Therapeutic Rescue of Misfolded Mutants: Validation of Primary High Throughput Screens for Identification of Pharmacoperone Drugs

Background Functional rescue of misfolded mutant receptors by small non-peptide molecules has been demonstrated. These small, target-specific molecules (pharmacological chaperones or “pharmacoperones”) serve as molecular templates, promote correct folding and allow otherwise misfolded mutants to pass the scrutiny of the cellular quality control system (QCS) and be expressed at the plasma membrane (PM) where they function similarly to wild type (WT) proteins. In the case of the gonadotropin releasing hormone receptor (GnRHR), drugs that rescue one mutant typically rescue many mutants, even if the mutations are located at distant sites (extracellular loops, intracellular loops, transmembrane helices). This increases the value of these drugs. These drugs are typically identified, post hoc, from “hits” in screens designed to detect antagonists or agonists. The therapeutic utility of pharmacoperones has been limited due to the absence of screens that enable identification of pharmacoperones per se. Methods and Findings We describe a generalizable primary screening approach for pharmacoperone drugs based on measurement of gain of activity in stable HeLa cells stably expressing the mutants of two different model G-protein coupled receptors (GPCRs) (hGnRHR[E90K] or hV2R[L83Q]). These cells turn off expression of the receptor mutant gene of interest in the presence of tetracycline and its analogs, which provides a convenient means to identify false positives. Conclusions The methods described and characterized here provide the basis of novel primary screens for pharmacoperones that detect drugs that rescue GPCR mutants of specific receptors. This approach will identify structures that would have been missed in screens that were designed to select only agonists or antagonists. Non-antagonistic pharmacoperones have a therapeutic advantage since they will not compete for endogenous agonists and may not have to be washed out once rescue has occurred and before activation by endogenous or exogenous agonists.


Introduction
G protein-coupled receptors (GPCRs), which include the gonadotropin releasing hormone (GnRH) receptor (GnRHR) and vasopressin type 2 receptor (V2R), comprise the largest family of validated drug targets; 30-50% of approved drugs derive their benefits by selective targeting of GPCRs [1]. Mutations in GPCRs are known to be responsible for over 30 disorders, including cancers, heritable obesity and endocrine diseases. Normally, GPCRs are subjected to a stringent quality control system (QCS) in the endoplasmic reticulum (ER). The QCS insures that only correctly folded proteins enter the pathway leading to the plasma membrane (PM). This system consists of both protein chaperones that retain misfolded proteins and enzyme-like proteins that participate in catalysis of the folding process. It has become apparent that point mutations may result in the production of misfolded and disease-causing proteins that are unable to reach their functional destinations in the cell because they are retained by the QCS even though they may retain function.
Pharmacoperone drugs (from ''pharmacological chaperone'') are small molecules that enter cells and serve as molecular scaffolding in order to cause otherwise-misfolded mutant proteins to fold and route correctly within the cell. Many pharmacoperones are also agonists or antagonists because they have come from high throughput screens that were originally designed with a view toward identification of such congeners as lead drug candidates, not pharmacoperones as such. Pharmacoperone activity has been identified in these targets post hoc by us [2,3] and others [4] for the GnRHR and V2R systems, respectively. Valuable drugs which effect the trafficking of GPCRs may have been overlooked because of this limitation.

Physiological Significance of the Targets Selected
The V2 receptor (V2R, also known as the arginine vasopressin receptor) is expressed in the distal convoluted tubule and the collecting ducts of the kidney. V2R responds to vasopressin by stimulating mechanisms that concentrate the urine and maintain water homeostasis in the organism. When the function of V2R is lost due to mutation, the disease nephrogenic diabetes insipidus (NDI) results. The gonadotropin-releasing hormone receptor (GnRHR, also known as the luteinizing hormone releasing hormone receptor) resides primarily in the gonadotrope cells of the pituitary and is responsible for producing responses to hypothalamic GnRH, such as the releasing of the gonadotropins, luteinizing hormone (LH) and follicle stimulating hormone (FSH). When the function of this receptor is lost due to mutation, the disease hypogonadotropic hypogonadism (HH) results.

