Analytical Biochemistry 385 (2009) 286–292

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/ locate /yabio

A combination of ultrahigh throughput PathHunter and cytokine secretion assaysto identify glucocorticoid receptor agonists

Amita Patel a,*, Justin Murray a, Sheila McElwee-Whitmer b, Chang Bai b, Priya Kunapuli a, Eric N. Johnson a

a Department of Automated Biotechnology, Merck & Co., Inc., 150 Wissahickon Avenue, North Wales, PA 19454, USAb Department of Molecular Endocrinology, Merck & Co., Inc., West Point, PA 19454, USA

a r t i c l e i n f o

Article history:Received 12 September 2008Available online 11 November 2008

Keywords:Glucocorticoid receptorUltrahigh throughput screenPathHunterCytokine secretion

0003-2697/$ - see front matter � 2008 Elsevier Inc. Adoi:10.1016/j.ab.2008.11.005

* Corresponding author. Fax: +1 267 305 8099.E-mail address: amita_patel@merck.com (A. Patel)

1 Abbreviations used: DMSO, dimethyl sulfoxide; GRlipopolysaccharide; NRs, nuclear receptors; PBS, phoultrahigh throughput screens.

a b s t r a c t

The use of ultrahigh throughput screens (uHTS) is a well-accepted mechanism to identify agonists andantagonists of target receptors. We used the Path Hunter [Path Hunter technology is a registered trade-mark of DiscoveRx Corporation.] technology from DiscoveRx to screen the entire Merck compound libraryfor glucocorticoid receptor (GR) agonists in a 2.2-ll total reaction volume assayed in a 3456-well plateformat. This single addition, homogenous assay which utilizes the principle of enzyme fragment comple-mentation (EFC) to detect nuclear translocation of GR, an initial step of receptor activation, was used tosuccessfully screen a large library of small molecules as indicated by an average signal to backgroundratio of approximately 4-fold and an average Z-factor value of 0.45. Hits from the HTS campaign werestudied in a cytokine secretion assay in primary human monocytes to gain functional information regard-ing these compounds in a phenotypic and physiologically relevant setting. Our data indicate that usingthe PathHunter assay, we successfully identified compounds that showed agonism for the GR receptorin primary human monocytes and due to their performance in a physiologically relevant model theylikely will have a better chance to evoke clinical efficacy.

� 2008 Elsevier Inc. All rights reserved.

1

Nuclear receptors (NRs) are ligand-inducible transcription fac-tors that can regulate the expression of target genes involved in di-verse processes such as metabolism, development, and reproduction[1]. Hormones such as the sex steroids (progestins, estrogens, andandrogens), adrenal steroids (glucocorticoids and mineralocortic-oids), vitamin D3, and thyroid and retinoid (9-cis and all-trans) hor-mones act as ligands for the NRs [1]. More than 100 NRs are knownto exist in the human genome, and, together, these proteins composethe single largest family of metazoan transcription factors, the nucle-ar receptor superfamily [1]. Because of their role in a wide variety ofimportant processes, nuclear receptors are interesting targets fordrug development [2].

The glucocorticoid receptor (GR), a member of the nuclear hor-mone receptor superfamily, is known to regulate numerous phys-iological pathways [3,4]. GR binds to specific DNA sequences,glucocorticoid response elements (GREs), of target genes whichmay either activate or repress transcription [3,4]. Steroidal gluco-corticoids act as ligands for GR [5,6]. They have long been usedfor the treatment of many autoimmune and inflammatory disor-ders including rheumatoid arthritis (RA). Despite their benefits,

ll rights reserved.

., glucocorticoid receptor; LPS,sphate-buffered saline; uHTS,

the use of GCs is limited by their undesirable effects such as hypo-thalamic-pituitary-adrenal (HPA) suppression leading to decreasedability to respond to stress, increased glucose production, and de-creased insulin sensitivity resulting in diabetes, skin/muscle atro-phy, osteoporosis, glaucoma, and cataracts [5,6]. Therefore, drugsthat would separate the anti-inflammatory effects of steroids fromsome of the side effects would be beneficial [7,8].

