GPCR ligand-dendrimer (GLiDe) conjugates: future smart drugs?

Abstract

Unlike nanocarriers that are intended to release their drug cargo at the site of action, biocompatibile polyamidoamine (PAMAM) conjugates are designed to act at cell surface G protein-coupled receptors (GPCRs) without drug release. These multivalent GPCR Ligand-Dendrimer (GLiDe) conjugates display qualitatively different pharmacological properties in comparison to the monomeric drugs and might be useful as novel tools to study GPCR homo- and heterodimers and higher aggregates. The structure of the conjugate determines the profile of biological activity, receptor selectivity, and physical properties such as water solubility. Prosthetic groups for receptor characterization and imaging can be introduced without a loss of affinity. The feasibility of targeting multiple adenosine and P2Y receptors for synergistic effects has been shown. Testing in vivo will now be needed to explore the effects on pharmacokinetics and tissue targeting.

Introduction

The mechanism of action of at least one quarter of the drugs on the market involves modulation of rhodopsin-like GPCRs on the cell surface [1]. A recurrent issue in the discovery and clinical development of new GPCR ligands, agonists and antagonists is the occurrence of a given receptor in multiple tissues, leading to side effects. The same receptor might occur in a target tissue, in which a desired pharmacological effect can be demonstrated, and in off-target tissues, in which the effect can be less than beneficial or even harmful. A means of designing GPCR modulators that function as “smart drugs” is needed [2]. This need is especially evident for receptor classes such as the purine receptors, which encompasses both adenosine receptors (ARs) [3] and purinergic P2Y receptors [4] for extracellular nucleotides that are expressed at varying levels and in diverse combinations in nearly every organ. Even with the promise of structure-based design of GPCR ligands [5,6] (now with an increasing number of X-ray crystallographic structures available), tissue- or organ-selectivity would provide a great advantage therapeutically.

Pro-drug approaches have sought to achieve this goal of organ-selective delivery of GPCR ligands [7]. However, their success is dependent on the presence of a specific enzyme for the release of the free drug species in the target tissue, which is often unfeasible. Instead, the field has adopted the approach of using covalent conjugates of chemically functionalized GPCR ligands tethered from macromolecular carriers, which emerged from the early work of Ringsdorf and others [8]. The polyamidoamine (PAMAM) dendrimers form a particularly versatile and biocompatible class of polymeric drug carriers. First introduced by Tomalia [9], these tree-like polymers have found use in diagnostic imaging and in research for delivering nucleic acids and peptides to cells for internalization [10,11]. Charge and size parameters that affect internalization of PAMAM dendrimers have been studied [12]. The PAMAM dendrimers can be tailored to a wide range of molecular weights by adding outer layers that bifurcate in successive “generations” that multiply the number of terminal groups geometrically, and the nature of the terminal functionality can be altered for conjugation purposes or for tuning pharmacokinetic properties in vivo. Enhanced stability of ligands when conjugated to PAMAM dendrimers has been noted [13]. Most of the examples of use of PAMAM dendrimers as drug carriers involved internalization of the complex followed by dissociation or cleavage of the drug cargo to act intracellularly. In one example, arginine-glycine-aspartic acid (RGD) peptides were conjugated to PAMAM dendrimers for binding to cell-surface integrin receptors prior to internalization [14]. With GPCR Ligand-Dendrimer (GLiDe) conjugates [15–21], we have taken a different approach that requires strategic functionalization of the small molecular drug such that it displays a desired biological property extracellularly while still tethered covalently to peripheral groups of the dendrimer. A peptide hormone was also coupled to a PAMAM dendrimer to serve as a targeting moiety through binding to its GPCR [22].

PAMAM conjugates as multivalent ligands of GPCRs

GLiDe conjugates have been described recently as interacting with a variety of adenosine and P2Y receptors (Table 1). These are the first covalent conjugates of PAMAM dendrimers with small molecular GPCR agonists and antagonists that display high potency at a target receptor (nM to µM range). Dissociation of the drug and/or internalization of the complex are generally not necessary or desirable by this approach, which maintains or enhances the binding affinity of the functionalized congeners while covalently bound [23]. Thus, it is now possible to have a multivalent display of potent tethered ligands, typically attached at the peripheral amino or carboxylic groups through amide linkages. Although the chemical structures are based on stoichiometrically-defined substitutions on well-defined polymeric carriers, i.e. averaged structures, the pharmacology of these conjugates can be characterized consistently and reproducibly. Nevertheless, there is a need for the preparation of more chemically defined or monodisperse GLiDe conjugates.

Table 1.

Examples of GLiDe conjugates for adenosine and P2Y receptors.

