Introduction

Top Abstract Introduction Materials and Methods Results and Discussion References

PGE₂ is a member of the 2-series of PGs that is derived de novo from arachidonic acid and constitutes the most abundant naturally occurring prostanoid (Campbell, 1990; Davies and MacIntyre, 1992). These autocrine and paracrine mediators are released in response to a variety of stimuli in many tissues, where they are involved in a broad spectrum of physiological and pathophysiological events (Coleman et al., 1989) and have been implicated in a number of therapeutic areas (Abramovitz and Metters, 1998).

PGE₂ elicits its effects primarily through interaction with four distinct prostanoid receptors: EP₁, EP₂, EP₃, and EP₄ (Coleman et al., 1994), members of the G-protein coupled receptor superfamily of integral serpentine plasma membrane proteins (Boie et al., 1995). The primary signaling pathways of the four EP receptor subtypes are as follows: EP₁ couples to elevation of [Ca²⁺]_i (Funk et al., 1993), EP₂ (Regan et al., 1994) and EP₄ (Bastien et al., 1994) couple to an increase in intracellular cAMP accumulation, and EP₃ couples to a decrease in intracellular cAMP levels (Boie et al., 1997; Jin et al., 1997).

When the four EP receptor subtypes are aligned by amino acid identity with the additional four members of the prostanoid receptor family (TP, Hirata et al, 1991; FP, Abramovitz et al., 1994; IP, Boie et al., 1994; DP, Boie et al., 1995), two subgroups are formed. Interestingly, the common element within each group is the signal transduction pathway to which the receptors couple rather than their preferred natural ligand: thus, G_q/G_i for the EP₁, FP, TP, and EP₃ receptors and G_s for the EP₂, EP₄, DP, and IP receptors (Boie et al., 1995; Toh et al., 1995). There are, in fact, no strictly conserved amino acid residues specific to the four EP subtypes that do not occur in one or more of the other four prostanoid receptors. This is reflected in preliminary molecular models of the EP receptors in which the few common residues found in the putative ligand binding pocket are also present in one or more of the non-EP prostanoid receptors (Yamamoto and Imai, 1996).

In view of this intriguing sequence information, it is relevant to identify the distinguishing structural features of PGE₂ and related prostanoid analogs that are important for the specific activation of each of the four EP receptor subtypes. This information would be particularly useful in the design of selective agonists. A number of SAR studies of prostanoids and prostanoid analogs had been conducted before the cloning of the receptors, using various bioassays from different mammalian species (Coleman et al., 1989 and references within). Unfortunately, the tissues chosen usually express mixed populations of two or more receptors, limiting interpretation of the results (Lawrence et al., 1992). Detailed SAR studies have yet to be published using recombinant systems expressing individual prostanoid receptors. The cloning of the prostanoid receptor family over the past several years (Coleman et al., 1994; Boie et al., 1995) has opened the door to evaluating compounds at all eight prostanoid receptors (Kiriyama et al., 1997; Abramovitz et al., 2000).

We describe in this report, for the first time, an SAR study of prostanoids and prostanoid analogs at the hEP₁ receptor using a well-defined recombinant hEP₁ heterologous expression system. A radioligand binding assay and a recently developed automated aequorin-based calcium functional assay (Ungrin et al., 1999) have been employed to obtain quantitative affinity and potency data for PGE₂ and related compounds at hEP₁. This data establishes that the 11 and 15(S) configuration of the hydroxyl groups, as well as the presence of the carboxylic acid moiety of the prostanoid structure, are crucial for hEP₁ receptor binding and activation.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion References

Chemicals. Chemicals were purchased from the following vendors: Biomol (Plymouth Meeting, PA) and Cayman (Ann Arbor, MI), with the exception of prostaglandin E₁ methyl ester (47) and prostaglandin F₂ methyl ester (50), which were purchased from Sigma (Oakville, ON, Canada). Concentrations of the two compounds obtained from Sigma were verified by NMR spectroscopy. The compounds used are listed in Table 1. The numbers assigned to the compounds in Table 1 are used throughout the article.

