Tebufenozide and other chemicals were purchased from Sigma-Aldrich (St. Louis, MO). DMEM medium and other cell culture supplies were obtained from Invitrogen (Carlsbad, CA). The 1536-well compound plates and black/clear assay plates were purchased from Greiner Bio-one (Monroe, NC). The CytotoxGlo protease release assay was purchased from Promega (Madison, WI), and the ATP-Lite assay was purchased from PerkinElmer (Waltham, MA).

Rat pheochromocytoma PC12 cell lines harboring HTT-Q103 or Q25 (a wild type control) fused to GFP under a tebufenozide inducible promoter were kindly provided by Dr. Eric Schweitzer [12 Varma H, Lo DC, Stockwell BR. High throughput screening for neurodegeneration and complex disease phenotypes Comb Chem High Throughput Screen 2008; 11(3 ): 238-48., 13 Aiken CT, Tobin AJ, Schweitzer ES. A cell-based screen for drugs to treat Huntington's disease Neurobiol Dis 2004; 16(3 ): 546-5.]. Cells were maintained in the phenol red free DMEM medium at 37γC under a humidified atmosphere containing 5% CO₂ and 95% air. The medium contained 5% calf serum, 5% horse serum (both from HyClone, Logan UT), 250 µg/ml geneticin, 1 ( pen/strep, and 2 mM L-glutamine. Cells were passaged when they reached 80 to 90 % confluency. The PC-12 cell lines have stringent growth conditions; if the cells become too confluent over time, the entire population will perish, and if the cells are seeded too sparsely, they will not grow efficiently. Phenol red was shown to slightly inhibit the luminescent signals so phenol red free media was chosen for the assay.

Cells were plated at a density of 1000 to 1500 cells/well in black, clear bottom, tissue culture treated and low base, 1536 well plates in 5 µl/well of complete DMEM media except geneticin using a Multidrop Combi dispenser (Thermo Scientific, Waltham, MA) and incubated for 24 hr at 37(C under a humidified atmosphere containing 5% CO₂. Tebufenozide was used at 200 nM final concentration to induce expression of Q103-GFP and Q25-GFP fusion proteins in these PC12 cell lines. Both tebufenozide and test compounds in DMSO solution were sequentially added to the assay plates at a volume of 23 nl/well using a Kalypsys pintool station. The final concentration of DMSO in assay plates was 0.46% and no significant cytotoxicity was observed in cells treated with this concentration of DMSO (data not shown). The assay plates were incubated at 37(C for 48 hours before being assayed in the following multiplexed dual read assay format.

Significant cytotoxicity and subsequent aggregation of the Q103-GFP fusion proteins were found after the induction of the fusion protein expression in the PC12 cell line. Cytotoxicity was first determined by a CytotoxGlo protease release assay kit by quantitating the activity of proteases released from impaired cell membrane. Briefly, 2ul/well of the reagent mixture from the protease assay kit was added to assay plates and the plates were centrifuged at 1500 RPM for 30 sec to remove air bubbles which would affect measurement of the luminescent signal. After 10 minutes of incubation at room temperature, the assay plates were measured in a luminescent mode of Viewlux plate reader for the activity of protease released to the media. The protease activity in the media increases with the increased damaged integrity of plasma membrane due to the cytotoxicity of protein aggregation in the Q103-GFP PC12 cell line.

Cells in the same assay plates after the above cytotoxicity measurement were then treated with 1 µl/well of Triton-X100 (0.25% final) for cell lysis that dissolved soluble proteins from the bottom of assay plates. Only the aggregated proteins that were insoluble to the detergent treatment still attached to the bottom of assay plates after this detergent treatment. The assay plates were then measured by a bottom reading mode in an Acumen Explorer reader (TTP Labtech Inc., Melbourn, UK), a laser scanning cytometer, for quantitation of the numbers of Q103-GFP aggregates. The measurement parameters in the Acumen Explorer reader included (1) excitation with a 488 nm laser and emission at 500-530 nm, (2) PMT voltage of 450 volts, and (3) the number of fluorescent objects in size of 1 to 25 µm with a signal of 6 standard deviations above background. The number of fluorescent objects was directly proportional to the severity of protein aggregation observed under a fluorescence microscope in the Q103-GFP cell line.

