The Prostaglandin Transporter: Eicosanoid Reuptake, Control of Signaling, and Development of High-Affinity Inhibitors as Drug Candidates

Abstract

We discovered the prostaglandin transporter (PGT) and cloned the human cDNA and gene. PGT transports extracellular prostaglandins (PGs) into the cytoplasm for enzymatic inactivation. PGT knockout mice have elevated prostaglandin E₂ (PGE₂) and neonatal patent ductus arteriosus, which reflects PGT's control over PGE₂ signaling at EP₁/EP₄ cell-surface receptors. Interestingly, rescued PGT knockout pups have a nearly normal phenotype, as do human PGT nulls. Given the benign phenotype of PGT genetic nulls, and because PGs are useful medicines, we have approached PGT as a drug target. Triazine library screening yielded a lead compound of inhibitory constant 50% (IC₅₀) = 3.7 μM, which we developed into a better inhibitor of IC₅₀ 378 nM. Further structural improvements have yielded 26 rationally designed derivatives with IC₅₀ < 100 nM. The therapeutic approach of increasing endogenous PGs by inhibiting PGT offers promise in diseases such as pulmonary hypertension and obesity.

THE HISTORY OF MANIPULATING ENDOGENOUS PROSTAGLANDIN LEVELS

The Ebers Papyrus from approximately 3,500 years ago, followed by Hippocrates, Celsus, Dioscorides, Pliny the Elder, and Galen 1,000 years later, all recommend that persons in pain — particularly women experiencing the pain of childbirth — as well as persons with fever ingest the bark or leaves of the willow tree (1). The willow is of the genus Salix, from which we derive the words salicylate and acetylsalicylic acid (aspirin) (1). Nonsteroidal anti-inflammatory drugs (NSAIDs), which are all derived from aspirin, block the synthesis of prostaglandins (PGs) (2). Thus, considering the trajectory starting with willow bark through aspirin to NSAIDs, humans have historically spent considerable time and effort reducing the synthesis of their endogenous PGs.

That said, topical PGs are used successfully in conditions such as glaucoma and obstetrics (3,4). When PGs of the E series are given systemically, they are generally well tolerated so long as the dose is not excessive (5,6). Analogues of prostaglandin I₂ (PGI₂, prostacyclin) are administered systemically in pulmonary artery hypertension (7). Unfortunately, the pharmacokinetic profile of most PGs renders them not useful for routine systemic clinical use (8,9).

MOLECULAR MECHANISMS OF PG INACTIVATION

Instead of administering exogenous PGs as medicines, our laboratory has focused on increasing the level of endogenous PGs by inhibiting their metabolism. Investigators had known since the 1970s that common prostaglandins (PGs) such as PGE₂ and PGF_2α are stable in blood (10–12). They are metabolized rapidly in tissues by a two-step process of carrier-mediated uptake across the plasma membrane followed by cytoplasmic enzymatic oxidation by 15-hydroxy prostaglandin dehydrogenase (HPGD) (13,14). Because PGs signal a diverse array of downstream physiological events (15), this two-step PG metabolism process must be broadly distributed and must function efficiently to keep PGs constrained to a local region, lest they diffuse from their site of action and signal promiscuously at a distance.

In 1995, our group reported discovering the rat PG transporter (PGT) (16) (subsequent gene names have been SLC21A1, SLCO2A1, and OATP2A1). PGT is a lactate/PG anion exchanger (17) that mediates the energetically active uptake into the cell of PGE₂, PGF_2α, PGD₂, and PGI₂, but not thromboxane (16). We subsequently cloned and characterized the mouse PGT cDNA (18) as well as the human PGT cDNA (19) and gene (20).

We showed that PGT is obligatory for PGE₂ metabolism: reconstitution experiments using a cell line null for both PGT and HPGD revealed that neither alone is sufficient for PG metabolism (21). When we expressed green fluorescent protein — tagged PGT in a polarized Madin-Darby canine kidney (MDCK) cell monolayer, it was sorted to the apical membrane, thus recreating the pattern of PGT expression in the native renal collecting duct, where it serves to direct the release of newly synthesized PGE₂ away from the apical, and toward the basolateral, compartment (22,23).