Advantages of the Approach Selected
Prior to establishing the cell-based assays described, we considered a number of alternative approaches. Because the assay is based on the redistribution of mutant GPCRs from the ER to the plasma membrane, we were unable to rely on membrane-type assays (e.g., GTPcS assays); intact and functional cells are required. Because well-characterized antiserum for GPCRs cannot be presumed to exist for every GPCR a priori (and, in fact, no wellcharacterized antiserum is available for the hGnRHR) we chose not to utilize ELISA or Western blot techniques. Fluorescently tagged (i.e. GFP) or HA-tagged mutants were not used, as this alters routing [19] and FLIPR TM -type assays were not used because of their difficulty to configure, potential artifacts and the need for expensive equipment.
Among the advantages of the approach we selected are that stable cell lines produce a reproducible response upon receptor stimulation and give rise to a high signal window. In transiently transfected cells a high proportion of non-transfected cells may be present. Untransfected cells reduce the maximum signal since they do not contribute to production of stimulated endpoint. In addition, stable cells are convenient since they do not require separate transfection for each experiment.
A special feature of the cell lines developed is that the GPCRs or GPCR mutants are expressed under the control of the tetracycline-controlled transactivator (tTA). The tetracyclineregulated expression system is based on two components: a Tetdependent transcription activator (tTA), which is a fusion between the Tet repressor of transposon TN10 and transcription factor binding domains of the herpes simplex protein VP16, and secondly, a tTA-responsive promoter, composed of seven Tet repressor binding sites (TetO7) immediately upstream of an RNA polymerase II transcriptional start site of the cytomegalovirus IE promoter (CMVm). When both elements are present in the cell, tTA binds to TetO7 and activates transcription at its neighboring initiation site. In the absence of tetracycline, the GPCR is expressed. In the presence of tetracycline, the GPCR is not measurably expressed. This model allows use of the GPCR to measure signal in the HTS (i.e. no tetracycline) and the identical background cell, lacking the expressed GPCR, to serve as a negative control (i.e. with tetracycline), thereby isolating false positives that may activate other cellular functions than the GPCR target.
We selected two model mutants of the human gonadotropin releasing hormone and vasopressin 2 receptors. These mutants are known to be misrouted, misfolded proteins [21], hGnRHR[E 90 K] [22] and hV2R[L 83 Q] [23] and are naturally occurring in patients with hypogonadotropic hypogonadism and nephrogenic diabetes insipidus, respectively. Stable HeLa cells were created expressing sequences for these proteins under control of a tetracycline-off promoter.
For both assays, we chose to use assays for effector coupling (IP or cAMP production) since, in vivo, effector activation would be the best measure of stimulation of the model systems under study. One could imagine receptor mutants that bound ligand, but failed to couple to effector. For such mutants, measuring receptor numbers (i.e. a radioligand assay) would provide a misleading measure of functional receptors. Moreover the need to develop a HTS is better served by using IP or cAMP as a screening assay, since it is

Primary Screen Model and Negative Control
The primary screen for pharmacoperones is rescue (gain) of function of stable HeLa cells expressing the indicated mutants (hGnRHR[E 90 K] or hV2R[L 83 Q]). A common confusion revolves around the use of an antagonist for receptor rescue which is then activated by an agonist. It is important to note that the pharmacoperone (a receptor antagonist) is not present at the time of agonist challenge. The general protocols for treatment with known pharmacoperones and determination of the resultant signal is described in Methods. Cell-containing wells were treated in replicates of 5-9 and then subjected to statistical evaluation. The format and protocol of the Assay Validation Guidelines by the Eli Lilly & NIH Chemical Center (http://www.ncgc.nih.gov/guidance/manual_toc.html), have been used in the present study.

Statistical Evaluation of the GnRHR Mutant Assay
Linearity. Using known pharmacoperone drugs, the relation between the signal (''reads'') and dose concentrations was evaluated. Figure 1 suggests a quadratic relation between the reads and concentrations of pharmacoperone drug.
Based on this observation, a linear regression analysis was attempted to evaluate the linearity/association of the doseresponse of pharmacoperone and concentrations of dose by estimating a polynomial function ( Figure 1). The p-value for this model was ,0.0001 as shown in Table 1, which indicates a significant relation between the dose-response and concentrations. The R-square value (0.37) suggests that only 37% the total variation is explained by the model. The estimated equation for this model is Signal = 701.94+1239.536Concentration2240.186 Concentration 2 .
The parameter estimates for both the linear and quadratic terms of concentrations were statistically significant with pvalues,0.0001 (Table 1). To assess the validity of the model  assumptions, the plot of the Studentized residuals versus the predicted values, along with 95% confidence interval, is performed and it is displayed in Figure 2. This figure exhibits systematic trends, suggesting that relation between the dose and read may be non-linear. We attempted some transformations and non-linear fits, but these alternative approaches have provided similar residual patterns. The estimated equation for the model has a maximum gain of activity in stable HeLa cells at the 0.5 mg/ml dose by using the first order derivative (Figure 3). This estimated equation demonstrates that the signal will monotone increase as concentration increases until it reaches up to 0.5 mg/ml, while the signal will decrease as concentration increases for the concentration larger than 0.5 mg/ml.