View this table: [in this window] [in a new window]   TABLE 1 Summary of all functional and affinity data, normalized relative to PGE2 as described under Materials and Methods The relative aequorin response data (Log10 relative aequorin) is presented as the average of triplicate experiments, ± 1 S.D., whereas for Log10 relative binding data (Log10 relative binding), both data points from duplicate experiments are presented. For ease of interpretation, the relative aequorin response is also presented as a simple nonlogarithmic ratio. Overall, in the activity assay, the EC50 value for PGE2 = 2.90 nM (n = 69), and in the binding assay, the Ki value for PGE2 = 9.9 nM (n = 18). Compound numbers are assigned based on rank order of potency in the aequorin assay. In the aequorin assay, all differences in activity of 0.5 log units ( 3-fold difference in EC50) are statistically significant (2-tailed t test,  = 0.05 for each individual comparison) with the exception of those involving 19(R)-hydroxy PGE2 (32). Statistical analysis was not possible for compounds 50 to 55 because the concentration-response curves did not reach a plateau at the maximum concentrations tested.

Aequorin Luminescence Assay. The aequorin luminescence assay was performed using the hEP₁ expressing clonal cell line hEP₁-5/293-AEQ17, as described previously (Ungrin et al., 1999). Briefly, holo-aequorin was reconstituted in intact cells by charging 85%-confluent cultures for 1 h at 37°C in Ham's F12 medium (with 0.1% fetal bovine serum, 25 mM HEPES, at pH 7.3) (Life Technologies, Missisauga, ON, Canada) containing 30 µM reduced glutathione (Sigma, St. Louis, MO) and 8 µM coelenterazine cp (Molecular Probes, Eugene, OR). After charging, the cells were washed from the growth surface by pipetting up and down, rinsed once, and resuspended in Ham's F12 medium (modified as above) at 5 × 10⁵ cells/ml. Experiments were performed using a Labsystems Luminoskan RS plate reading luminometer (Labsystems, Franklin, MA) with 3 integral peristaltic pumps and an internal orbital mixer. The luminometer was controlled by the Lskan Controller, custom software written in LabView (National Instruments, Austin, TX) and data was analyzed using a dedicated package of Excel Macros referred to as the Luminometer Data Analysis Macros (LDAM) developed at Merck Frosst.

1

Data Analysis. Peak integration values were obtained by summing the half-second integrations from the raw trace. Fractional luminescence for each well was determined by dividing the area under peak 1 by the total area under peaks 1 and 2. These calculations were performed using the Lskan Controller program, and a data file was generated containing both the raw traces, the calculated results for each well, drug concentrations, and the start time for each well. This data file was then subsequently analyzed using the LDAM software employing a modified version of the Levenberg-Marquardt four-parameter curve-fitting algorithm to calculate EC₅₀ values.

Radioligand Binding Assays. Radioligand binding assays were carried out on membranes prepared from the hEP₁-expressing clonal cell line hEP₁-HEK 293 (EBNA), as described previously (Abramovitz et al., 2000). Briefly, cells were harvested by incubation in enzyme-free cell dissociation buffer (Life Technologies, Inc.), washed in ice-cold phosphate-buffered saline, pH 7.4, and resuspended in 10 mM HEPES/KOH at pH 7.4 containing 1 mM EDTA. Membranes were prepared from harvested cells by lysis followed by differential centrifugation (1,000g for 10 min, then 160,000g for 30 min, all at 4°C). The 160,000g pellets were resuspended in 10 mM HEPES/KOH, pH 7.4, containing 1 mM EDTA at approximately 5 to 10 mg/ml by Dounce homogenization (Dounce A; 10 strokes), frozen in liquid nitrogen and stored at 80°C.

