CellSensor^TM CRE-bla CHO-K1 cell line, obtained from Invitrogen (Madison, WI), stably expresses a β-lactamase reporter gene under the regulation of a cAMP response element (CRE). Cells were cultured in DMEM medium supplemented with 10% dialyzed fetal bovine serum, 2 mM L-glutamine, 0.1 mM non-essential amino acids, 1mM sodium pyruvate, 25 mM HEPES, 50 U/ml penicillin and 50 μg/ml streptomycin, and 5 μg/ml of blasticidin at 37°C under a humidified atmosphere and 5% CO₂.

CellSensor^TM CRE-bla CHO-K1 cells were suspended in assay medium (DMEM with 1% dialyzed fetal bovine serum, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 25 mM HEPES pH 7.3, 100 U/ml penicillin, and 100 µg/ml streptomycin), and were dispensed at 2000 cells/ 5 µL/well in 1536-well tissue culture treated black/clear bottom plates (Greiner Bio-One North America, NC) using a Flying Reagent Dispenser (Aurora Discovery, Carlsbad, CA). After the cells were incubated at 37ºC overnight, 23 nL of compound or DMSO was transferred into the assay plates by a pin tool (Kalypsys, San Diege, CA) resulting in a 261-fold dilution. One µl of 30 nM NKH 477 (final concentration) was dispensed in all the columns except column 3 which one µl medium was added. The positive control plate format was as follows: Column-1 IBMX concentration-response titration from 2 nM to 67 µM, column-2, IBMX 38 µM +30 nM NKH 477, column 3, DMSO only, and columns 4 to 48, 30 nM NKH 477. The plates were incubated 3 hours at 37°C. One µl of LiveBLAzer^™ (Invitrogen) detection mix was added, the plates incubated at room temperature for 2 hrs, and fluorescence intensity at 460 and 530 nm emission was measured at 405 nm excitation by an Envision (Perkin Elmer, Boston, MA) detector. Data was expressed as the ratio of 460nm/530nm emissions.

Approximately 1280 compounds from Library of Pharmacologically Active Compounds (LOPAC), which was purchased from Sigma (St. Louis, MO), were screened in this qHTS. Compound plates were prepared as interplate titrations of fifteen dilutions with the four left-most columns left empty in each plate. Pin tool transfer of compounds to assay plates resulted in a 261-fold dilution. The final compound concentration in the 6 µl assay volume ranged from 0.5 nM to 38 µM. The primary data analysis was performed as previously described [18Inglese J, Auld DS, Jadhav A, et al. Quantitative high-throughput screening: a titration-based approach that efficiently identifies biological activities in large chemical libraries Proc Natl Acad Sci USA 2006; 103: 11473-8.]. Briefly, raw plate reads for each titration point were first normalized relative to the IBMX control (38 µM, 100%) in the presence of 30 nM NKH 477 and DMSO only wells (basal, 0%), and then corrected by applying a pattern correction algorithm using compound-free control plates (DMSO plates). Concentration-response titration points for each compound were fitted to the Hill equation yielding concentrations of half-maximal activity (EC50) and maximal response (efficacy) values. The concentration response curves of the compounds were classified into four major classes (1-4) based on the completeness of curve, goodness of fit, and efficacy [18Inglese J, Auld DS, Jadhav A, et al. Quantitative high-throughput screening: a titration-based approach that efficiently identifies biological activities in large chemical libraries Proc Natl Acad Sci USA 2006; 103: 11473-8.]. The compounds with class 1.1, 1.2, and 2.1 curves are statistically the most reliable, while the compounds with class 2.2 and 3 curves are less reliable. Class 4 compounds which show no concentration response were defined as inactive compounds.

As a start to identifying small molecule potentiators of the CREB signaling pathway, we conducted experiments to validate the CRE β-lactamase reporter gene assay using CRE-bla CHO-K1 cells expressing β-lactamase under the control of CRE.

