Toward Multivalent Signaling across G Protein-Coupled Receptors from Poly(amidoamine) Dendrimers

Abstract

Activation of the A_2A receptor, a G protein-coupled receptor (GPCR), by extracellular adenosine, is antiaggregatory in platelets and anti-inflammatory. Multiple copies of an A_2A agonist, the nucleoside CGS21680, were coupled covalently to PAMAM dendrimers and characterized spectroscopically. A fluorescent PAMAM-CGS21680 conjugate 5 inhibited aggregation of washed human platelets and was internalized. We envision that our multivalent dendrimer conjugates may improve overall pharmacological profiles compared to the monovalent GPCR ligands.

Graphical abstract

Small-molecule ligands for a G protein-coupled receptor (GPCR) generally bind within its heptahelical transmembrane domain to trigger intracellular signaling pathways (1, 2). Accordingly, the surface contact of a distal structural unit of a bound ligand, away from the pharmacophore region, with a GPCR at/near its extracellular domain (i.e., a secondary interaction) may impart additional selectivity and affinity. An approach to designing ligands for GPCRs by applying structural modification at their permissive distal sites has led to the concept of functionalized congeners (3, 4). For instance, ligands for GPCRs may require attachment to a carrier molecule for delivery, and the chemical functional groups and geometry of a linker moiety can be varied, as guided by receptor affinity.

Dendrimers (5–7) are treelike macromolecules which possess favorable characteristics: structural integrity, control of the component functional groups and their corresponding physical properties by chemical synthesis, feasibility to conjugate multiple functional units at the periphery and the interior, and a low enzymatic degradation rate. For the past 15 years, dendrimers have been used for biomedical research (8–12) including gene transfection (polycationic nature) (13–16), drug delivery (targeted/controlled release, encapsulation, or covalent/electrostatic attachment) (17–21), protein–carbohydrate interactions (multivalent effect) (22–27), medical diagnostics (signal amplification) (12, 28), and tissue engineering (29–31). Poly(amidoamine) (PAMAM) dendrimer is composed of aliphatic amino and amido groups, and its relatively biocompatible nature has found many applications in biomedicine (32–35). Although the heterogeneity of commercial PAMAM dendrimers may hinder the structural characterization at the monomolecular level (36), usage of PAMAM dendrimers may provide quick and easy access to gauge the applicability of dendrimers as advanced therapeutic agents before progressing to tailor-made dendrimers.

Here, we report the first example of dendrimer applications to induce the multivalent intracellular signal transduction across GPCRs. Adenosine receptors (ARs) are members of the Group A GPCR family that are involved in various disease states such as inflammation, cancer, cardiovascular damage, and nervous system disorders (37–39). Four subtypes of ARs, A₁, A_2A, A_2B, and A₃, have been identified to date, with each manifesting a unique pharmacological profile, tissue distribution, and effector coupling. To test the feasibility of our approach, the carboxylic acid derivative CGS21680¹(1 (40–42)), an A_2A AR agonist, was selected as a ligand for the attachment to the peripheral amino groups of a third generation (G3) PAMAM dendrimer with an ethylenediamine core (2). Here, the PAMAM dendrimer linked to CGS21680 through its aliphatic amino end group can be considered a part of the distal modification under the functionalized congener concept to modulate the biological activity of the original ligand (3, 4).