A library collection of 220,581 compounds was screened at four compound concentrations. All compounds were dissolved in DMSO at 10 mM as stock solutions that were serially diluted in DMSO in 384-well plates at a ratio of 1:5. The four sets of 384-well inter-plate dilution plates were subsequently reformatted into one set of 1536-well compound plates at 7 µl/well using a Cybi-well dispensing station with a 384-well head (Cybio, Inc.). The final compound concentrations in 5 µl/well assay volume ranged from 500 nM to 40 µM in our screen.

The reagent dispensing and compound transfer in 1536-well plates were handled by automated work stations. A BioRAPTR work station (Beckman Coulter, Brea, CA) using solenoid valves for dispensing was employed to deliver reagents to 1536-well plates ranging from 0.5 to 5 µl per well. A pintool compound transfer station was used to transfer 23 nl compounds in DMSO solution to the media of 5 µl/well cells in 1536-well assay plates.

The primary screening data was first analyzed using customized software developed internally [21 Wang Y, Jadhav A, Southal N, Huang R, Nguyen DT. A grid algorithm for high throughput fitting of dose-response curve data Curr Chem Genom 2010; 4: 57-66.]. Clustering analysis of active compounds from the primary screening was performed with Leadscope^® Hosted Client (Leadscope Inc., Columbus, OH). The results of concentration response curves were analyzed and plotted with the Prism program (GraphPad Software, San Diego, CA).

The Q103-GFP cell line has been used in previous screening assays as a cellular model of HD [12 Varma H, Lo DC, Stockwell BR. High throughput screening for neurodegeneration and complex disease phenotypes Comb Chem High Throughput Screen 2008; 11(3 ): 238-48., 13 Aiken CT, Tobin AJ, Schweitzer ES. A cell-based screen for drugs to treat Huntington's disease Neurobiol Dis 2004; 16(3 ): 546-5.]. We used a PC12 cell line that inducibly expresses the first 17 amino acids of exon 1 of the HTT gene with 103 glutamines fused to a GFP reporter. The induction of Q103-GFP expression in the cells causes the formation of perinuclear, fluorescent aggregates (Fig. 1A). The size and fluorescence intensity of these aggregates increase with cell incubation time in the presence of inducer. Additionally, expression of the Q103-GFP fusion protein is cytotoxic, resulting in approximately 40-50% cell death after 48 hr induction of expression of the fusion protein [13 Aiken CT, Tobin AJ, Schweitzer ES. A cell-based screen for drugs to treat Huntington's disease Neurobiol Dis 2004; 16(3 ): 546-5.]. However, the protein aggregation and resultant cytotoxicity are not observed in the cells of a control line with Q25-GFP fusion protein (Fig. 1B) after 48 hr incubation with the inducer.

Fig. (1) Schematic representation of a cell based Huntington’s disease model. (A) In a Q103-GFP PC12 cell line, induction of Q103-GFP fusion protein expression causes formation of protein aggregates and resultant cytotoxicity. Typical screening assays measure either the cytotoxicity caused by the expression of Q103-GFP fusion proteins or protein aggregates. (B) In a Q25-GFP control cell line, the Q25-GFP fusion proteins are diffusely distributed in cytosol that do not form aggregates and are not toxic to cells. The images of Q103-GFP and Q25- GFP in (A) and (B) were obtained with a fluorescence confocal microscope using a 40! objective. Nuclei (blue color) were stained with Hoechst 33342 and the GFP fusion proteins (either soluble proteins or aggregates) are shown in green color.

