An Efficient Microwave-Mediated Synthesis of Hexavalent Sialic Acid Sulfoglycodendrimers as Potential Anti-HIV Agents

Abstract

A series of four sialic acid-containing hexavalent sulfoglycodendrimers (SGDs) were synthesized in excellent yields using an efficient strategy involving multiple microwave-mediated reactions. Four sugars, sialic acid, and the dimer through tetramer of α-2→8-linked oligosialic acid were added to an aminooxy-terminated hexavalent dendrimer core using a chemoselective oxime-forming reaction. This method resulted in substantial improvements in reaction time and product yields over previous methods. These multivalent glycopolymers were designed as potential topical agents for preventing the sexual transmission of HIV-1. While inactive against HIV-1, the SGDs were also not cytotoxic, opening a pathway for the further development of anti-HIV SGDs.

Graphical Abstract

The interest in pursuing macromolecules as potential therapeutic agents is on the rise. In the realm of polymeric materials, dendrimers and glycodendrimers represent an interesting niche of nanomaterials that can be implemented for a variety of biological applications such as drug carriers, anti-viral agents, and gene delivery vehicles, to name a few.^1–4 Dendrimers are multivalent, monomolecular polymers, that are protein-like in their globular shapes, and flexible in terms of design features such as size, number/type of end groups and degree of hydrophilicity/hydrophobicity. Like dendrimers, glycodendrimers have versatile structures and have been incorporated in widespread applications in biology and medicine due to their ability to establish multivalent interactions with proteins/glycoproteins.³ The scaffolding of carbohydrates on the dendrimer cores is responsible for the observed increase in the strength of the interactions from millimolar, which is a typical binding constant (K_d) observed for a monovalent protein-carbohydrate interaction, to micromolar or better when the binding partners are present in >1:1 ratios.

While glycodendrimers have excellent potential as drug candidates, the synthesis of these large molecules is not always simple. This is primarily due to the similar reactivities of the hydroxyl groups, which can lead to longer syntheses because of multiple protection/deprotection steps, often resulting in lower overall product yields. In order to circumnavigate some of these issues, we endeavored to simplify glycodendrimer syntheses using streamlining strategies that include a chemoselective oxime-forming reaction to eliminate the protection/deprotection steps, and the use of a microwave-focused synthetic method to ensure the complete and rapid functionalization of the target glycodendrimers. These glycodendrimers are being pursued as potential thereapeutics to prevent the sexual transmission of HIV-1.

Exceptional hurdles abound in HIV prevention. There are large numbers of individuals unaware of their positive HIV status (15% in the United States, and 21% worldwide)^{5, 6} who may unknowingly spread the disease, primarily through sexual intercourse.⁷ Therefore, until a vaccine can be developed, new preventative strategies must be advanced. One promising approach involves the development of microbicide agents to prevent the sexual transmission of HIV. A microbicide is a chemical that is applied topically to the rectum/vagina to prevent the transmission of sexually transmitted diseases such as HIV.⁸ Currently, there are no FDA-approved anti-HIV microbicides.

To accomplish the goals of developing a potent, non-cytotoxic anti-HIV microbicide, our lab has designed glycodendrimers consisting of a multivalent dendrimer core with pendant sulfated carbohydrates, or sulfoglycodendrimers. These SGDs are designed to bind a polybasic region within the V3 loop, presented on the surface protein gp120, and serve as molecular mimics of the host cell binding partner. The binding partners consist of either the sulfated HIV-1 coreceptors, CCR5 and CXCR4, or heparan sulfate proteoglycans (HSPGs). which can also be hijacked by HIV-1 for the early stages of viral entry.^{2, 9–11} The SGDs described herein were envisioned to have increased potential to interact with the V3 loop, due to the similarities in the compact globular shapes of the SGDs and the gp120 trimers.

