The development of a dedicated polymer support for the solid-phase oligosaccharide synthesis

Abstract

Reported herein is the development of a novel polystyrene-based resin that we named PanzaGel. The resin was equipped with diethylene glycol-derived cross-linker with the dedicated application to polymer supported glycan synthesis in mind. After investigating its swelling properties and obtaining encouraging data for its chemical and thermal stability we accessed the amenability of PanzaGel to the HPLC-based platform for the automated synthesis. Comparable glycosylation results to those with traditional supports have been obtained in the synthesis of glycans up to pentasaccharide that was obtained in 30% overall yield. The automated synthesis set-up implemented a common analytical autosampler for delivering all reagents for all steps of the glycan synthesis and cleavage.

With the advances in the field of glycosciences and an increasing number of structures elucidated and applied in all areas of the field, the need for reliable approaches to the synthesis of glycans has grown exponentially.¹ Traditional glycan synthesis in solution involves iteration of glycosylation and deprotection steps with interim purification for practically every intermediate. Some advanced strategies based on either chemoselective or selective activation of building blocks help to streamline the oligosaccharide assembly significantly.² However, no universal route to the chemical synthesis of glycans can be established, which dramatically hinders progress in glycosciences, whereas other biopolymers, peptides^3,4 and oligonucleotides,⁵ can be produced by machines. Solid-phase synthesis eliminates the need for conventional reaction work-up and purification of intermediates,^6–8 and offers promising automation amenability. Since early efforts in 2001,⁹ Seeberger et al. developed a dedicated automated oligosaccharide synthesizer in 2012¹⁰ that was later commercialized as Glyconeer 2.1.¹¹

Also in 2012, we reported High Performance Liquid Chromatography equipment-based Automation (HPLC-A) of solid-phase synthesis.¹² The general idea for developing the HPLC-A is that a computer interface coupled with standard HPLC components will allow recording a successful automated sequence as a computer program that can then be reproduced with the “press of a button.”

However, solid-phase synthesis of glycans suffers from certain inherent weaknesses. “The chemistry and biology of carbohydrates has been a Cinderella field,”¹³ and the area of solid-phase synthesis wherein everything has been “borrowed” from other fields illustrates this problem very clearly. For example, instead of developing dedicated supports, commercial resins designed specifically for peptide or nucleotide solid-phase synthesis^14–16 are meticulously evaluated to determine their possible suitability for glycan synthesis, which often demands different characteristics.¹⁷ While HPLC-A and other automated approaches have a potential to revolutionize the way glycan synthesis is conducted, all current platforms suffer from inherent weakness of solid-phase synthesis as applied to glycans, among which is poor compatibility of existing commercial resins. Reported herein is the first step of our strategic goal to tackle all key weaknesses of the solid-phase synthesis, both automated and manual.

All traditional resins are based on polystyrene, but differ in functionalization/crosslinking, albeit none were designed specifically with the solid-phase synthesis of glycans in mind.¹⁸ Seeberger automation platform relies on the Merrifield resin. Our original HPLC-A set-up was based on Tentagel, a polystyrene grafted with PEG chains.¹² Later on, we identified JandaJel polystyrene resin crosslinked with tetrahydrofuran-derived chains as a more suitable support for HPLC-A¹⁹ due to its better swelling properties compared to the traditional Merrifield resin. Cross-linked polystyrenes are mechanically stable and can be obtained on a large scale from broadly available and inexpensive precursors. However, phenol-based cross-linking used in JandaJel (1, Fig. 1) may potentially have marginal stability under strongly (Lewis) acidic conditions, such as large amounts of trimethylsilyl trifluoromethanesulfonate (TMSOTf)²⁰ commonly employed in the solid-phase synthesis of glycans.

Fig. 1.

Selection of the attachment strategies and cross-linkers.

Although this degradation was not thought to significantly hinder general progress, we came to the realization that beads designed for peptides do not have the necessary properties suitable for glycan solid-phase synthesis. Researchers working with polymer-supported glycan synthesis often encounter limited swelling, marginal chemical stability, resin poisoning, reagent trapping, etc. To address this potential weakness, described herein is the investigation of benzyl alcohol-derived ethylene glycol chain cross-linking (2, Fig. 1) due to the anticipated greater stability of such structures under (strongly) acidic reaction conditions.^21,22 This type of resin was first synthesized by Itsuno et al.,²² but its application in solid-phase synthesis is unknown. Interestingly, benzylic attachments have been considered by Janda and co-workers,^23,24 but were deemed inferior due to their marginal stability toward strongly basic reagents, such as BuLi. To increase the swelling properties in polar protic solvents, we chose to replace the tetrahydrofuran chain used by Janda with diethylene glycol (DEG) derived structure 2 depicted in Fig. 1. Ethylene-glycol,²⁵ its homologs, and other similar molecules²⁴ have been previously investigated as crosslinkers, but the application of such resins in solid-phase synthesis is unknown.²²

Crosslinked monomer 2 was synthesized using paravinylbenzyl chloride 3 and diethylene glycol 4 in the presence of NaH in DMF, affording the desired compound in a good yield of 85% (Scheme 1). The cross-linked monomer 2 was then used in the synthesis of a 2% cross-linked polymer 6. For this purpose, a mixture of monomers 2, 3, and 5 in a respective ratio of 2/10/88 (w/w/w) was polymerized in the presence of benzoyl peroxide, sodium chloride, and Arabic gum in water at 85 °C for 16 h following a procedure previously described by Janda.²⁰ The desired polymer 6 was then purified using a Soxhlet extractor and i-PrOH. The chlorobenzyl group was further functionalized following Gabriel’s synthesis protocol, wherein the chloride was displaced with phthalimide in DMF at 55 °C. The resulting phthaloyl group was then removed with hydrazine hydrate in a mixture of MeOH and chloroform. This two-step procedure produced PanzaGel 7 equipped with the primary amine functional groups (Scheme 1) with a nominal loading capacity of 1.0 mmol g⁻¹.

