa-Torres

, Fabelo-Rosa

, Riverol

, et al.: The role of protein denaturation energetics and molecular chaperones in the aggregation and mistargeting of mutants causing primary hyperoxaluria type I. PLoS One, 2013; 8:e71963.

24.

Lalonde

, Weise

, Walsh

, Ward

, Frommer

: Fusion to GFP blocks intercellular trafficking of the sucrose transporter SUT1 leading to accumulation in companion cells. BMC Plant Biol, 2003; 3:8.

25.

Norton

, Titus

, Dexter

, Austin

, Zheng

, Huang

: Automated high-content screening for compounds that disassemble the perinucleolar compartment. J Biomol Screen, 2009; 14:1045–1053.