Our first series of glycodendrimers utilized a PAMAM (poly(amidoamine)) core in 3 different generations (G=0, 1 and 2), consisting of 4, 8 and 16 ends capped with the unprotected sugar, sialic acid (Sia), or N-acetylneuraminic acid (Neu5Ac). The Sia was linked through carboxyl group in an amide bond, either directly or indirectly to PAMAM, followed by sulfation of the glycodendrimer, yielding the target SGDs.¹² Sia was selected because of a report by Yang et al. demonstrating anti-HIV activity in the sulfated Sia-containing polysaccharide, sulfated colominic acid (SCA).¹³ Colominic acid (CA) is a polydisperse bacterial polysaccharide consisting of α-2→8-linked sialic acid (Figure 1A), that may be chemically modified through a non-selective sulfation reaction. We made these glycodendrimers for two primary reasons: first to find out if multivalent scaffolding of sialic acid onto PAMAM cores resulted in good anti-HIV activity without cytotoxicity, and second, to ascertain which dendrimer core size was most effective in yielding active glycodendrimers. Through this work, we were able to determine that while the G=1 SGD showed weak, qualitative anti-HIV activity, the G=2 SGD, Sulfo-6 (MW=8777 g/mol, 4% sulfur) gave quantitative, low micromolar activity against 4 different pseudovirus strains of HIV-1 (1.6–5.1 μM) in a live cell HIV neutralization assay.^{12, 14} This finding compared well with the activity of the 8000 g/mol linear SCA with 6% sulfur content (0.4 μM) from the Yang study.¹³ Our SGDs had the added advantage of being monomolecular, rather than polydisperse like SCA. This study therefore provided an important proof of concept by showing that smaller sugar fragments, in this case the monosaccharide Sia, when attached multivalently on PAMAM, revealed glycodendrimers with low micromolar anti-HIV activity, equivalent to SCA of similar size. One drawback to this study, however, was the low yields obtained (11–28%) and extremely long reaction times necessary (up to two weeks) to form the fully substituted amide-linked glycodendrimers.¹²

Figure 1.

A. Structures of colominic acid (CA) and sulfated colominic acid (SCA). CA is comprised of Sia with α-2→8 glycosidic linkages. B. Sequence of events showing the formation of the different isomers possible between the ketose, sialic acid, and an aminooxy-containing molecule in aqueous solution.

Moving forward, we sought to build on these results using longer fragments of α-2→8 polysialic acid, namely the dimer (di-Sia), trimer (tri-Sia) and tetramer (tetra-Sia). However, this could not be accomplished using amide coupling chemistry, due to the presence of a carboxyl group on each sugar, which could give rise to multiple regioisomers of the glycodendrimers. To simplify this issue, we investigated a chemoselective oxime-forming reaction (Figure 1B) between a trivalent aminooxy-containing molecule and the ketone-containing sugars, Sia and di-Sia.¹⁵ This methodology would also allow us to continue to incorporate the sialic acid sugars without protecting groups. Oximes are readily formed in aqueous buffers and are both acid-stable and glycosidase-resistant, making them an outstanding choice for the coupling of sugars to dendrimer cores.^16–18 Using the combination of a microwave-mediated reaction with the chemoselective oxime-forming reaction, we determined that the test-case trivalent clusters could be efficiently formed in as little as 30 minutes in excellent yields.¹⁵

Using our microwave-based oxime-forming methodology, herein we report the synthesis, and anti-HIV properties of four new oxime-linked SGDs, utilizing an aminooxy-terminated dendrimer core, in this case, a G=1 hexavalent core, and four different chain lengths of sialic acid-containing sugars, from sialic acid (Sia) through the tetramer (tetra-Sia) of α-2→8-linked oligosialic acid, derived from a microwave mediated acid hydrolysis of colominic acid.¹⁹ To our knowledge, this is the first report of the synthesis of oligosialic acid-conjugated glycodendrimers, which is also notable because of the use of a microwave-mediated reaction for the sugar oxime conjugation step. We wanted to demonstrate that the synthesis of these large glycodendrimers could be achieved in a much shorter total assembly time and in improved yields over our previous amide-linked Sia PAMAM glycodendrimers. Additionally, we hypothesized that we could achieve equivalent or improved anti-HIV activity as with our previously synthesized Sia-PAMAM amide conjugates¹² on a smaller dendrimer core through the clustering of oligomeric sugars on the surface.

The synthesis of the oxime-linked hexavalent glycodendrimers first required the preparation of a hexavalent aminooxy-terminated core (Scheme 1). 1,3-diaminopropane was added to Compound 1 in methanol and was adjusted to pH 8–9 using magnesium hydroxide, yielding the hexa-amino core, Compound 2 in 95% yield. Subsequently, Compound 2 was combined with Compound 3, prepared according to Pochon and coworkers,²⁰ with EDCI/HOBt in DMF and subjected to microwave irradiation for 120 minutes at a power setting of 30% of 400W, providing the Boc-protected aminooxy-terminated dendrimer core, which was subsequently deprotected in trifluoroacetic acid in dichloromethane, giving the desired hexavalent aminooxy-terminated core, Compound 4, in a two-step yield of 55%.

Scheme 1.

Synthesis of a hexavalent aminoxy-terminated dendrimer core, Compound 4.