Precision: Summary Signal & Plate Acceptance
Criteria. There was a statistical significant effect of the dates (pvalue = 0.04), and plates effect was not significant (p-value.0.05). These results are consistent with above tables and graphs.
Spatial Uniformity Assessment. The signals are plotted ( Figure 4) against the well number. The wells are ordered by the rows and columns in order to explore any patterns of the edge, drift effect and other systematic source of variability. There was no significant edge effect (p-value = 0.79) using ANOVA. This figure suggests no trends (drift effect). The simple linear regressions were used to explore the trend in 'signal over the location within the plate.' There was no association between 'the signal and location.'

Statistical Evaluation of the V2R Mutant Assay
Linearity. Using pharmacoperone drugs, the relation between the dose response of pharmacoperone and dose concentrations of pharmacoperone drug V2R was evaluated. Figure 5A suggests that there is a near linear relation between the signals and concentrations with increased variation for higher concentrations.
Based on these observations, linear regression was attempted to evaluate the linearity/association of the dos-response of pharmacoperone by estimating a linear function. The linear regression models did not fit data very well. We attempted a log transformation with base 10 for the concentrations. Figure 5B displays a strong linear relation between the signal and the log transformed concentration.
Therefore, a linear regression was performed to evaluate the linearity/association of the dos-response of pharmacoperone by estimating a linear function of signals as a function of log 10 (concentrations). Figure 5C displays the result of linear regression.
There was significant relation between the signals and the log transformed concentrations (p-value,0.0001). The R-square value indicates that the model accounts for 76% of the variation in the signals. Table 3 summarizes the parameter estimates along with their descriptive statistics. The null hypothesis that the slope is 0 was rejected at the level of 0.05 significance (p-value,0.0001). The fitted equation for this model is Signal = 0.0305+0.06796 log 10 (Concentration).
The plot of the Studentized residuals versus the predicted values is displayed in Figure 5D. When a model provides a good fit and does not violate any model assumptions, this type of residual plot exhibits no marked pattern or trend. This residual plot exhibits no specific trends, indicating that model fits well.

Precision: Summary Signal & Plate Acceptance
Criteria. Tables 4 shows the descriptive statistics including averages (AVG), standard deviations (SD), standard errors (SE), and coefficient of variation (CV). These were calculated as described above for the GnRHR mutant. CV values for this study range from 3.4% to 7.1% which are well below the acceptance criterion.
Uniformity Assessment. Temporal uniformity and plate uniformity assessment. Line plots ( Figures 6A and B) can reveal patterns of date by date variations, and plate to plate systematic variations.
There was a statistically significant effect of the date of experiment (p-value = ,0.0001), and plate effect (p-value,0.001) using ANOVA. By inspection, this effect will not impair the ability of the assay to detect positives, but suggests that internal controls should be included for each date of experiment. It is also possible to compare data between dates if the discrimination level for differences is modestly increased.
Spatial Uniformity Assessment. The signal was plotted against well number, where the wells are ordered by row first then by column in order to explore any pattern of edge, drift effect and other systematic source of variability ( Figures 7A and B). This figure suggests no trends (drift effect). The simple linear regressions were used to explore the trend in signals over the location (wells). There was no significant association between signals and location (p-value = 0.33).

Assessment of Compounds that Would Not Be Expected to Provide a Signal in the Assays
As a test of the specificity of the two assays and to compare the signal to noise ratio for known positives with non-specific compounds, a broad range of substances were evaluated in the assays (Figures 8, the GnRHR pharmacoperone assay and Figure 9, the V2R pharmacoperone assay). The non-specific compounds were selected from reagents that alter the level of cyclic nucleotides, block Ca 2+ and Na channels, inhibit calmodulin, inhibit transcription, activate a range of receptors, cross link membrane proteins and otherwise perturb the cell. In both assays, no non-specific compound produced more than 26 signal above background, which the positive control signals were on the order of 106 above basal. This indicates that these assays produce a satisfactory range of discrimination that enables the determination of ''positives.''