Our plan for the conjugation of CGS21680 to the PAMAM dendrimer was to utilize its distal carboxylic acid group extending from the C-2 position of the adenine base, which was shown to be a site suitable for attachment that is tolerated in receptor binding (3). The synthetic feasibility to make PAMAM-CGS21680 conjugates through a peptide coupling method was first examined at a monomeric level using CGS21680 in its unprotected form. For peptide coupling, PyBOP¹ was adopted as a coupling reagent. PyBOP generally allows rapid condensation in high yield unless a significant steric challenge is involved. Unlike a somewhat sluggish and unselective carbodiimide-mediated coupling, which acts essentially through dehydration (i.e., forming urea as a byproduct), PyBOP does not generate any esterified products between a free hydroxyl and a carboxylic acid. Thus, CGS21680 1 was treated with an excess of base, followed by an equimolar amount of N-Boc-ethylenediamine¹ in DMF¹ (Scheme 1). PyBOP (0.8 equiv) was added to this mixture to form an activated intermediate in situ and to achieve the coupling exclusively at the primary aliphatic amine. Thus, the preactivation step of the carboxylic acid group commonly applied before the addition of amine was avoided in order to minimize the formation of intra- or intermolecular cross-coupled products involving the adenine N⁶amine. Indeed, neither the ¹H NMR nor the mass spectrum of the crude mixture showed the formation of any unwanted side products, and this reaction produced the desired compound 3 in 90% isolated yield.

Scheme 1

Next, this PyBOP coupling strategy was applied to the PAMAM dendrimer. We began by preparing the dendrimer conjugate 4, which was fully amide-coupled with CGS21680 (Scheme 2). A slight excess of CGS21680 was added to a basic solution of PAMAM 2, which was then treated with 32 equiv of PyBOP as a DMSO-d₆¹ solution. Deuterated DMSO was used as the reaction solvent in order to avoid later complications in the NMR analysis based on integration in DMSO-d₆. The reaction continued for 22 h, and the product was purified by size exclusion chromatography (SEC) in DMF to give the desired compound 4 in quantitative yield as confirmed by ¹H NMR. Here, the peak corresponding to the methylene α- to the carbonyl at the CGS21680 linker terminus was well-resolved at 2.33 ppm and exhibited the maximal shift upon coupling, similar to the result observed for the monomeric analogue 3. The integration value of a methyl peak from CGS21680 at 0.95 ppm (96.3 H) relative to a PAMAM methylene peak “c” at 2.18 ppm (120 H)—an internal standard of PAMAM G3 for characterization by integration—indicated that the reaction proceeded completely to produce the desired product 4 bearing 32 CGS21680 groups.

Scheme 2

To aid in the microscopic visualization in biological assays, a fluorescent dendrimer derivative 5 was then prepared. We chose Alexa Fluor 488 as a fluorophore, which has similar absorption/emission maxima to those of fluorescein, but is reported to have a higher photostability, pH insensitivity, and good water solubility (43, 44). Thus, an equimolar amount of Alexa Fluor 488 was added to PAMAM dendrimer 2 as a 5-carboxy tetrafluorophenyl ester 6 in DMSO-d₆ (Scheme 2). A small amount of red precipitate was observed almost instantly upon addition of 6. However, removal of this precipitate by filtration did not affect the recovery of an anticipated mass balance when calculated after the full conjugation of the ligands in the next step to prepare 5. Analysis of the fluorescent intermediate 7 by ¹H NMR integration was difficult, and thus, the conjugation of 1 equiv of Alexa Fluor 488 was assumed on the basis of the stoichiometry of addition. The filtered crude mixture 7 was used for the next step without any purification. Subsequent attachment of CGS21680 to 7 was carried out using the same PyBOP coupling strategy as for 4, but by adding 31 equiv of PyBOP instead. The fluorescent PAMAM-GS21680 conjugate 5 was purified by SEC and was similarly characterized by ¹H NMR integration to confirm that, on average, 31 CGS21680 groups were attached to the dendrimer (Supporting Information).