Receptor(s) (ligand linkage and PAMAM generation)	Cell type	Effect	Potency (nM)a	Ref.
A2A agonist (amide, G3)	Human platelet HEK-293 cells	Antithrombotic	151	15,17
		cAMP generation	11.2, 151b
A2A agonist (thiourea, G3)	HEK293 cells	cAMP generation	96–152b	16
A3 agonist (amide, G2.5)	Cardiomyocytes	Antiapoptotic protection	35	21
A3 agonist (triazole, G4)	CHO cells	Inhibition of cAMP	0.14b	20
P2Y1 antagonist (triazole, G4)	Human platelet	Antithrombotic	1000–4000c	28
P2Y14 agonist (amide, G6)	HEK293 cells	Inhibition of cAMP	0.8	19
P2Y14 agonist (amide, G4) + A3 agonist (triazole)d	HEK293 and CHO cells	Inhibition of cAMP	2.24 (P2Y14), 1.36 (A3), 39.5b (A3)	20

^aCalculated on the basis of the concentration of the conjugate.

^bBinding affinity at human AR.

^cAt 4 µM inhibited 94% of platelet aggregation induced by 5 µM ADP.

^dStructures of the terminal groups of this bifunctional conjugate are shown in Figure 3.

A rationale for the GLiDe conjugate approach is shown in Figure 1. The application of nanocarriers for selective tissue delivery in cancer and other diseases, including in the brain, has been reviewed [24,25]. Similarly, GLiDe conjugates may provide advantages over monomeric drugs. However, the biocompatibility of GLiDe conjugates in vivo and the feasibility of these drugs to cross the blood-brain barrier (and other membrane barriers) in order to treat central nervous system (CNS)-associated diseases remain unexplored.

Figure 1.

Rationale for exploring GPCR Ligand-Dendrimer (GLiDe) conjugates. (a) Bridging of two adjacent protomers in homo- or heterodimeric GPCR aggregates. A single strategically tethered ligand B might span two binding sites. By incorporating multiple ligands A and B, a preference for the heteromer may be anticipated. Thus, there may be selective binding in tissues expressing combinations of GPCRs. (b) Molecular model of a GLiDe conjugate (e.g. having A_2AAR agonist properties in Kim et al. [15]) bridging the two protomers of a homodimeric A_2AAR [reprinted from ref. 33 with permission from Elsevier]. (c) Prosthetic groups may be incorporated for targeting (e.g. an antibody) or therapeutics or as reporter groups for diagnostics (e.g. a fluorescent or luminescent group) without loss of receptor binding. (d) Fluorescent microscopic image of A₃AR expressed in CHO cells [18]. The left panels are fluorescent images, and right panels are light microscopic images of the same fields; upper panels – in cells expressing the human A₃AR, lower – untransfected cells.

Dimers of GPCRs might be homomeric or heteromeric, and this association, as well as higher order aggregation, can have a major effect on the pharmacological behavior [26]. The multivalent GLiDe conjugates might prove to be useful tools for investigation of GPCR aggregates, given the predicted ability of these conjugates to bridge multiple binding sites within the transmembrane region. Functional negative allostery between protomers in dimeric GPCRs has been detected [27]; this raises the issue of how dendrimer-bound agonists occupying neighboring binding sites would compare with agonist/antagonist combinations. Furthermore, the design of potent GLiDe conjugates is suggestive as a means of drug targeting, perhaps on the basis of an antibody complex. It is also possible that conjugates containing two different pharmacophores would show enhanced affinity for the corresponding heterodimers or permit co-activation of two separate receptors on the same cell. Reporter groups, intended for imaging or diagnostic purposes, might also be incorporated in the conjugates without loss of the high affinity in binding to the receptor.

The first example of a GLiDe conjugate in the purine receptor family is a nearly fully substituted third generation (G3) PAMAM conjugate of a potent, carboxy functionalized A_2A AR agonist, CGS21680 [15]. This conjugate not only binds to the receptor but also activates it to inhibit platelet aggregation (a known action of this AR subtype), thus serving as a prototypical example of a novel antithrombotic agent. At moderate (µM) concentrations, no internalization in the platelets was observed using flow cytometry, thus serving the overall strategy of direct action at cell surface receptors.

Another means of inhibiting platelet aggregation pharmacologically is to antagonize the action of the native agonist ADP at either P2Y₁ or P2Y₁₂ receptors, the latter being the protein target of the blockbuster drug Plavix® (clopidogrel) [4]. We recently overcame synthetic difficulties to demonstrate that a P2Y₁ receptor antagonist, an adenine nucleotide derivative, could be incorporated into a GLiDe conjugate that resulted in an anti-aggregatory effect on human platelets [28]. The inability to maintain in a PAMAM conjugate the nM affinity of the parent monomeric P2Y₁ receptor antagonist MRS2500 was ascribed using ligand docking to a conformational flip of the adenine moiety in the receptor-bound form in this strucutural series. Nevertheless, the moderate antithrombotic activity demonstrated that the GLiDe conjugate approach is equally applicable to nucleotides as well as nucleosides, to receptor antagonists as well as agonists.