It is challenging to measure these protein aggregates in a high throughput screening format as all existing assays require extensive manipulation of the samples including several complicated assay steps. During the course of assay development, we observed that the fluorescent GFP aggregates in Q103-GFP cells remained on the bottom of assay plates after cell lysis by 0.25% Triton X-100. In contrast, no fluorescent signal was observed on the bottom of assay plates with the Q25-GFP control cells after the same detergent treatment (Fig. 2). The results indicate that the soluble fusion proteins were released into the solution after the detergent treatment, while the protein aggregates were insoluble to detergent and remained on the bottom of assay plates. Importantly, this finding enabled us to develop a homogenous assay for the rapid quantitation of protein aggregates without any cell wash steps. In addition, we found that our observation is similar to a previous report that the HTT polyQ aggregates were resistant to detergent treatment whereas non-aggregated polyQ proteins were solublized by detergent [14 Kazantsev A, Preisinger E, Dranovsky A, Goldgaber D, Housman D. Insoluble detergent-resistant aggregates form between pathological and nonpathological lengths of polyglutamine in mammalian cells Proc Natl Acad Sci USA 1999; 96(20 ): 11404-9.].

Fig. (2) Differentiation of the fluorescent signals from aggregated Q103-GFP and non-aggregated G25-GFP fusion proteins by a laser scanning cytometer plate reader before and after detergent treatment. Microscopic images were captured using a 20! objective on an inverted epifluorescence microscope. In Q103-GFP cells, the detergent treatment did not disrupt the protein aggregates attached to the bottom of an assay plate, while soluble proteins in Q25-GFP control cells diffused away from the bottom of the plate and were not visible after detergent treatment. (A) In Q103-GFP cells, the concentration response curves of the inducer tebufenozide were similar in the presence and absence of detergent. (B) In Q25-GFP control cells, no fluorescent signal was detected after detergent treatment because the fusion protein was dispersed into the medium and only the fluorescence signal at the bottom of an assay plate was measured by the reader.

We also found that the fluorescence intensities from whole wells of assay plates measured by a conventional plate reader were not able to detect the difference between the Q103-GFP and Q25-GFP control cells. This could be due to the similar expression levels of total GFP fusion proteins in both cell lines and the fluorescence intensities from soluble proteins and aggregates were not distinguishable by measuring a whole well in assay plates. We then used a laser scanning plate cytometer (Acumen Explorer reader) which read from the bottom of assay plates and applied a signal threshold algorithm to identity fluorescent objects (e.g. aggregates) on the bottom of plates against the fluorescent background in solution. We found that the GFP tagged aggregates in Q103-GFP cells were distinguishable from the soluble GFP fusion proteins in Q25-GFP control cells after the detergent treatment. Thus, from optimizing the measurement parameters, only the intensity of fluorescent objects ranging in size of 1 to 25 µm were counted from the bottom of assay plates. The EC₅₀ values of tebufenozide, an inducer for the fusion protein expression, were 25.5 and 28.9 nM in the Q103-GFP cells prior to and after the detergent treatment, respectively (Fig. 2A). The result indicated that the detergent treatment did not change the fluorescence signals from Q103-GFP fusion proteins. Although the concentration response of tebufenozide in Q25-GFP cells was observed before the detergent treatment, no detectable signal was found by using the Acumen Explorer reader because the Q25-GFP fusion proteins were dispersed in solution after the detergent treatment (Fig. 2B). Thus, this protein aggregation assay based on the laser scanning cytometer detection is able to distinguish the aggregated fusion proteins from the un-aggregated proteins in assay plates after the treatment with 0.25% Triton X-100 detergent.

Expression of long polyglutamine expansions such as in the Q103-GFP cells has been shown to be a cause of protein aggregation and resulting cell death in many cell types [15 Li SH, Li XJ. Huntingtin and its role in neuronal degeneration Neuroscientist 2004; 10(5 ): 467-75.]. Compounds that inhibit the cytotoxicity of polyglutamine expression are of interest as research tools to study the pathophysiology of many diseases related to protein aggregation. In order to identify lead compounds which inhibit cytotoxicity caused by expression of long polyglutamine expansions, we multiplexed the assay readout to measure both protein aggregation and cytotoxicity in the same well of the same assay plate.