The use of ultrahigh throughput screens is a well-acceptedmechanism to identify agonists and antagonists of target receptors.Previously, these screens would consist of evaluating a bindingevent between purified proteins or using a cell-based reporter genesystem. While the binding assay is effective at identifying com-pounds that bind the target, the fact that it occurs with purifiedproteins may reveal false positives or negatives due to the fact thatappropriate cofactors may be absent from the assay. In addition,competition binding assays identify only those compounds thatprevent binding of a radiolabeled ligand, and could therefore missallosteric modulators. A reporter gene system offers the advantageof being a functional assay that occurs in a cellular environment.However, maintaining proper cell health during the relatively longduration required for transcription of the reporter can be a chal-lenge. In addition, reporter gene assays are prone to identify off-target hits. Inhibition of any part of the transcription machinerycould reveal a hit without interaction with the NHR of interest.

The PathHunter system from DiscoveRx [9] is a single-addition,homogeneous assay that exploits the properties of enzyme

Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292 287

fragment complementation (EFC) associated with a tagged targetprotein. In this system, a CHO cell expressing a portion of beta-galactosidase in the nucleus is stably transfected with a GR con-struct fused to a complementary portion of beta-galactosidase,prolabel. Upon ligand binding, the GR-prolabel fusion proteintranslocates to the nucleus, allowing the complementary portionsof beta-galactosidase to interact and form a functional enzyme thatis capable of hydrolyzing a chemiluminescent substrate and pro-ducing light.

While this assay provides the ability to screen large libraries ofsmall molecules very quickly, it cannot distinguish between anagonist that stimulates transcription and a compound that will in-duce translocation, but fail to evoke transcription. In addition, be-cause of the nature of overexpressed systems, this assay lacks ahigh relevance to physiology.

As noted above, the stimulation of GR is an effective treatmentfor inflammation. Primary human monocytes stimulated with lipo-polysaccharide (LPS) secrete inflammatory cytokines. Pretreatmentwith a glucocorticoid receptor agonist, such as dexamethasone, de-creases the secretion of IL1-b, TNFa, and IL-6. Therefore, by imple-menting a follow-up assay to hits selected from the PathHunterHTS campaign, we were able to further profile compounds identi-fied as stimulating GR translocation to the nucleus, in an anti-inflammatory assay in a physiologically relevant system that canenable the evaluation of efficacy.

Here we report the results of a complete uHTS campaign usingthe PathHunter technology to identify GR agonists and a follow-upstudy in primary human monocytes measuring the inhibition ofLPS-evoked cytokine secretion.

Materials and methods

Cell lines

Frozen vials of Chinese hamster ovary K1 cells engineered tooverexpress a GR-prolink fusion protein were purchased from Dis-coveRx (Freemnont, CA). Frozen primary human monocytes wereobtained from Lonza (Lonza Walkersville Inc., MD; No. 2W-400A).

Cell culture for primary HTS

Frozen vials of Chinese hamster ovary K1 cells were thawed andcultured in growth medium (HAM/F12 (Invitrogen, Carlsbad, CA;No. 11765-062) containing 10% heat-inactivated FBS (Invitrogen;No. 10082-147), 1X penicillin/streptomycin/200 mM L-glutamine(Invitrogen; No. 10378-016), 500 lg/ml Geneticin (Invitrogen; No.10131-035) and 300 lg/ml Hygromycin B (Invitrogen; No. 10687-010)). When the cells reached 70–80% confluence in a T150 flask,the cells were washed with phosphate-buffered saline (PBS) anddissociated using trypsin/EDTA (Invitrogen; No. 25300-054) for5 min. Growth medium was then added to the flask and the cellswere split as needed. For the assay, after dissociation, the cells wereresuspended in assay buffer (HAM/F12, charcoal-stripped, heat-inactivated FBS (HyClone, Logan, UT; No. SH30068.03), 1X penicil-lin/streptomycin/200 mM L-glutamine, 500 lg/ml Geneticin,300 lg/ml Hygromycin B) at the required density.

The 3456-well plate assay protocol

Increasing concentrations of control compounds were dis-pensed into white-tissue culture-treated 3456-well plates at a vol-ume of 5 nl/well using the Pico Raptr (Beckman Coulter, Inc.,Fullerton, CA).