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion References

This SAR study was undertaken to elucidate the structural features of PGE₂ (Fig. 1) and 54 additional prostanoids and prostanoid analogs (structures shown in Fig. 2 and trivial names shown in Table 1) that significantly impact potency at hEP₁. Two previously described clonal cell lines (Ungrin et al., 1999, Abramovitz et al., 2000) that stably express recombinant hEP₁ (Funk et al., 1993) were used in this study (see Materials and Methods). To compare the two cell lines, saturation analysis was performed on hEP₁-5/AEQ17-HEK 293 cells to determine the affinity of PGE₂ for hEP₁ (K_D) as well as the level of receptor expression, as defined by the maximum number of detectable PGE₂ specific binding sites (B_max). The K_D value for PGE₂ was approximately 10 nM and the data conformed to a single-site binding model (data not shown). The level of hEP₁ receptor expression corresponded to a B_max value of 2.8 pmol/mg of protein. Membranes from parental AEQ17-293 cells were also tested for their ability to bind [³H]PGE₂ under the same conditions used for hEP₁-5/AEQ17-293 and were negative (data not shown). The K_D and B_max values for the hEP₁-HEK 293 EBNA membranes used to obtain binding data were 25 nM and 7.3 pmol/mg, respectively, as reported previously (Abramovitz et al., 2000).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Structure of PGE2. Carbon atoms are numbered from 1 to 20 starting from the C-1 carboxyl group.

View larger version (37K): [in this window] [in a new window]   Fig. 2.   Molecular structures of compounds used in the study, with compound numbers as in Table 1.

The rank orders of affinity and potency for all of the compounds tested at hEP₁ are shown in Table 1. In the functional assay, the EC₅₀ value for PGE₂ was 2.90 nM ± 1.86 nM (n = 69). All of the compounds whose responses reached a plateau within the range of concentrations tested were as efficacious as PGE₂ with respect to the maximum response in the aequorin assay (90 to 100% effective relative to  3 µM PGE₂), behaving as full agonists at hEP₁ in this recombinant assay system. A representative example of the data generated in the aequorin assay for PGE₂ and three compounds, enprostil, cloprostenol, and fluprostenol, is shown in Fig. 3. It should be noted that partial agonism could potentially be masked in a cell line with a high receptor reserve such as the one used in this study. Of the 55 compounds tested at hEP₁, only four showed higher affinities than PGE₂ in the binding assay, whereas seven were more potent than PGE₂ in the functional assay.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Concentration response curves from hEP1-5/293-AEQ17 cells challenged with PGE2 (8), cloprostenol (22), enprostil (18), and fluprostenol (49) from three different experiments shown in A, B, and C. The fractional luminescence responses are plotted as a function of log10 agonist concentrations and the EC50 values shown on the graphs are calculated as described under Materials and Methods.

Figure 4 depicts a direct comparison of receptor affinity and agonist potency for all the compounds for which K_i and EC₅₀ values were obtained. Receptor binding results are presented on the x-axis, whereas results from the aequorin assay are presented on the y-axis. Theoretically, the points would be expected to lie along a straight line of slope 1 if agonist affinities and potencies were to correlate perfectly. This relationship is seen for the majority of the prostanoids tested, giving a best fit line of slope 0.87, and a correlation coefficient R² of 0.86 (Fig. 4). The hypothesized "barrier effect" (see below) affecting compounds with stronger binding than the natural ligand would be expected to interfere with linear correlation of the data. If we do not consider the compounds involved in this hypothesis (1, 3, 4, 5 and 17), the slope becomes 1.01 and R² becomes 0.91 (data not shown). The effects of various ligand modifications on receptor activation are presented in Tables 2 through 9, organized by location on PGE₂. Effects on receptor binding are also discussed where they differ from the changes in activity.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Correlation of receptor binding and aequorin activity data. Both data sets are normalized to PGE2 as described under Materials and Methods. Binding results for each compound are shown on the x-axis, whereas functional results from the aequorin assay are shown on the y-axis. The relevant compound reference number (see Table 1 and Fig. 2) is displayed adjacent to each data point. Ki and/or EC50 values for compounds 46 and 50 to 55 could not be ascertained because the concentration-response curves did not plateau at the maximum concentrations tested and hence are not included in the figure.

View this table: [in this window] [in a new window]   TABLE 2 The effects of modifications at the C-15 position on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. For compounds in which no root structure was available (bold), results are reported relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 3 The effects of modifications to the carboxylic acid moiety on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. For compounds where no root structure was available (bold), results are reported relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 4 The effects of modifications at the C-11 position on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. For compounds where no root structure was available (bold), results are reported relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 5 The effects of -tail modifications on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. For compounds where no root structure was available (bold), results are reported relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 6 The effects of chiral inversion at the C8 position on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 7 The effects of saturation at the 5,6 double bond on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 8 The effects of saturation at the 13,14 double bond on potency Results are reported as fold reduction in potency relative to the root structure in each group with parenthetical values relative to PGE2. Compounds are numbered according to Table 1.