With the hexavalent aminooxy core Compound 4 in hand, the microwave-mediated oxime-forming reactions were undertaken to yield the desired glycodendrimers, Compounds 9–12 (Scheme 2). Compound 4 was the limiting reactant, and an excess (1.1–2.5 equivalents per reactive site) of the sugar (Compound 5–8) were combined in 150 mM ammonium acetate buffer, pH 4.5, containing 100mM aniline as a catalyst. For the Sia reaction, the coupling was initially attempted in a 42°C oil bath for two days, however, as Compound 4 was still present at the end of 48 hours, the reaction was carried forward to a microwave-mediated reaction for a further 45 minutes at 25% of 400 Watts to yield 94% of Compound 9 after purification. For the oligosialic acids (Compounds 6–8), the coupling reactions were irradiated in the microwave for 120 minutes at 28–30% of 400W, resulting in the di- through tetra-Sia conjugated glycodendrimers in yields of 94, 61 and 97% yields, respectively for Compounds 10–12 after purification.

Scheme 2.

Microwave synthesis of four hexavalent glycodendrimers using sialic acid and the dimer, trimer and tetramer of α-2→8-linked oligosialic acid (Compounds 9–12), and subsequent sulfation, yielding SGDs (Compounds 13–16).

For the confirmation of Compounds 9–12, ¹H and ¹³C NMRs were acquired and assessed for purity. Due to the large size of the glycodendrimers, the peak shape and splitting becomes broadened, and the peak size increases, inflating the integration values.^{21, 22} To determine whether the glycodendrimers were fully substituted, two peaks were selected from the outer region of the glycodendrimer, one for the core and one for the sugar, and the ratio of the integrated values was determined and compared to the theoretical value. For the sugar peak, the N-acetyl peak at ~2.1 ppm was selected, and for the dendrimer core, the ethylene group at ~1.7 ppm was chosen. It should be noted that the H_3a signal from the sugar also falls under the peak at 1.7 ppm, so this was included in the proton count. The N-acetyl peak was normalized to the theoretical value for a fully substituted glycodendrimer. The theoretical ratios for these peaks for Compounds 9–12 were 6:6, 12:8, 18:10 and 24:12, respectively, for 1/3 of the total proton count for the full glycodendrimer (Figure S18, Supporting Information). The experimental values found were 6:6.4, 12:9.4, 18:10.7 and 24:13.5, respectively, in good agreement with the theoretical ratios.

Beyond NMR experiments, we also sought confirmation by high-resolution mass spectrometry (HRMS) using both electrospray mass spectrometry in both the positive and negative modes (ESI⁺ and ESI⁻), matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF) and liquid chromatography-mass spectrometry (LC-MS). Unfortunately, none of these methods were successful. This has been reported previously for compounds similar to these.^{23, 24} Given that we could not obtain MS confirmation of these compounds, we turned to gel filtration chromatography (GFC) for molecular weight determination of the glycodendrimers (See Supporting Information). The experimental values were in good agreement with the theoretical for Mw, the weight average molecular weight, and Mn, the number average molecular weight, with all 4 PDI’s falling below a value of 1.5.

The next step in the preparation of the SGDs involved the sulfation of Compounds 9–12, using a modification of the method reported by Kunou and coworkers.²⁵ Here, Compounds 9–12, were dissolved in dry DMF, to which sulfur trioxide-pyridine was added, and the reaction stirred at 0°C for one hour, after which the reaction solution was neutralized with NaOH, then precipitated in acetone (Scheme 2). Purification of the SGDs (Compounds 13–16) was achieved by a combination of dialysis and size exclusion chromatography. Characterization was accomplished through the determination of the % sulfur (%S) present. Compounds 13–16 yielded %S values of 10, 15, 7 and 7, respectively (Figure 2A).

Figure 2.

A. Summary of the structural details, anti-HIV activity and cytotoxicity data for SGDs 13–16 and positive assay controls, dextran sulfate (DS) and the HIV neutralizing antibody cocktail, CH01–31. B. Thermophoresis response of labeled gp120 (100 nM) in the presence of Compounds 13–16. The change in normalized fluorescence (y-axis) indicates a weak signal with increasing SGD concentration for all four compounds. This suggest a weak or negligible binding interaction with gp120 relative to the positive control, dextran sulfate (DS). For the DS control, a K_d value of 0.52 nM was obtained from this analysis.