For the cell titration assay, cells at a range of densities (500–2000 cells per well) were resuspended in assay buffer and dis-pensed to designated wells in 1.4-ll volume. After the addition

of control compounds diluted in assay buffer (0.2 ll of varied con-centrations) the plate was incubated for 3 h in a humidified atmo-sphere at 37 �C and 5% CO2. PathHunter detection reagent (0.6 ll)was added to each well prior to incubation for 1 h at room temper-ature and reading relative luminescence on the ViewLux (Perkin-Elmer Life and Analytical Sciences, Inc., Waltham, MA) with a90 s exposure (High Gain, 1X binning). For the HTS 5 nl of 2 mMcompounds in DMSO was preplated into 3456-well assay plates(columns 7–66). Controls were dispensed in columns 1–6 and42–48.

The 1536-well plate assay protocol

In the 1536-well plate assays, 4 ll of cells (6000 cells per well)was added via a Flying Reagent Dispenser (Beckman Coulter, Inc.,Fullerton, CA). A 30-nl volume of compounds or controls wasadded to the cells using a Pintool (Kalypsys Systems, San Diego,CA) and the plate was incubated for 3 h in a humidified atmo-sphere at 37 �C and 5% CO2/95% O2. Detection reagent (3 ll) wasadded to the whole plate followed by incubation at room temper-ature for 1 h. Luminescence was read on the ViewLux (Perkin-Elmer Life and Analytical Sciences, Inc., Waltham, MA) for a 90 sexposure (High Gain, 1X binning).

Cytokine secretion assay

Frozen primary human monocytes (Lonza Walkersville Inc.; No.2W-400A) were thawed according to manufacturer’s instructions[10] and resuspended in RPMI 1640 medium (ATCC, Manassas,VA; No. 30-2001) containing 10% FBS (Hyclone, Logan, UT; No.SH3007003) at a density of 6.67e6 cells/ml. Cell were plated(45 ll, 30,000 cells per well) in 384-well plates ((Fisher Scientific,Newark DE; No. 07-200-650) and the plates were incubated for4 h at 37 �C, 5% CO2, 95% O2. Test compounds or controls such asdexamethasone (Sigma Aldrich Inc., St. Louis, MO; No. D-1756)were incubated with the monocytes for an additional hour. Bacte-rial lipopolysaccharide (Sigma Aldrich Inc.; No. L-2630) was dis-pensed using a multichannel pippetter (Finnpipette FocusMultichannel Plus, Thermo Fisher Scientific Inc., MA) to achieve afinal concentration of 200 ng/ml and the plates were incubatedovernight for about 18 h at 37 �C. The plates were centrifuged at228g for 10 min at room temperature and the supernatant washarvested and stored at �20 oC until being assayed.

Meso Scale detection using the Sector Imager 6000

The supernatant from cell treatments was thawed at room tem-perature and added (10 ll) to a 384-well MSD Multispot humancytokine assay plate (Meso Scale Discovery, Gaithersburg, MD;No. K21025B-2) for the quantification of four cytokines- IL-1b, IL-6, IL-8, and TNFa. The plate was sealed and incubated at room tem-perature for 2 h with shaking. An MSD detection antibody cocktail(1 pg/ml) was added to all wells and the plate was sealed and incu-bated at room temperature with shaking for 2 h. After beingwashed 3 times with 1X PBS containing 0.05% Tween 20, an MSDRead Buffer was added to each sample and the plate was immedi-ately read on the MSD Sector Imager 6000. Because the assay is amultispot read, there are four sets of data from each well of a384-well plate corresponding to the four distinct cytokines. Therange of detection for cytokines in this assay is 2.4 to 10,000 pg/ml.

Data analysis

Dose response curves were plotted using Prism (GraphPad Soft-ware, SanDiego, CA). The Z factor, a measure of assay quality con-trol [11], was determined by the following equation:

288 Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292

Z factor ¼ 1� ð3 � ðSDsample þ SDCþÞÞðAbsðmeansample �meanCþÞÞ

SDsample and SDC+ refer to standard deviation of sample and positivecontrol regions, respectively.