PAMAM-CGS21680 conjugates 4 and 5 were then characterized by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (see Supporting Information). Various MALDI matrices were attempted; however, in all cases, the intensity of the desired peak was generally weak, and the peak shape was overly broad due to the heterogeneity and possible fragmentation of PAMAM under the applied MALDI condition. Among those attempted matrices, DHB¹ and THAP¹ gave the best results for analysis. A broad peak corresponding to the half-size of the desired compound 4 or 5 was detected, which may have originated from either a doubly charged species or the fragmentation of the desired compound at the middle of the ethylenediamine core as suggested previously (45). In comparison to the ¹H NMR integration-based characterization, MALDI underestimated the average molecular weights (MWs) for both 4 and 5 in either matrix when the broad peak region above the MW of a half-size dendrimer was considered for calculations (Table 1). MALDI of a commercial PAMAM dendrimer 2 obtained under similar conditions also underestimated the average MW by ca. 1000 Da. When the MALDI-estimated average MW of PAMAM 2 was subtracted from those of 4 and 5, dendrimers 4 and 5 were found to have 28–30 and 27–28 CGS21680 attachments, respectively, thus a slight underestimation when compared to those calculated by NMR. The deviations in MALDI-estimated MWs from those determined by NMR integration may have resulted from the difference in the tendency to form multiple matrix-adducts between 2, with many free amino groups, and the fully conjugated dendrimer 4 or 5. In either matrix, the polydispersity of PAMAM-CGS21680 conjugate 4 or 5 was relatively low, ca. 1.02–1.03, and was similar to the value obtained for PAMAM dendrimer 2. This result further supports that the conjugation of CGS21680 was nearly complete for both 4 and 5, by reacting at most of the available peripheral amino groups of the PAMAM dendrimer.

Table 1.

Comparison of Molecular Weights and the Number of CGS21680 Groups Attached Determined by NMR and MALDI MS

			MALDI
	NMR		DHB matrix				THAP matrix
cmpd	no. of CGS21680	MWb	M n c	M w d	PDIe	no. of CGS21680f	M n c	M w d	PDIe	no. of CGS21680f
2	0	6909	5772	5909	1.02	0	5956	6085	1.02	0
4	32	22317	18839	19391	1.03	28	19789	20449	1.03	30
5 a	31	22352	19050	19502	1.02	27	19576	20044	1.02	28

^aCounterions of sulfate groups for Alexa Fluor 488 were assumed to be protons.

^bPAMAM dendrimer 2 used for calculation here was assumed to have 32 peripheral amino groups and no structural defects.

^cNumber-average molar mass.

^dWeight-average molar mass.

^ePolydispersity index.

^fNo matrix adduct was considered for the estimation.

Preliminary conformational studies of PAMAM-CGS21680 conjugate 4 were performed by nuclear Overhauser enhancement (NOE) experiments and computer modeling (see Supporting Information). PAMAM interiors were predicted to maintain a relatively compact structure by forming exchangeable multiple hydrogen bonds in water or methanol (32). Unlike water molecules, which may interact mostly with the charged ammonium ions at the surface of PAMAM, polar aprotic DMSO may be more likely to penetrate into the PAMAM interior to disrupt the hydrogen-bonded network. Indeed, NOE cross-peaks were observed between methylene groups of PAMAM (“b” and “c”) and the protons from the ribose moiety (i.e., methyl, H-1′, H-2′, H-3′, and H-4′) of CGS21680 in DMSO-d₆ (Supporting Information), suggesting a possible internalization of the ligand into the PAMAM interior. In addition, the energy-minimized structure of 4 (HyperChem7.5.2, Amber force field) exhibited a relatively large volume in the interior, which is sufficiently spacious to accommodate several internalized ligands simultaneously (46). Eighteen hydrogen bonds were detected in the interior of the PAMAM-CGS21680 conjugate 4. The overall shape of 4 was somewhat ellipsoidal (32), with the diameter ranging ca. 65–85 Å.