In 384-well plate format, forskolin, an adenylate cyclase activator, and NKH 477, a water soluble forskolin analogue [19Morinobu S, Fujimaki K, Okuyama N, Takahashi M, Duman RS. Stimulation of adenylyl cyclase and induction of brain-derived neurotrophic factor and TrkB mRNA by NKH477, a novel and potent forskolin derivative J Neurochem 1999; 72: 2198-05.], concentration dependently induced CRE coupled β-lactamase activity after 5 hr incubation with the cells. The EC₅₀s of forskolin and NKH 477 were 0.14 μM and 0.05 μM, respectively (Fig. 1). The signal-to-background ratios were 4.5 and 4.3 for forskolin and NKH 477, respectively. NKH 477 was chosen as a control potentiator in the assay due to its water solubility.

Fig (1) Forskolin and NKH 477 stimulated β-lactamase activity in CRE-bla CHO cells in 384-well format.

The assay was miniaturized into 1536-well plate format with a final 6 μl assay volume. To optimize the incubation time, the cells were incubated with various concentrations of NKH 477 for 2 to 5 hrs. The EC_50s of NKH 477 are 0.11, 0.21, 0.22 and 0.26 μM for 2, 3, 4, and 5 hr incubation, respectively (Fig. 2A). The signal-to-background ratio from 3 hr incubation was 4.7 fold, which was the highest compared to other time points so was chosen for use in the assay. The effect of DMSO on NKH 477 stimulated β-lactamase activity was also evaluated, and showed that DMSO at concentrations up to 1% did not alter NKH 477 stimulated β-lactamase activity (Fig. 2B). Since the final DMSO concentration in the screening conditions chosen was 0.38 %, the assay is clearly tolerant of DMSO in the compound solution.

Fig (2) NKH 477 stimulated β-lactamase activity in CRE-bla CHO cells in 1536-well format. (A). Time course of NKH 477 stimulated β-lactamase activity in CRE-bla CHO cells in 1536-well format. (B). DMSO effect on the NKH 477 stimulated β-lactamase activity assay.

It has been reported that the over expression of CREB alone had no effect on memory enhancement, but the synergetic effect on CREB signaling enhanced LTM [3Tully T, Bourtchouladze R, Scott R, Tallman J. Targeting the CREB pathway for memory enhancers Nat Rev Drug Discov 2003; 2: 267-77.]. Thus, the potentiator of CREB function might be a better approach to enhance LTM. We have optimized a cell-based assay for compound screening in the presence of small amount of NKH477, an activator of CREB signaling. To find an appropriate concentration of NKH 477 that could be used to capture the response of potentiators without a significant reduction in the signal-to-background ratio, the concentration response of 3-Isobutyl-1-methylxanthine (IBMX) [20Wagner B, Jakobs S, Habermeyer M, et al. 7-Benzylamino-6-chloro-2-piperazino-4-pyrrolidino-pteridine, a potent inhibitor of cAMP-specific phosphodiesterase, enhancing nuclear protein binding to the CRE consensus sequence in human tumour cells Biochem Pharmacol 2002; 63: 659-8.], a known PDE inhibitor, was determined in the presence of several different sub-EC₅₀ concentrations of NKH 477. Fig. (3) showed that IBMX concentration-dependently increased β-lactamase activity in the presence of NKH 477. The highest signal-to-background ratio of approximately 2 fold was observed in the presence of 30 nM of NKH 477. The EC₅₀ of IBMX in the present of 30 nM NKH 477 was 10.5 μM. In the screen 30 nM of NKH477 was used to stimulate a low level of cAMP production and IBMX was chosen as a positive control.

Fig (3) The concentration response curves of IBMX in the presence of indicated concentrations of NKH 477. CRE-bla CHO cells were incubated with various concentrations of IBMX in the presence of indicated concentrations of NKH 477 for 3 hours, and then the β-lactamase activity was measured.

To evaluate the assay performance, DMSO plate without compounds in the presence of 30 nM NKH 477 was tested in 1536-well plate format. The signal-to-background ratio (S/B) was 2.0 fold, CV (coefficient of variation) was 8.6% and Z factor was 0.5 indicating that this β-lactamase assay in a 1536-well format was robust and ready for high throughput screening (Fig. 4).

Fig (4) Screening test of β-lactamase reporter gene assay in CREbla CHO cells in 1536-well format. Column-1 is a concentration response curve of IBMX, the known potentiator in duplicate, in the presence of 30 nM NKH 477. Column-2 is a concentration response curve of IBMX in the absence of NKH 477. Column-3 was the negative control with assay buffer only. Column-4 to 48 was DMSO with 30 nM NKH 477.