In Q103-GFP cells, only a subpopulation of cells (usually ~40-50%) is adversely affected after 48 hr induction of expression of Q103-GFP proteins. Previous reports indicated that signal-to-basal ratios of MTT or LDH release assays in these cells were not robust with lower signal-to-basal ratios [13 Aiken CT, Tobin AJ, Schweitzer ES. A cell-based screen for drugs to treat Huntington's disease Neurobiol Dis 2004; 16(3 ): 546-5.]. We initially tried an ATP content assay for the measurement of ATP levels in viable cells with a luminescence readout. The signal-to-basal ratio was around 1.5 to 2 fold, presumably due to the large percentage of viable cells (Fig. 3A). We then applied a protease release assay that measures the activity of released proteases from damaged plasma membrane due the toxicity of Q103-GFP proteins. We found that the signal-to-basal ratio of this protease release assay in Q103-GFP cells was 11.1 fold, indicative of a robust assay for high throughput screening (Fig. 3A). The high signal-to-basal ratio was found in this protease release assay that may be due to the lower basal signal in this assay compared with the high basal signal in the ATP content assay [16 Cho MH, Niles A, Huang R, et al. A bioluminescent cytotoxicity assay for assessment of membrane integrity using a proteolytic biomarker Toxicol in Vitro 2008; 22(4 ): 1099-6.]. The cellular proteases only leak out to media when the integrity of plasma membrane is perturbed in apoptotic or necrotic cells. The cytotoxicity of GFP-Q103 protein expression after 48 hr induction with the inducer tebufenozide was much more significant compared with that of 24 hr induction, which correlated with the time course of protein aggregation formation detected by the above assay (Fig. 3B).

Fig. (3) (A) Comparison of an ATP content assay and a protease release assay for measurements of cytotoxicity resulted from the expression of Q103-GFP fusion proteins in the Q103 PC12 cells. The expression of Q103-GFP fusion proteins in these cells was induced by 48 hr treatment with the inducer tebufenozide. (A) In the ATP content assay, 30 to 50 % reduction in ATP levels caused by the fusion protein cytotoxicity was observed in the cells treated with the inducer tebufenozide compared with that in untreated cells, resulting in a signal-to-basal (S/B) ratio of 1.4 fold. In the protease release assay, the protease activity due to the cytotoxicity was dramatically increased in the cells treated with the inducer tebufenozide compared with that in untreated cells (n=4 for each data point). The average S/B ratio was 9 fold. Thus, the protease assay produced a better assay window and was selected for the measurement of cytotoxicity in our experiments. (B) Concentration response curves of inducer tebufenozide determined in Q103-GFP cells after 24 or 48 hr treatment. The average S/B ratios in both protease and aggregation assays were much higher after 48 hr incubation with the inducer tebufenozide than these after 24 hr incubation. The EC50 values of tebufenozide determined from the protease release cytotoxicity assay and protein aggregation assay were 9.5 and 29.6 nM, respectively (n=12 for each data point). (C) Scatter plot of a DMSO plate test in the protease release assay. (D) Scatter plot of a DMSO plate test in protein aggregation assay. Both results in C and D were obtained from the same assay plate in a multiplexed assay experiment. In both panel C and D, the wells in columns 1 and 2 were not treated with inducer (e.g. negative control) and wells in columns 3 through 48 were treated with 200 nM Tebufenozide for 48 hours. Uninduced cells exhibited virtually no detergent insensitive fluorescence signal (D). All the wells also received 0.46% DMSO, a solvent for compounds.

Because the measurements of cytotoxicity and protein aggregation described in the above experiments utilize two different detection modes (e.g. luminescence and fluorescence intensity, respectively), we tried to multiplex them into sequential measurements in a single assay plate. The main advantage for using the same assay plate to measure compound effects on both cytotoxicity and protein aggregation is that the compound activities in these two assays are determined from the cells treated under identical experimental conditions. The results from such a multiplexed assay would be more comparable for the similarity or difference in compound activities on protein aggregation and cytotoxicity caused by the aggregated proteins. This multiplexed assay also has very few steps and is therefore relatively simple for the robotic screening (Table 1). Briefly, the cells are seeded into a 1536 well plate and incubated for one day. Expression of the Q103-GFP or Q25-GFP fusion proteins is induced by the addition of 200 nM tebufenozide (with or without library compound addition). After a 48 hr treatment with the inducer, cytotoxicity is first measured by the protease release assay in the luminescence detection mode with a single reagent addition step. After the cytotoxicity measurement, the assay plates are treated with detergent to lyse the cells that subsequently allows all soluble proteins to diffuse into the surrounding media. The fluorescent protein aggregates are then measured using a laser scanning plate cytometer (Fig. 4).