The polymeric multivalency of GLiDe conjugates provides unusual pharmacodynamic and pharmacokinetic properties. One of the objectives of this approach is to enhance receptor potency and selectivity in the multivalent conjugates in comparison to the monomers, and such qualitative differences in selectivity were already observed. A nucleoside conjugate of an N⁶-derivatized adenosine attached to a G2.5 PAMAM dendrimer, intended for selectivity at the A₁AR, instead displayed unexpectedly high affinity at the A₃AR [18]. A₃AR agonists are already in clinical trials for treating inflammatory conditions and cancer [29]. This nM affinity permitted the detection by fluorescence microscopy of the punctate occurrence of the A₃AR expressed in CHO cells (Figure 1d). A larger molecular weight variation of this structure was shown to be protective in HL-1 mouse cardiomyocytes exposed to H₂O₂ for induction of apoptosis [21]. The GLiDe conjugate was at least 100-fold more potent than the monomeric nucleoside in protection, which was dependent on the transfection of the cells with the human A₃AR. This enhanced protective potency was not manifested by increased binding affinity but might result from functional amplification of the activation signal.

If multiple anchor points exist for the GLiDe conjugates, a large increase in affinity in comparison with the monomers would be expected, which was demonstrated for agonists of the A₃AR and the P2Y₁₄ receptors [19,20]. The affinity enhancement of the A₃AR GLiDe conjugates depended on the chemical linkage used. Thus, a functionalized nucleoside was thethered through a triazole ring formed using Cu^I-catalyzed click chemistry, which is a convenient and efficient method for joining molecules that contain azido (-N₃) and terminal alkyne (-C≡CH) groups [30]. The resulting conjugates displayed nM or sub-nM affinity, whereas amide-linked conjugates of the same pharmacophore were considerably less potent. In the case of the Gi-activating P2Y₁₄ receptor, amide-linked GLiDe conjugates of the agonist UDP-glucuronic acid [19] were sufficient for high potency in inhibition of cyclic AMP in transfected HEK cells, which varied widely as a function of the percent substitution of the PAMAM dendrimer and the generation (Figure 2). This potency enhancement is suggestive of simultaneous binding to more than one P2Y₁₄ receptor, but this aspect must be further explored for an unequivocal demonstration. Nevertheless, the nM potency of these nucleotide conjugates allows the incorporation of reporter groups in the drug-bearing dendrimers of fluorescent, metal-chelating, or biotin moieties, without a loss of biological activity.

Figure 2.

Nucleotide dendrimer agonists of the P2Y₁₄ receptor: Dependence of potency on the generation of the PAMAM dendrimer carrier and its percent substitution [19]. G2.5 and G5.5 PAMAM dendrimers have a total of 64 and 256 terminal carboxylic acid groups prior to ligand coupling. G3 and G6 dendrimers have a total of 64 and 256 terminal primary amino groups prior to ligand coupling.

PEGylated unimolecular micellar conjugates

A difficulty encountered with the initial amide- or thiourea-linked A_2AAR agonists was reduced aqueous solubility, likely due to H-bonding-induced aggregation of the dendrimer [15,16]. Aggregation can reduce bioavailability and increase nonspecific membrane binding. Therefore, we tested whether appending polyethylene glycol (PEG) chains to free terminal groups of the carrier would still allow efficient binding of the nucleosides to their binding sites [17]. The effects of increasing PEG chain length and the degree of substitution on the receptor-binding affinity and activation were systematically probed. Some of the conjugates were modestly enhanced in potency compared to the monomeric agonist. Although there was a small reduction in affinity with the longer (2000 MW) PEG chains, solubility could be enhanced without losing receptor binding. Therefore, these antithrombotic PEGylated GLiDe conjugates can be considered unimolecular micelles, in which the outer hydrophilic layer determines the interface for interaction with medium, membrane surfaces, and biopolymers. The degree of internalization will generally depend largely on the cell type and other conditions, but in platelets it was promoted by amine terminal groups on the carriers, which could be reduced by acetylation of the remaining peripheral amines. Therefore, physicochemical properties of the conjugates, such as size, water solubility and charge, could be modulated leading to predictable changes in the biological activity.