Results

Optimization of the 3456-well plate format assay for uHTS

In order to identify the optimal conditions for the GR PathHun-ter assay in a 3456-well plate format, a cell titration experimentwas performed. Cells at various concentrations were delivered tothe assay plate containing 5 nl of dexamethasone, a control ago-nist. Evaluation of the EC50 concentrations of the agonist and theresultant signal to background ratios revealed 1000 cells per wellto be the optimum for this assay (Fig. 1a). With the optimal celldensity selected, several GR modulators (agonists and a non-GRbinder) were evaluated in the GR translocation assay. As expected,the negative control, a non-GR binder, did not cause nuclear trans-location of the receptor, whereas other controls exhibited varyingpotencies of receptor nuclear translocation (Fig. 1b).

To evaluate the stability of cell response over a period of 4 h,cells were harvested and assayed at 5 time points, immediately,

Fig. 1. Standardization of PathHunter assay conditions. (a) Cell and dexamethasone titculture-treated plates in 5 nl volumes. Cells (1.4 ll) and 0.2 ll media were then added awas then added and the plates were incubated for 1 h at room temperature before readi(expressed as relative light units) is graphed as a function of compound concentration. (band S/B ratios. Five nanoliters of varying doses of dexamethasone, 6-a-methyl prednisoldihydrotestosterone) were plated into 3456-well plates. Cells, controls, and detection regraphed as a function of compound concentration.

represented by t=0, and at times 1, 2, 3, and 4, h of stirring(400 rpm) at room temperature. The cells in this assay demon-strated consistent signal to background ratios for up to 4 h (datanot shown), indicating that two deliveries per day will suffice formaximal performance and throughput of this assay.

Primary HTS

The primary screen for GR modulators was performed using the3456-well plate uHTS system. To facilitate rapid quality control ofthe data, we calculated four parameters for each plate: (A) ‘‘signal-to-background ratio,” the ratio of the median luminescence of allwells containing EC50 concentration of the positive control com-pound to the median luminescence of all wells containing the sam-ple compounds, (B) ‘‘% CV,” the coefficient of variance of the‘‘sample region” (C) ‘‘Z-factor” value, obtained with the meansand variance of luminescence of the ‘‘sample wells” and the‘‘EC50” wells and (D) ‘‘mean + 3 sigma” obtained from the meanof % activation of ‘‘sample wells” and standard deviation of % activ-ity of sample wells. Plates of low quality, possibly due to mechan-ical failures of liquid handlers or a misalignment of a reader, wererescheduled and assayed on a subsequent day. To better visualizerelatively weak GR agonists, we normalized the agonist data to

ration. Dexamethasone, at a range of concentrations, was plated into white tissuend the plates were incubated at 37 �C for 3 h. PathHunter detection reagent (0.6 ll)ng luminescence on the ViewLux with a 90-s exposure. Nuclear translocation of GR) The GR PathHunter assay detects known GR agonists with appropriate EC50 valuesone, a compound from the Merck collection, and a negative control compound (5a-agents were added. Nuclear translocation of GR (expressed as relative light units) is

Fig. 2. A representative compound plate from primary HTS. For primary HTS cells (1000 cells per well) and controls (0.2 ll of EC50 or EC100 concentration of dexamethasone)were dispensed into white 3456-well sample plates, containing 5 nl of 2 mM compounds. Controls were dispensed in columns 2–6 and 67–71, as indicated. The ‘‘% activity” ofeach sample was calculated using EC50 value of control agonist as reference for 100% activity and graphed as a function of column number of the plate.

Fig. 3. Summary of screen statistics. Panel a shows the signal–basal ratio (open circles) and % CV (filled circles) for individual plates assayed during HTS. Panel b shows the Zfactor (open circles) and mean + 3 sigma (filled circles) for each plate of the screen. The average S/B ratio and% CV values for all plates were approximately 4.5 and 16%,respectively, whereas the Z factor and mean + 3 sigma values for the screen were 0.45 and 16%, respectively.

Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292 289

Fig. 4. A representative dose response curve of a hit from the HTS campaign. Hitsfrom the 3456-well HTS campaign were titrated in 8-point dose titrations in 1536-well plate format according to the assay protocol detailed under Materials andmethods. In the graph,% activity (normalized to dexamethasone control) is plottedas a function of compound concentration.

Fig. 5. Effect of dexamethasone on cytokine secretion IL-1b (a), TNFa (b), IL-6 (c),and IL-8 (d) from LPS-treated human monocytes. Primary human monocyteswere plated at a density of 30,000 cells/well and treated with increasingconcentrations of dexamethasone in the absence (filled circles) or presence (opencircles) of 200 ng/ml LPS. Cell supernatants were analyzed for the presence of IL-1b (a), TNFa (b), IL-6 (c), and IL-8 (d). The cytokine concentration expressed aselectrochemiluminescence (ECL) units is graphed relative to dexamethasoneconcentration.

290 Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292

an EC50 concentration of dexamethasone. Because the efficacy ofdexamethasone is so great, the ratio of median raw luminescenceunit (RLU) for the EC50 concentration region to that of basal wasapproximately 4.

A scatter plot of a representative compound plate from thescreen (Fig. 2) demonstrates the low variation in the sample fieldand the ease at which ‘‘hits” can be selected by eye, demonstratingthe robustness of the assay. The signal to background ratio for thisplate, calculated as the median of the% activity value stimulated by10 nM dexamethasone divided by the median% activity of the sam-ple field, was 3.9-fold. Those compounds that evoked a percentageactivity of greater than the mean + 3 standard deviations of thesample field (calculated as 13.83% activity for this plate) were con-sidered as hits.

The summary statistics of the entire uHTS campaign indicatedconsistent performance of the PathHunter assay throughout thescreen. The mean S/B ratio,% CV, Z-factor value, and mean + 3 stan-dard deviations value were 4.3-fold, 16%, 0.45, and 15.3%, respec-tively. Graphical representations of the signal to backgroundratios and% CV for each plate of the uHTS campaign are shown inFig. 3. Those compounds that were selected as active in the pri-mary screen were confirmed in triplicate and yielded 248 con-firmed hits.

Titration of confirmed hits in 1536-well plate format

Dose response curves better characterize the compounds bytesting a range of concentrations rather than a single concentrationyielding relative efficacy in this translocation assay. Therefore, fol-lowing a confirmation screen, a limited set of compounds was pro-filed in 8-point dose titrations in 1536-well plate format using thesame PathHunter technology used for the primary screen. Fig. 4shows the dose response curve for a representative hit from thescreening collection with an EC50 value of 693 nM.

Cytokine secretion assay using MSD technology

To evaluate agonist activity of the hits generated from our HTScampaign in a biologically-relevant cell line with endogenous GR,we exploited the fact that GR is expressed in primary humanmonocytes. Stimulation of monocytes with lipopolysaccharide in-duces the secretion of several cytokines including TNFa and IL�1b.We took advantage of this response to create an assay using theMeso Scale Discovery multiplexed cytokine detection platform

Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292 291

which can measure the levels of IL-1b, IL-6, IL-8, and TNFa simul-taneously when the monocytes are challenged with GR agonistsand lipopolysaccharide. The MSD assay to quantify cytokines isbased on the principle of electrochemiluminescence [12].

Dexamethasone is a well-known GR agonist known to suppresscytokine release in LPS-challenged monocytes [13]. Assay optimi-zation revealed that dexamethasone dose-dependently reducedthe secretion of IL-1b, IL-6, and TNFa whereas it dose-dependentlyincreased IL-8 secretion from monocytes at 30,000 cells per well,challenged with 200 ng/ml LPS (Fig. 5). A control antagonist hadno effect on cytokine secretion (data now shown), suggesting thatthis functional assay could distinguish between agonists andantagonists of GR.

Fig. 6. Compounds identified in the translocation assay demonstrate an agonist-like phena tagged GR from the cytosol to the nucleus in the PathHunter assay were evaluated in a cexpressed as ECL units is graphed as a function of compound concentration. The correscytokines evaluated.