Table 1

Protocol for Multiplexed Protein Aggregation and Cytotoxicity Assay in 1536-Well Plate Format

Fig. (4) Schematic illustration of a multiplexed assay approach that sequentially measures cytotoxicity by a protease release assay and protein aggregates by a laser scanning cytometer plate reader in the same assay plate. From left to right, the uninduced Q103-GFP cells were seeded into an assay plate and incubated for 24 hr (1). The inducer tebufenozide, as well as compounds to be screened, were added to the cells and incubated for 48 hr (2). The protease detection reagent was added to the assay plate to detect the luminescent signal for cytotoxicity resulting from Q103-GFP expression (3). Detergent (final 0.25% Triton X100) was then added to lyse the cells and release soluble proteins into solution. Fluorescent protein aggregates remained attached to the bottom of assay plate were measured by the laser scanning cytometer plate reader (4).

Using a DMSO plate as a control , we found that the signal-to-basal ratios for cytotoxicity and protein aggregate measurements were 9.6 and 250 fold, the CVs were 7.5% and 9.8%, and the Z’ factors were 0.80 and 0.83, respectively (Figs. 3C and 3D). The results from this multiplexed assay demonstrate the robustness of assay performance, which is suitable for high throughput screening of large compound collections. We then screened a library of 220,581 small molecule compounds at four concentrations ranging from 0.5 to 40 µM. Several groups of active compounds with different effects on cytotoxicity and protein aggregation were identified from this screen. The primary screening results have been deposited in PubChem (AID 1471, 1482 and 1688).

Representative compounds identified from the compound library screening are shown in Fig. (5). The cytoprotective compounds identified were classified into three groups; (1) compounds inhibiting cell death but increasing protein aggregation (top panel), (2) compounds inhibiting both cell death and protein aggregation (middle panel), and (3) compounds inhibiting cell death without an effect on protein aggregation (bottom panel). While all three representative compounds exhibited cytoprotective effect in the protease release assay as well as an ATP quantitation assay (data not shown), compound 1 (securinine) increased the number of protein aggregates (green color in the image), compound 2 (hycanthone) decreased the number of protein aggregates and compound 3 (ipriflavone) had no effect on the protein aggregates compared with the control cells. These active compounds could be useful research tools for the further studies on the mechanism of protein aggregation and its relation with cytotoxicity as well as further studies in animal models. Some of these compounds are currently under investigation to identify their potential mechanisms of action. The results of structure-activity studies as well as the potential mechanism of action will be reported in a separate article.

Fig. (5) Three representative compounds that modulate cytotoxicity and protein aggregation identified from the compound library screen using the multiplexed cytotoxicity and protein aggregation assay. All three compounds inhibited the cytotoxicity caused by the expression of Q103-GFP fusion proteins in the Q103-GFP cells. The Compound -1 (on the top panel) had an IC50 of 11.2 µM on inhibition of cytotoxicity but it increased the aggregates formation with a potency of 14.1 µM. Compound -2 (in the middle panel) had an IC50 of 12.5 µM on inhibition of cytotoxicity and an IC50 of 5.01 µM against the aggregates formation. Compound -3 (on the bottom panel) had an IC50 of 1.41 µM on inhibition of cytotoxicity without effect on protein aggregation. The potency and percentage of maximal control response for cytoprotection and aggregates formation are listed in columns 3 though 6, respectively. The last column shows the images (magnified by 20×) of unlysed cell and their associated GFP aggregates in the presence of 25 µM compounds.