Discussion
Pharmacoperones are small molecules that enter cells and serve as a ''molecular scaffold'' to promote correct folding of otherwisemisfolded mutant proteins [22,24]. Because these drugs are frequently selected from candidates that were originally identified as target specific antagonists, these also show high target specificity as pharmacoperones, although competition for endogenous ligands is a therapeutic complication. Accordingly we sought to develop assays that would identify molecules that were not necessarily agonists or antagonists.
Although the use of pharmacoperone drugs in vivo is very recent and has not had the benefit of dose regime optimization, there are some in vivo successes that suggest the value of this approach. In a mouse model, Pey et al. [25] used compounds obtained from a chemical screen to treat rodents with phenylketonuria (PKU), an inherited metabolic disease caused by mutations in phenylalanine hydroxylase (PAH). This enzyme converts Phe to Tyr. Presently, restriction of access to Phe is an accepted therapy in humans. When WT-PAH or PKU-associated mutants were transiently expressed, treatment with such compounds increased PAH activity by up to 100%. This effect was associated with an increase in PAH synthesis and a decrease in its degradation. These compounds were effective when given orally and were able to stabilize PAH in the liver, increasing PAH activity and protein levels.
Another success involved patients with X-linked nephrogenic diabetes insipidus. Mutant vasopressin 2 receptors in NDI result in   misrouted proteins that are trapped in the ER, degraded and do not reach the plasma membrane in the collecting ducts of the kidney where they would normally reabsorb water. In vitro studies indicated that a non-peptide V1a receptor antagonist rescued cell surface expression and function of mutant V2 receptors. When applied in vivo, a short-term treatment with a V1a receptor antagonist showed that patients given this molecule decreased both 24-h urine volume and water intake. Maximum increase in urine osmolality was observed on day 3 and sodium, potassium, and creatinine excretions and plasma sodium were constant throughout the study [26]. It is interesting and useful to note that, within a family of mutants of a single receptor, pharmacoperone drugs that rescue one mutant, often rescue most other mutants, even if the mutations are located distally from one another [3,23]. This observation suggests that pharmacoperones bind at multiple sites and stabilize a conformation that is acceptable to the cell's quality control system, allowing them to traffic to the plasma membrane. Moreover, pharmacoperones from different chemical classes appear to bind to many of the same sites [27,28]. These observations suggest that pharmacoperone drugs identified in the screens described here will identify lead structures that will be useful in treatment of human and animal mutational disease.

Stable (tTA+GPCR) Cells
The stable HeLa (tTA; tetracycline-controlled transactivator) cell line was obtained from Peter Seeburg (Max-Planck-Institut für Medizinische Forschung, Molekulare Neurobiologie Jahnstraße 29, 69120 Heidelberg, Germany). Cells were maintained in DFG growth medium (DMEM/10%FCS/20 mg/ml Gentamicin) and grown at 37uC, 5% CO 2 in a humidified atmosphere until about 90% confluent. The cells were washed with Dulbecco's PBS, and then trypsinized to detach the cells. Growth medium will be added to the cells to dilute out the trypsin which will be centrifuged to pellet the cells. The hGnRHR[E 90 K] or hV2R[L 83 Q] mutants (in pTRE2-Hygromycin vector) were transfected into the stable HeLa (tTA) cell line. Selection antibiotics were used at 400 mg/ml G418 plus 200 mg/ml Hygromycin. Single colonies were selected and screened for expression of the mutant GPCRs. The dual stable cell lines were maintained using 200 mg/ml G418 plus 100 mg/ml Hygromycin in growth medium. Sub-cloning was used to select the best-expressing lines.
GnRHR Mutants: Stably Transfected HeLa Cells for measuring Inositol Phophates (tTA+hE 90 K-GnRHR) 10 4 cells per well were plated in 125 ml with or without 2 mg/ml Tetracycline. After 24 h, the medium was changed and fresh Tetracycline was added for another 24 h period. The cells were then treated with DMSO (vehicle), 0.01, 0.05, 0.1, 0.5, 1, or 5 mg/ml In3 (Merck) prepared in 1% DMSO (final) and incubated for 4 h at 37 C with or without 2 mg/ml Tetracycline. The cells were then washed twice with 175 ml of DBG containing 1% DMSO with or without 2 mg/ml Tetracycline using a 10 min incubation at room temp for each wash. The cells were pre-loaded with 125 ml 3 H-inositol (4 mCi/ ml) with or without 2 mg/ml Tetracycline for 18 h in inositol free DMEM. The cells were washed with 175 ml of inositol free DMEM/ 5 mM LiCl with or without 2 mg/ml Tetracycline and stimulated with 10 27 M Buserelin with or without 2 mg/ml Tetracycline for 2 h. Total IPs were determined as previously described [29].  [23]. These were gifts of Merck and Company and Sanofi Recherche, Exploratory Research Department, Toulouse, France, respectively. Each is highly selective for the corresponding receptor and each binds in the nM range.