To test the data quality on each plate screened in the qHTS process, the concentration titration of IBMX, the positive control compound, was performed in each plate. The concentration response curves of IBMX from 51 plates seem to reproduce well in each plate, with an average EC₅₀ value of 11 ± 2 μM. The average signal-to-background ratio from 51 plates was 2.7 ± 0.24 and average CV (%) was 7 ± 4. The average Z’ factor and Z factor were 0.64 ± 0.07 and 0.60 ± 0.07, respectively. The high data quality with small interplate variation obtained from this screening could be accredited to the use of high quality liquid handlers, plate readers and the robust nature of this β-lactamase reporter gene assay. This ensured the quality of results from the interplate compound dilutions used in the qHTS campaign.

To minimize false positives and false negatives that are common with single-concentration screening paradigms, and obtain immediate information on compound potencies and efficacies, we screened the compounds in 7 different concentrations ranging from 0.5 nM to 38 μM in quantitative high-throughput screening (qHTS) format [18Inglese J, Auld DS, Jadhav A, et al. Quantitative high-throughput screening: a titration-based approach that efficiently identifies biological activities in large chemical libraries Proc Natl Acad Sci USA 2006; 103: 11473-8.]. The qHTS analysis and concentration-response curve classification schema has been published previously [18Inglese J, Auld DS, Jadhav A, et al. Quantitative high-throughput screening: a titration-based approach that efficiently identifies biological activities in large chemical libraries Proc Natl Acad Sci USA 2006; 103: 11473-8.]. In this study, compounds producing curve classes 1.1, 1.2, or 2.1 were defined as active, those producing class 2.2 or 3 curves were defined as inconclusive, and compounds producing class 4 curves defined as inactive.

To evaluate assay robustness, traditional Z’ and Z factors [21Zhang JH, Chung TDY, Oldenburg KR. A simple statistical parameter for use in evaluation and validation of high throughput screening assays J Biomol Screen 1999; 4: 67-73.], coefficients of variation (CV), and signal-to-background ratio were used, but were interpreted somewhat differently given the nature of the concentration-response data generated by qHTS. In traditional HTS in which compounds are tested at a single concentration, Z factors less than 0.5 or CV (over 10%) are frequently used as cut-offs for acceptable assays, since they indicate assay performance below which compound activities cannot be reliably determined. However, we have found that the Z factor and CV are less important in qHTS comparing to the single concentration screening because the hits are selected based on the significance of concentration-response of the compounds. In this qHTS, robust and reproducible results were obtained though the Z factor was 0.41 at the 38 µM compound concentration, and CVs were 12% and 18% at the 7.6 and 38 µM compound concentrations – both outside the usually acceptable HTS parameters. This illustrates a general feature of qHTS, which is that it allows screening of assays that would be unscreenable in traditional HTS paradigms.

To evaluate data reproducibility, the LOPAC library was screened independently in CRE-bla CHO cells on three separate days. Of these 1280 compounds, 44 compounds (3.4%) showed a concentration dependent response (curve classes 1-3) in all three independent runs. The EC₅₀ values from the three runs correlated well with an average R² of 0.94 (Fig. 5A). Example replicate concentration-response curves for some of these compounds are shown in Fig. (5B). 1185 compounds were inactive in all three runs. Twenty-five compounds showed activity in only one or two of the three runs, but the curve classes of these compounds were mostly class 2.2 or 3, indicating a lower degree of data reliability.

Fig (5) qHTS reproducibility of β-lactamase reporter gene assay. (A) LOPAC library was screened independently in CRE-bla CHO cells on three separate times. Linear correlations of EC50s from 48 compounds with curve classes 1-3 in three independent screening yielded an average R2 = 0.94 ± 0.02 ( 0.95, run 1 vs run 2; 0.94, run 1 vs run 3; and 0.92, run 2 vs run 3). 47 compounds were inactive in one or two runs and are colored gray in the figure. 1185 compounds were inactive in all three runs. (B) Examples of triplicate concentration-response curves for some of known PDE inhibitors, forskolin, rolipram and trequinsin hydrochloride, identified from the LOPAC library.