From the PathHunter assay results, 100 compounds that evokeddose-dependent nuclear translocation of GR were profiled in thisfunctional multiplexed cytokine secretion assay in primary humanmonocytes. Because the assay is multiplexed and resulted in thesimultaneous detection of 4 different cytokines, a more completeprofile of compound activity was revealed including EC50 valuesfor the alteration of secretion of 4 different cytokines (Fig. 6). Sim-ilar to dexamethasone, some compounds dose dependently inhib-ited secretion of IL-1b, IL-6, and TNFa, whereas they dosedependently increased IL-8 secretion from monocytes, suggestingthat those compounds potentially activate the glucocorticoidreceptor in a manner similar to that of dexamethasone and couldpotentially be GR agonists.

otype in the cytokine secretion assay. Two compounds that evoked translocation ofytokine secretion assay. The concentration of TNFa (a), IL-8 (b), IL-6 (c), and IL-1b (d)ponding EC50 values and S/B ratios for both compounds are shown for each of the

292 Nuclear translocation and cytokine secretion assays identify GR agonists/A. Patel et al. / Anal. Biochem. 385 (2009) 286–292

Discussion

An assay for ultrahigh throughput screening has many require-ments. It should be robust in order to detect weak activators orinhibitors. It must be reproducible as measured by appropriateplate statistics. To achieve optimum throughput it should be rela-tively rapid and able to be miniaturized into very small volumes.The PathHunter assay to evaluate nuclear translocation of GR pos-sesses each of those attributes. It has been recently reported thatthis assay was miniaturized to a 1536-well plate format and uti-lized in a quantitative screening paradigm of approximately10,000 compounds [14]. We were able to miniaturize the PathHun-ter assay into a 3456-well plate format with a final reaction vol-ume of 2.2 ll. The mean signal to background ratio for the entirescreen was 4-fold and the mean Z-factor value was 0.45. The assaywas very consistent across the entire screening campaign (Fig. 3).An additional attribute of the PathHunter nuclear translocation as-say is the fact that the target receptor is expressed as a fusion pro-tein with prolabel. Unlike reporter gene assays in which interactionwith many proteins in any signaling cascade that leads to tran-scription or translation of a reporter gene can appear as a hit, thesignal measured in the well of this PathHunter assay is the nucleartranslocation of only the receptor of interest. Therefore, this assaytype is less prone to the detection of off-target false positiveresults.

The combination of an ultrahigh throughput HTS campaign thatemploys overexpression of a target of interest with a primary cellassay with a phenotypic end-point can be an extremely effectivescreening paradigm [15]. Here we implemented that strategy bycombining our nuclear translocation of GR assay with the alter-ation of cytokine secretion in LPS-stimulated primary humanmonocytes. The cytokine secretion assay in human monocyteswas chosen for several reasons. It used a primary human cell thatis relatively easy to obtain and it endogenously expresses the tar-get receptor. In addition, it has been well documented that some ofthe therapeutic benefits from the treatment of glucocorticoids is asa result of their anti-inflammatory activity on monocytes and a po-tential surrogate marker for inflammatory activity is the secretionof proinflammatory cytokines [16–18].

The ability to evaluate 4 different cytokines from the same wellof stimulated supernatant is an added benefit of the MSD platform.Because known agonists of GR like dexamethasone decrease thesecretion of IL-1b, TNFa, and IL-6, it is difficult to know whetherthe decrease in cytokine secretion from LPS-stimulated monocytesis a result of the activation of GR or toxicity of the compounds. Themeasurement of a fourth cytokine, IL-8 in this case, that is in-creased in the presence of known GR agonists allows us to inferthat increases in GR activity could explain the decrease in IL-1b,TNFa, and IL-6. If a compound were toxic and resulting in celldeath, we would expect a decrease in all 4 of the cytokinesmeasured.

Thus we have shown the successful implementation of a screen-ing strategy in which an ultrahigh throughput assay was used torapidly screen a large collection of small molecules to vastly nar-row the pool of potential agonists. That pool was further evaluatedusing a much lower throughput but physiologically meaningful as-say in which we have greater confidence that the compounds that

showed agonism will have a better chance to evoke clinical effi-cacy. While this system worked very well for a nuclear receptor in-volved in an inflammatory disease, we are hopeful that itsapplication to other target classes and other therapeutic indica-tions will follow. GPCRs that were profiled in a beta-arrestin assaycould be further studied in insulin secretion assays as potentialtreatments for diabetes or antagonists of tyrosine kinase receptorsprofiled in a dimerization assay could be evaluated for their abilityto induce cell death in cancer cells. We are hopeful that this para-digm of screening will create a better link between the highthroughput assays in drug discovery and efficacy in patients whohave a clinical need.