PGT CONTROLS PG SIGNALING

We performed a global knockout of the mouse PGT gene by flanking exon 1 with LoxP sites and crossing with a mouse transgenic for EIIA-Cre recombinase. Although the pups from PGT heterozygotes were born in a Mendelian genotypic ratio, several days later the PGT null pups were dead. Necropsy revealed that the PGT nulls had patent ductus arteriosus (24), which placed PGT squarely in the PG signaling pathway (25–27). Because systemic PGE₂ levels were high in the PGT null mice (24), the most likely mechanism for the patent ductus is that persistently high post-natal PGE₂ levels had opposed forces causing normal post-natal ductus contraction.

We have also examined directly the ability of PGT to compete for cell-surface PGE₂ and thus control access of this ligand to its plasma membrane receptors. In a reconstituted system, PGT expression reduced PGE₂ signaling through either EP₁ or EP₄ receptors (28).

PGT is also coordinately regulated with cyclooxygenase and/or various other components of the PG signaling systems (29–37).

Taken together, our studies and those of others indicate that PGT plays an essential role in controlling the metabolism of PGs in a broad variety of tissues and organs.

PG SIGNALING MODEL: RELEASE/REUPTAKE

Our present working model is that PG signaling is akin to synaptic signaling (22) (Figure 1). Both neurotransmitters and prostanoids are synthesized by inducible enzymes (38–41). Both systems are characterized by triggered release of ligand into the extracellular compartment (42,43). Both sets of G protein-coupled receptors (GPCRs) use similar molecular signaling and regulatory mechanisms (44–47). And both sets of ligands undergo reuptake by plasma membrane carriers that are located on the cell that released the signaling molecule (16,22,48–50).

Fig. 1.

Model comparing neurotransmission with proposed eicosanoid signaling. In neurotransmission, an action potential triggers the release of neurotransmitter that has been compartmentalized in vesicles. The neurotransmitter undergoes reuptake by a carrier, followed by cytoplasmic oxidation or reuptake into vesicles. In prostaglandin signaling, an agonist induces triggered release via activation of cytoplasmic phopholipase A2 (cPLA2), causing delivery of archidonate to cyclooxygenase (Cox). We propose that prostaglandin signaling also involves PG reuptake via PGT and that, as in neurotransmission, PG reuptake is followed by cytoplasmic oxidation. The dotted vertical line implies compartmentalization between the PG synthesis/ release pathway on the one hand and the PG reuptake/oxidation pathway on the other. The nature of the compartmentalization remains to be determined. Adapted from Nomura et al (22) with permission.

In the neurotransmitter model system, there is good evidence that varying the rate of ligand reuptake varies the signaling (51–53). Indeed, these effects form the basis for treatment of psychiatric conditions with serotonin reuptake inhibitors (54,55). Accordingly, we asked whether small molecule inhibitors of PGT might be useful medicinally insofar as they would increase the levels of endogenous PGs.

PGT NULL MICE AND HUMANS

If we are to block PGT medicinally, it would be helpful to know phenotypic details of PGT null mice and humans, in other words, a “worst case scenario” of PGT inhibition.

Mice null at the PGT locus, once rescued through the perinatal period of ductus arteriosus closure, appear to grow normally without obvious pathology (24). Admittedly, more detailed phenotyping awaits improved methods of rescuing the null pups.

On the other hand, in the past 2 years there have been reported some 57 human subjects who have been null at the PGT locus from birth (56–64). These reported subjects, ranging in age from 7 to 88 years, have had no symptoms until around the time of puberty. Then, at an average age of 16 years, the males (but not the females) have developed thickened cephalic and facial skin, digital clubbing, and periosteal calcification. This condition has been termed variously “pachydermoperiostosis” or “hypertrophic osteoarthropathy.”

Systemic PGE₂ levels are elevated, and five cases of 57 have been reported to have bone marrow fibrosis and anemia (61).