The triple LOPAC library screen revealed 44 potentiators of the CREB pathway. The relatively high active rate (3.4%) is likely due to both the nature of the LOPAC library comprised of pharmacologically active compounds, and the nature of the assay, which detects compounds active against any molecular target in the CREB pathway. Among the 44 compounds, twelve were known PDE inhibitors, including papverine [22Menniti FS, Chappie TA, Humphrey JM, Schmidt CJ. Phosphodiesterase 10A inhibitors: a novel approach to the treatment of the symptoms of schizophrenia Curr Opin Investig Drugs 2007; 8: 54-9.], trequinsin [23Lugnier C. Cyclic nucleotide phosphodiesterase (PDE) superfamily. A new target for the development of specific therapeutic agents Pharmacol Ther 2006; 109: 366-98.], Ro20-1724 [23Lugnier C. Cyclic nucleotide phosphodiesterase (PDE) superfamily. A new target for the development of specific therapeutic agents Pharmacol Ther 2006; 109: 366-98., 24Teixeira MM, Gristwood RW, Cooper N, Hellewell PG. Phosphodiesterase (PDE)4 inhibitors: anti-inflammatory drugs of the future? Trends Pharmacol Sci 1997; 18: 164-70.], rolipram [23Lugnier C. Cyclic nucleotide phosphodiesterase (PDE) superfamily. A new target for the development of specific therapeutic agents Pharmacol Ther 2006; 109: 366-98., 24Teixeira MM, Gristwood RW, Cooper N, Hellewell PG. Phosphodiesterase (PDE)4 inhibitors: anti-inflammatory drugs of the future? Trends Pharmacol Sci 1997; 18: 164-70.], zardaverine [24Teixeira MM, Gristwood RW, Cooper N, Hellewell PG. Phosphodiesterase (PDE)4 inhibitors: anti-inflammatory drugs of the future? Trends Pharmacol Sci 1997; 18: 164-70.], diprydamole [23Lugnier C. Cyclic nucleotide phosphodiesterase (PDE) superfamily. A new target for the development of specific therapeutic agents Pharmacol Ther 2006; 109: 366-98.], IBMX, methoxymethyl-IBMX, etazolate [25Chasin M, Harris DN, Phillips MS, Hess SM. 1-Ethyl-4-(isopropylidenehydrazinol-H-pyrazolo-(3,4-b)-pyridin-5-carboxylic acid, ethyl ester, hydrochloride (SQ 20009) - potent new inhibitor of cyclic 3'5-nucleotide phosphodiesterase Biochem Pharmacol 1972; 21: 2443-50.], ibudilast [26Kishi Y, Ohta S, Kasuya N, Sakita S, Ashikaga T, Isobe M. Ibudilast: a non-selective PDE inhibitor with multiple actions on blood cells and the vascular wall Cardiovasc Drug Rev 2001; 19: 215-5.], T-1032 [27Inoue H, Koji Yano K, Noto T, Takagi M, Ikeo T, Kikkawa K. Acute and chronic effects of T-1032, a novel selective phosphodiesterase type 5 inhibitor, on monocrotaline-induced pulmonary hypertension in rats Biol Pharm Bull 2002; 25: 1422-6.], and propentofylline [28Meskini N, Némoz G, Okyayuz-Baklouti I, Lagarde M, Prigent AF. Phosphodiesterase inhibitory profile of some related xanthine derivatives pharmacologically active on the peripheral microcirculation Biochem Pharmacol 1994; 47: 781-8.]. Inhibition of PDE leads to a decrease in the cAMP degradation, which enhances CREB signaling. A direct activator of adenylate cyclase, forskolin, and two cAMP analogues, 8-bromo-cAMP and 8-(4-chlorophenylthio)-cAMP were also found active. The EC₅₀ values of these known compounds from the screen are listed in Table 1. These data suggest that this CRE qHTS assay is robust and can be used to identify small molecule compounds that potentiate CREB signaling from large compound libraries.

Table 1

Summary of Compound Potencies (µM) with Known Activity in the CREB pathway identified from screening of the LOPAC library in qHTS