Acknowledgments

We thank Azriel Schmidt and Carlo Gambone for meaningfuldiscussions regarding this project.

References

[1] N.J. McKenna, B.W. O’Malley, Nuclear receptors, Co regulators, ligands, andselective receptor modulators, Ann. NY Acad. Sci. 949 (2001) 3–5.

[2] D.P. McDonnel, E. Vegeta, M.A.G. Gleeson, Nuclear hormone receptors astargets for new drug development, Nat. Biotechnol. 11 (1993) 1256–1261.

[3] M.R. Yudt, J.A. Cidlowski, The glucocorticoid receptor: coding a diversity ofproteins and responses through a single gene, Mol. Endocrinol. 16 (2002)1719–1726.

[4] C. Stefanaki, G.P. Chrousos, A. Katsambas, Glucocorticoid and sex hormonereceptors: clinical implications and therapeutic relevance, Dermatol. Clin. 25(2007) 503–513.

[5] J. Rosen, J.N. Miner, The search for safer glucocorticoid receptor ligands,Endocr. Rev. 26 (2005) 452–464.

[6] J. Saklatvala, Glucocorticoids; do we know how they work?, Arthritis Res 4(2002) 146–150.

[7] B.M. Vayssiere, S. Dupont, A. Choquart, F. Petit, T. Garcia, C. Marchandeau, H.Aronemeyer, M. Resche-Rigon, Synthetic glucocorticoids that dissociatetransactivation and AP-1 transrepression exhibit anti-inflammatory activityin vivo, Mol. Endocrinol. 11 (1997) 1245–1255.

[8] P. Herrlich, Cross-talk between glucocorticoid receptor and AP-1, Oncogene 20(2001) 2465–2475.

[9] http://discoverx.com/.[10] www.lonza.com.[11] J. Zhang, T.D.Y. Chung, K.R. Oldenburg, A simple statistical parameter for use in

evaluation and validation of high throughput screening, J. Biomol. Screen. 4(1999) 67–73.

[12] www.meso-scale.com/.[13] D. Franchimont, H. Martens, M.T. Hagelstein, E. Louis, W. Dewe, G.P. Chrousos,

J. Belaiche, V. Geenan, Tumor necrosis factor-a decreases, and interleukin-10increases, the sensitivity of human monocytes to dexamethasone: potentialregulation of the glucocorticoid receptor, J. Clin. Endocrinol. Metab. 84 (1999)2834–2839.

[14] P.J. Zhu, W. Zheng, D.S. Auld, A. Jadhav, R. MacArthur, K.R. Olson, K. Peng, H.Dotimas, C.P. Austin, J. Inglese, A miniaturized glucocorticoid receptortranslocation assay using enzymatic fragment complementation evaluatedwith qHTS, Comb. Chem. High Throughput Screen. 11 (2008) 545–559.

[15] E.N. Johnson, High throughput screening follow-up studies usingphysiologically relevant cells, Am. Drug Discov. 3 (2008) 12–22.

[16] M. Schmidt, H. Pauels, N. Lugering, A. Lugering, W. Domschke, T. Kucharzik,Glucocorticoids induce apoptosis in human monocytes: potential role of IL-1b,J. Immunol. 163 (1999) 3484–3490.

[17] D.A. Joyce, J.H. Steer, L.J. Abraham, Glucocorticoid modulation of humanmonocyte/macrophage function: control of TNF-a secretion, Inflamm. Res. 46(1997) 447–451.

[18] L.M. Breuninger, W.L. Dempsey, J. Uhl, D.M. Murasko, Hydrocortisoneregulation of interleukin-6 protein production by a purified population ofhuman peripheral blood monocytes, Clin. Immunol. Immunopathol. 69 (1993)205–214.