Next, our fluorescent PAMAM-CGS21680 conjugate 5 was subjected to in vitro functional assays. We specifically chose the platelet aggregatory system (47–49) to examine the potential pharmacological value of our dendrimer-based multivalent ligand carriers for GPCR signaling. CGS21680, an A_2A AR agonist, is already known to display a potent antiaggregatory effect in human platelet preparations (48, 49). The degree of platelet aggregation was determined by aggregometry after addition of ADP¹ to a washed human platelet suspension (50) containing either the PAMAM-CGS21680 dendrimer conjugate 5 or CGS21680 monomer 1 as a DMSO solution (2 μM) or the same volume of pure DMSO as a control (Figure 1) (Supporting Information). Our preliminary findings indicated that ADP-induced platelet aggregation, which occurs through the nucleotide binding to P2Y receptors on the platelet surface, was inhibited in the platelet suspension containing either dendrimer 5 or monomer 1 to a similar degree (i.e., 61% inhibition by 5; 57% inhibition by 1, both at 2 μM), whereas platelets treated with DMSO alone aggregated in response to ADP. The inhibitory effect was the same whether the dendrimer conjugate was incubated for 5 or 60 min. Interestingly, the platelet aggregation response in the presence of dendrimer 5 exhibited a slower onset in comparison to that of control or the platelet suspension treated with monovalent CGS21680. Moreover, the fluorescent dendrimer conjugate 5 was clearly observed inside the platelets under a fluorescent confocal microscope (Figure 2), when the platelet suspension was incubated with dendrimer 5 and washed to remove any free dendrimer 5 in the extracellular medium. Thus, the extracellularly administered dendrimer conjugate is effectively internalized in platelets. The internalization of intravenously injected macromolecules into platelet granules has been described (53). Future studies will be directed toward determining the structural features of the dendrimer conjugates that affect internalization.

Figure 1.

Platelet aggregation studies at 37 °C in the presence or absence of dendrimer 5 (Supporting Information). A suspension of human washed platelets was incubated with DMSO (top), dendrimer 5 (middle), or CGS21680 1 (bottom) (each added as a DMSO solution), to which ADP was then added to induce the platelet aggregation. An increase in light transmission indicates increased aggregation. The final concentration of dendrimer 5 or CGS21680 1 in the media was 2 μM, and the total content of DMSO in the media was 0.4% (v/v) in all cases.

Figure 2.

Fluorescence confocal microscopy of human washed platelets incubated with either (A) DMSO or (B–E) dendrimer 5 (Supporting Information). Incubation time: (B) 5 min, (C) 15 min, (D) 30 min, and (E) 60 min. The final concentration of dendrimer 5 in the media was 2 μM, and the total content of DMSO in the media was 0.4% (v/v) in all cases.

In summary, in an effort to develop highly potent agonists acting through multivalent signaling across GPCRs, PAMAM dendrimer conjugates 4 and 5 loaded with multiple copies of CGS21680 were synthesized using a peptide coupling method and purified by SEC. The average MWs and the degree of ligand attachments for dendrimers 4 and 5 were determined by ¹H NMR integration and MALDI. Both analyses support the formation of nearly fully loaded dendrimers. Our preliminary functional assay clearly demonstrated a potent antiaggregatory effect of the PAMAM-CGS21680 dendrimer 5 on the platelet aggregation induced by ADP. In addition, the fluorescent dendrimer 5 was found to be internalized into the platelet as shown by the fluorescent confocal microscopy studies. The approach described here to develop multivalent ligands for GPCRs can be extended to create a dendrimer containing a set of heterogeneous agonists/antagonists, in order to induce combined synergistic biological effects via binding to different receptor subtypes simultaneously on a cellular membrane (51, 52). A detailed investigation on the pharmacological effects of our dendrimers on platelet aggregation will be reported in due course.

Abbreviations:

CGS21680

2-[4-(2-carboxylethyl)phenylethylamino]-5′-N-ethylcarboxamidoadenosine

PyBOP

benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate

Boc

tert-butoxycarbonyl

DIEA

N,N-diisopropylethylamine

Et₃N

triethylamine

DMF

N,N-dimethylformamide

DMSO

dimethyl sulfoxide

DHB

2,5-dihydroxybenzoic acid

THAP

2,4,6-trihydroxyacetophene

ADP

adenosine 5′-diphosphate