Despite the well-known role of PGs in pain signaling (65,66), and although approximately 40% of reported PGT null subjects do report pain, it is not generalized pain, rather the pain is confined to sites of periosteal calcification. Similarly, despite the well-known role of PGs in fever induction (67), subjects null for PGT do not have fever.

In the context of a large literature on the role of PGE₂ in colon carcinogenesis (68–70), to date one PGT null subject has been reported to have colon carcinoma (56), a rate among the 57 reported subjects that appears to be equivalent to the rate of lifetime colon cancer development in the general population (71). That PGT gene expression is markedly turned off in colon cancers (72) suggests that further decreases in PGT, either from genetic mutations or from small molecule inhibitors, may not further influence the process of carcinogenesis. Clearly, however, further work in this area is needed.

Thus, at present it appears that the clinical consequences of systemically inhibiting PGT, either genetically or pharmacologically, are relatively minor. Based on these finding, we have pursued PGT as a prospective drug target.

DEVELOPMENT OF PGT INHIBITORS

To develop PGT-specific inhibitors with high affinity, we screened a library of 1842 triazine compounds using Madin-Darby canine kidney cells stably expressing rat PGT. In this screening, we found several effective PGT inhibitors. Among them, the most potent inhibitor had an IC₅₀ of 3.7 μM. These inhibitors allowed us to isolate the efflux process of PGE₂ and to show that PGT does not transport PGE₂ outwardly under physiological conditions (73). The molecule with the highest affinity from the initial library screening was used in vivo to show that PGT regulates local cerebral blood flow via lactate-PG exchange (74).

We performed subsequent structural activity relationship (SAR) studies and developed a better inhibitor, T26A, with IC₅₀ = 378 nM (75). In a proof of principle set of experiments, T26A injected into rats doubled the circulating [PGE₂]; reduced circulating PGE₂ metabolites by 50%; and slowed the metabolism of exogenously injected PGE₂ (75). At the conclusion of our report on T26A, we wrote:

“In addition to being a powerful basic research tool for investigating the fundamental role of PGT in PG metabolism and signal termination, a potent PGT inhibitor such as T26A provides a starting point for the development of therapeutic agents targeting PGT. Because…PGT regulates the metabolism of PGs, a specific inhibitor of PGT could potentially be developed for clinical applications” (75).

Since then, our further SAR work has yielded more than two dozen rationally designed derivatives with IC₅₀ < 100 nM (unpublished observations).

We are currently using these small molecular PGT inhibitors in preclinical studies of disease for which the literature suggests that PGs would be beneficial (eg, pulmonary artery hypertension and obesity).

DISCUSSION

Zeidel, Boston: Is there only one of these transporters, or are there different types? Because you could imagine that if you knock this out all over the body there might be all kinds of adverse effects. But — like the Cox-2 which didn't turn out quite so well — the idea of being able to inhibit one kind or another might have a significant impact on the therapeutic efficacy of this strategy.

Schuster, New York: There are about five or six other transporters that have been shown to transport prostaglandins. Usually people just look at PGE2, but the affinities are quite low. This has an affinity of about 80 nanomolar, and those others have affinities up in the low micromolar range. And, if you do a head-to-head flux comparison, the other transporters really are not in the game. So this does seem to be the one. I would point out that the phenotype is pretty much the same whether you take out the transporter or you take out the downstream oxidase enzymes. It looks like there is pretty much one pathway for this inactivation.

Tweardy, Houston: At this point in time as you propose or are in the process of doing your IND-enabling toxicity studies in rats and, I presume, dogs, what is the duration of exposure you are going to use for those studies in anticipation of a phase 1?

Schuster, New York: There are a couple different ways to think about this as a drug strategy. One is an acute strategy; a one-time use. So we are engaged in some studies, for example, in prophylaxis of contrast-induced nephropathy. Or there are topical sorts of applications. And then long-term applications so we are exploring a number of these, and the question is, where is the efficacy maximized and where is the toxicity minimized?