View this table: [in this window] [in a new window]   TABLE 9 The effects of modifications at the C-9 position on potency Results are as fold reduction in potency reported relative to the root structure in each group with parenthetical values relative to PGE2. For compounds where no root structure was available (bold), results are reported relative to PGE2. Compounds are numbered according to Table 1.

C-15 Position of PGE₂. The receptor is most sensitive to modifications at the C-15 position of prostanoid analogs (Tables 1 and 2). Methylation at the 15-position of PGE₂ (8) or PGF₂ (15), while maintaining the orientation of the hydroxyl group, yielded the 15(S)-15-methyl derivatives (12, 21) and caused minor reductions in potency of 2.5- and 2.7-fold, respectively. However, inversion of the stereochemistry at this position to the 15(R) conformations of 15(R)-PGE₂ (42), 15(R)-PGE₁ (38), 15(R)-PGF₂ (52), and 15(R)-15-methyl-PGE₂ (40) resulted in substantial reductions in potency of 870, 140, >400, and 220-fold, respectively. Oxidation of the hydroxyl to a ketone, giving the primary metabolite of PGE₂, 15-keto-PGE₂ (28), resulted in 93-fold lower potency than the parent compound [the subsequent metabolite, 13,14-dihydro-15-keto-PGE₂ (33) is 250-fold less active]. In the case of misoprostol free acid (36), a 15-deoxy-16-hydroxy-16-methyl derivative of PGE₁ (14) racemic at the 16 position, potency was reduced 100-fold relative to PGE₁ (14) and 330-fold relative to PGE₂ (8), and binding was reduced 130- and 1500-fold relative to the same two compounds. Interestingly, in competition for [³H]PGE₂ radioligand binding at the human EP₂, EP₃, and EP₄ receptors, misoprostol free acid, is 7, 24, and 29-fold less effective than PGE₂ (Abramovitz et al., 2000). This suggests that at these receptor subtypes, the 15-hydroxyl may not play as critical a role for agonist affinity as it does at hEP₁.

C-1 Position of PGE₂. Modifications to the carboxylic acid moiety at C-1 were also found to have extremely negative effects on the affinity and potency of the compounds tested, although the functional effects were generally more profound (Tables 1 and 3). In all cases in which a direct comparison could be made between compounds with and without such a modification, the modified compound was at least 60- and up to >400-fold less potent as an agonist. All such compounds also exhibited reduced affinity. Similarly, data with various methyl ester compounds supports the importance of the C-1 carboxylic acid for binding to the EP₂ (Stillman et al., 1998; Abramovitz et al., 2000), EP₃ (Audoly and Breyer, 1997; Abramovitz et al., 2000), and EP₄ (Stillman et al., 1998; Abramovitz et al., 2000) receptor subtypes.

2

C-11 Position of PGE₂. Both potency and affinity are sensitive to alterations at the C-11 position (Tables 1 and 4). Removal of the hydroxyl group reduced potency by 43 and 84-fold for PGE₂ (8) and PGE₁ (14), respectively. A much smaller effect was observed in the case of 16,16-dimethyl-PGE₂ (4), a slightly more potent agonist than PGE₂, in which removal of the 11-hydroxy group reduced potency only 2.6-foldanother example of a ligand in which an -tail modification seems to protect against loss of potency (see below). PGA₂ (41) (11-deoxy-10,11-enyl-PGE₂) exhibited 748-fold less potency than PGE₂ (8). Inversion of the stereochemistry at the 11-position also significantly reduced the potencies of compounds, as exemplified by analogs of PGE₂ (8) [11-PGE₂ (26)], PGE₁ (14) [11-PGE₁ (31)] and PGF₂ (15) [9,11-PGF₂ (48)], whose activities dropped 61, 44, and 150 -fold, respectively. Replacement of the 11-hydroxyl with a keto group again greatly reduced the potencies of analogs of PGE₂ (8) [PGK₂ (53)], PGE₁ (14) [PGK₁ (46)] and PGF₂ (15) [PGD₂ (45)] resulting in reductions of >8000, 387, and 110-fold, respectively, further highlighting the importance of the 11-hydroxyl group of PGE₂ for potency at hEP₁.