The four resultant SGDs were then evaluated for their gp120 binding and subsequent anti-HIV activity utilizing a combination of MST (microscale thermophoresis)^{26, 27} and an HIV neutralization reporter gene assay,¹⁴ respectively, along with a cytotoxicity assessment using a commercially available assay (CellTiter Glo, Promega).²⁸ MST is a relatively new fluorescence-based technique for quantifying molecular interactions in solution and has the advantage of utilizing very small quantities of materials (nanomolar), as compared to the larger volumes and concentrations (millimolar) required for a complementary technique, isothermal titration calorimetry (ITC).²⁶ Consistent with the lack of measurable anti-HIV activity for any of the current SGDs in the HIV neutralization assay (Figure 2A), MST measurements indicated weak or negligible binding of the SGDs to HIV-1 gp120 (Figure 2B). When compared to the positive control, dextran sulfate (K_d=0.52 nM), all four SGDs yielded shallow thermophoresis responses, from which it was not possible to derive meaningful values for the binding affinities. However, despite the lack of binding/inhibition of gp120 or HIV-1, respectively, the SGDs displayed no measurable cytotoxicity in concentrations up to 50 μg/mL against the same cell line (TZM-bl) used in the inhibition of infectivity assay (Figure 2A), as compared to DS, which showed significant cytotoxicity at 25 μg/mL (50 nM).

To attempt to ascertain if there was a structural explanation for why Compound 16 was inactive relative to our previously reported 16-valent active glycodendrimer, Sulfo-6, we investigated the energy minimized structures of both SGDs using MMFF94s, and displayed the structures using an electrostatics density map (See Supporting Information, Figures S26). From the two structures, it can be noted that while both consist of a polyamide dendrimer core with surface decoration by sulfated sialic acids, Compound 16 shows a more open star-shaped configuration, and Sulfo-6, a more closed globular structure. Additionally, Compound 16 has more negative charge density at the larger sulfated sugar termini relative to a smaller dendrimer core, while Sulfo-6 shows a more diffuse and lower level of negative charge density at the smaller sulfated sugar termini relative to the larger dendrimer core. Based on these factors alone, it would be easy to conclude that the shape is the primary determinant of the activity, where the more globular shape of Sulfo-6 is preferred. However, we also theorize that for the star-shaped molecule, Compound 16, having two amide linkages per arm is making the molecule too rigid to contact gp120 in the correct location and conformation. Therefore, we believe that with structural modifications to the dendrimer core to increase the flexibility of the arms, these molecules could still show promise as anti-HIV agents.

In conclusion, we set out to expand our microwave-mediated oxime-forming method to synthesize Sia-containing glycodendrimers efficiently and in excellent yields to serve as potential topical anti-HIV microbicide agents.¹⁵ Our previously reported Sia-conjugated glycodendrimers suffered from long reaction times of up to two weeks in very low yielding amide forming reactions.¹² The amide-linked method was also limited to single sialic acid conjugation, given the linkage ambiguity that would result if oligosialic acids were used. With our new method, we were able to successfully synthesize four oxime-linked glycodendrimers containing the monomer through tetramer of α-2→8-linked Sia in excellent yields of 94, 94, 61 and 97%, respectively, with the longest microwave reaction requiring only 2 hours, as compared to 2 weeks for the amide-linked glycodendrimers. Upon sulfating the sugars to yield the SGDs, we evaluated them for gp120 binding affinity using a relatively new technique, microscale thermophoresis (MST), and for anti-HIV properties through a well-established HIV neutralization assay. Here we desired to use the SGD gp120 binding affinities to be able to predict the HIV neutralization activities, thereby facilitating the development of more potent anti-HIV SGDs. Surprisingly, neither the MST experiments, nor the HIV neutralization assays showed any measurable gp120 binding or anti-HIV activity, respectively, for the four SGDs reported here, even though the largest of the four (Compound 16) was similar in size and % sulfur to our best Sia-PAMAM amide conjugate (Sulfo-6, G=1, hexadecavalent).¹²

To try to better understand the observed differences in biological activity, we performed energy minimization calculations on both compounds and displayed the molecules showing the charge densities. From this, we were able to show that Compound 16 was star-shaped, with a high negative charge density at the sugar termini, while Sulfo-6 bore a more globular structure with a lower negative charge density at the sugar ends. While the globular shape of Sulfo-6 may offer some clear advantages in binding to the surface of the polybasic gp120 surface, we don’t believe it is the only factor influencing activity. We also hypothesize that Compound 16 may be too rigid and this may prevent the SGD from aligning properly with gp120, something that would not be an issue with the more globular Sulfo-6. We are currently revising the dendrimer core to make it more conformationally flexible, and to determine if the anti-HIV activity is restored. We are also further pursuing MST as a screening tool where strong gp120 binding affinities can be equated to the SGD potential in the HIV neutralization assay, thereby streamlining the continued development of more potent anti-HIV SGDs.