Multiple pharmacophores on the same carrier

These two examples of agonists of the A₃AR and the P2Y₁₄ receptor were used as a test of the ability to combine two different pharmacophores into the same GLiDe conjugate [20]. Thus, the conjugate shown in Figure 3 was able to bind to and activate either receptor with high potency. We recently reported that the RBL-2H3 mast cell line, known previously to express functional A₃ARs, also expresses the P2Y₁₄ receptor, and both receptors facilitate the release of inflammatory mediators [31]. It is not yet known whether A₃AR/P2Y₁₄ heterodimers occur or whether this bi-functional multivalent conjugate acts in a synergistic fashion in RBL-2H3 cells. Nevertheless, the ability to potently activate both receptors suggests that disease treatment using specific combinations of GPCR ligands is possible.

Figure 3.

A nucleoside/nucleotide conjugate of a PAMAM G4 dendrimer, which functions as a dual agonist of the P2Y₁₄ receptor and the A₃AR [20]. The synthetic sequence required the preparation of the bifunctional GLiDe conjugate (c). First, the nucleotide UDP-glucuronic acid (a) was linked to peripheral amino groups on the dendrimer by amide bond formation, followed by conjugating the nucleoside (b) to peripheral azido groups by facile click chemistry [30]. The latter step involved cyclization of a terminal acetylene moiety on the prefunctionalized nucleoside (b) with a terminal azido group on the dendrimer in the presence of cuprous sulfate to form a stable triazole ring. The alkyne group that is adjacent to the adenine ring in (b) maintains the A₃AR affinity, but does not participate in the click reaction. The stoichiometry of substitution was on average ~31 nucleotide moieties and ~33 adenosine moieties per 64 possible terminal positions. The bifunctional conjugate (c) potently activated the human A₃AR expressed in CHO cells and the human P2Y₁₄ receptor expressed in HEK-293 cells.

Bivalent ligands have been reported as specific pharmacological tools for homo- and heterodimeric GPCRs [32]. However, tethering two pharmacophores in a manner that is expected to achieve the precise geometry needed to reach both binding sites is a risky endeavor, i.e. difficult to predict a priori due to uncertainty in spatial characteristics of receptor dimers in relation to conformation of the tether. A GLiDe conjugate that contains a statistical distribution of geometries separating two pendant ligands is more likely than a pure substance with a single chain length to contain a suitable combination to bridge two binding sites on adjoining GPCRs protomers. Also, many of the GPCR biological systems have “switchlike” properties that are dictated by interacting ligands. These properties of GPCRs might make predicting a priori GLiDe conjugate actions difficult.

Conclusions

We have introduced a general methodology for designing in stages multivalent GLiDe conjugates that might display qualitatively different pharmacological properties in comparison with the monomeric drugs. Although the ability of these polymeric derivatives to cross biological membranes is likely reduced and structural heterogeneity is present, there are advantages over the monomeric drugs. Nevertheless, large hurdles must be addressed before significant therapeutic advances are achieved. The potential of this approach for use in disease models in animals must still be demonstrated, and basic questions remain concerning cellular uptake, the pharmacological effects of ligands binding to neighboring GPCR proteins, the alteration of selectivity based on composition of the conjugates, and side effects. This methodology has been applied to study adenosine and P2Y receptors, but potentially it could be used for any GPCR. Potent biological activity has been observed in vitro, and the receptor selectivity can be modulated by the structure of the conjugate. The physical properties of these biocompatibile PAMAM conjugates, such as surface polarity, which affects aggregation and internalization, can also be readily modulated by structural changes. Prosthetic groups for receptor characterization and imaging can be introduced without loss of affinity. Novel tools are needed to study GPCR aggregates and these conjugates are also of interest in this context. Testing in vivo will now be needed to explore the effects on pharmacokinetics and tissue targeting, in the hope that future “smart drugs” will arise from the GLiDe conjugate approach.

Glossary

AR

adenosine receptor (family of 4 GPCRs activated by extracellular adenosine and its nucleoside analogues)

CGS21680

2-[p-(2-carboxyethyl)phenyl-ethylamino]-5′-N-ethylcarboxamidoadenosine

GLiDe

covalently conjugated GPCR Ligand-Dendrimer

GPCR

G protein-coupled receptor

Ligand docking

process of computational identification of an energetically favorable binding mode of a small molecule in its receptor site

MRS2500

(1′R,2′S,4′S,5′S)-4-(2-iodo-6-methylamino-purin-9-yl)-1-[(phosphato)-methyl]-2-(phosphato)-bicyclo[3.1.0]hexane

PAMAM

polyamidoamine

P2Y

family of 8 GPCRs by various extracellular adenine and uracil nucleotides

RBL

rat basophilic leukemia