The -Tail of PGE₂. The -tail is the primary region of interest for modifications yielding increased potency (see Table 5). In particular, modifications that result in a phenyl ring commencing at the former C-18 position, such as the 16-phenoxy and 17-phenyl compounds, exhibit a significant positive influence, although it is not clear whether this is caused by steric or electronic effects. It is noteworthy that these phenylic modifications are capable of contributing significant improvements in potency (more than an order of magnitude) to structures that contain other deleterious substitutions, for example 11-deoxy-16,16-dimethyl-PGE₂ (10) is 24-fold more potent than 11-deoxy-PGE₂ (24). These effects, however, are not observed when the parent structure is a highly potent compound, such as PGE₂.

2

2

2

C-8 Position of PGE₂. Stereochemistry at the C-8 position is important for affinity and potency (Tables 1, 6), with chiral inversion at this position in PGE₂ (8) [to 8-iso-PGE₂ (25)] and PGF₂ (15) [to 8-iso-PGF₂ (39)] giving 53- and 51-fold reductions in potency, respectively. These compounds, known as isoprostanes, are thought to exert some, if not all, of their biological effects through the TP receptor. However, a recent study (Sametz et al., 2000) suggests that they may also act through an EP₁ subtype. Our data also suggests a potential role for EP₁, although the isoprostanes are weak agonists when compared with PGE₂. Specific receptors for isoprostanes have yet to be identified.

The 5,6 Double Bond of PGE₂. Saturation of the double bond present at the 5,6 position has relatively minor effects on affinity and potency (Tables 1 and 7). In general, this modification results in a small decrease in both parameters, exemplified by the reduction of the 5-6 double bond of PGE₂ (8) to form PGE₁ (14) which resulted in a 3-fold decrease in activity. In certain cases, however, the effect is slightly positive [e.g., PGK₂ (53) to PGK₁ (46) and 15(R)-PGE₂ (42) to 15(R)-PGE₁ (38)]. Interestingly, both these compounds are derived from a parent structure bearing a single modification that heavily impairs affinity and potency (see Fig. 2) and that might be expected to result in significant effects on the general conformation of the molecule. One possible explanation is that the increased flexibility resulting from the saturation of the double bond may allow these compounds to adopt a more active conformation.

13,14 Position of PGE₂. Saturation of the double bond at the 13,14 position also had variable consequences for activity (see Table 8), ranging from a slight increase for 13,14-dihydro-PGE₁ (11) to a 45-fold reduction for latanoprost free acid (23) compared with their parent compounds, PGE₁ (14) and 17-phenyl--trinor-PGF₂ (7), respectively. In all cases, the effects on receptor affinities are detrimental, to a greater or lesser degree, as are the effects on potency with the exception of the conversion of PGE₁ (14), already one degree more saturated than the optimal natural ligand, PGE₂ (8), to 13,14-dihydro-PGE₁ (PGE₀) (11), where the difference is not statistically significant.

C-9 Position of PGE₂. Finally, modifications at the C-9 position have small effects on the potency of a given structure, although in certain cases larger effects are seen on the receptor binding results (see Tables 1 and 9). In general, conversion of the 9-keto group to a hydroxy group results in a loss of approximately 1 order of magnitude in potency. Comparison of results from PGF₂ (15) and 9-PGF₂ (16) would suggest that the stereochemistry of the resulting chiral center is not of great significance. The effects of keto-to-hydroxy conversion are not statistically significant on the structure incorporating the phenyl substitution at the C-17 position [17-phenyl--trinor-PGE₂ (5) versus 17-phenyl--trinor-PGF₂ (7)]. Because 17-phenyl--trinor-PGF₂ (7) is potent, this is thought to reflect the limit of improvement by -tail modifications (see discussion above) rather than an absence of effect from the conversion.

2