Scheme 1.

Synthesis of PanzaGel 7.

Then, the swelling properties of the polymer were investigated. Swelling was measured following described procedures,²⁶ using a syringe equipped with a sintered frit. The volume of dry resin was measured, then solvent was introduced in the syringe and the mixture was allowed to equilibrate for 1 h. After being vortexed, extra solvent was removed using another syringe. As listed in Table 1, polymer 7 has much better swelling properties when compared to the Merrifield resin, both in polar and non-polar solvents. In addition, polymer 7 showed similar swelling volumes to those reported for JandaJel.²⁰ Notably, PanzaGel exhibits better swelling in MeOH than that of both Merrifield’s resin and JandaJel.

Table 1.

Comparison of measured swelling properties of PanzaGel to those reported for Merrifield’s resin and JandaJel

Solvent	Volume of resin (mL g−1)
	PanzaGel	Merrifield20	JandaJel20
None	1.75	1.5	1.5
CH2Cl2	9.0	5.8	11.8
MeOH	2.5	1.8	1.5
DMF	8.5	4.2	14
THF	10.0	5.4	12
Toluene	8.0	ND	ND

A set of experiments was then conducted to determine physical properties of the resin. Thermal degradation properties of PanzaGel were assessed by thermogravimetric analysis (TGA). As shown in Fig. 2, the thermal decomposition of PanzaGel 7 was investigated at various temperatures. PanzaGel was found thermally stable up to 360 °C, but about 97% weight loss occurs by 460 °C. To assess the chemical stability of PanzaGel, the polymer was treated with trimethylsilyl trifluoromethanesulfonate, as this Lewis acid was found to degrade JandaJel.²⁰ Upon treatment of 100 mg of resin with 100 μL of TMSOTf in 1.0 mL of methylene chloride for 1 h, which exceeds the typical conditions employed in chemical glycosylations using polymer supports, no decomposition was observed. The thermal decomposition temperature and the SEM images of the polymer showed no significant difference before and after the treatment with TMSOTf. PanzaGel treated with TMSOTf decomposed by about 85% upon heating to 600 °C leaving behind a residue, indicating a chemical structure unaltered by the prolonged contact with the acid.

Fig. 2.

Thermogravimetric analysis (TGA) plots (black line) and weight derivatives curves (red line) of PanzaGel 7.

The prepared polymer was also investigated by scanning electron microscopy. As shown in Fig. 3, polymer 7 was prepared in a shape of monospheres with an average size of 157.64 μm (average of 30 polymer beads). Nitrogen adsorption/desorption isotherm plots have been used to evaluate the pore volume (V_p) and the surface area (S) of the synthesized material. The total pore volume was calculated as 0.0033 mL g⁻¹. The BET surface area was found to be 0.84 m² g⁻¹.

Fig. 3.

SEM images of PanzaGel 7 polymer beads at 100× (a) and 150× magnification (b).

To evaluate synthetic applications of the novel resin, we moved on to synthesize disaccharide 8 and pentasaccharide 9 (Fig. 4). Our labs have recently achieved a full automation of the synthesis of oligosaccharides using a two-way split valve.²⁷ Differently from our previous experimental set up involving reagent delivery with a preparative autosampler,^19,27,28 herein we investigated an HPLC system equipped with an analytical autosampler instead. This module is common on HPLC systems, which would allow to broaden the scope of HPLC-A.

Fig. 4.

Target compounds as showcase of the new resin.

To accommodate this adjustment, a higher concentration of donor 11 (0.12 mmol, 7.2 equiv.) were dissolved in 2.7 mL of methylene chloride and the donor solution was split in two vials, containing 1.6 and 1.1 mL of solution, respectively. The autosampler was programmed to draw 500 μL (0.024 mmol) of donor 11 solution per injection. The autosampler was also programmed to draw 100 μL (0.025 mmol) of TMSOTf solution in methylene chloride per injection. Donor and promoter were mixed in the needle seat and injected towards the HPLC column, loaded with 50 mg of acceptor-bound resin 10 (0.0165 mmol). The activated donor is allowed to saturate the column for 1.0 min at a flow rate of 1.0 mL min⁻¹. After that, the flow rate was lowered to 0.05 mL min⁻¹ to maximize the time of exposure of the resin-bound glycosyl acceptor to the glycosyl donor. The flow rate was kept at the minimum level (0.05 mL min⁻¹) for 10 min, then increased back to 1.0 mL min⁻¹, using fresh CH₂Cl₂ to wash the column from the unreacted and hydrolyzed donor. This washing step is needed to prepare the resin for new injections of donor 11 and TMSOTf solutions. The same operation was repeated four additional times and the glycosylation sequence was completed by alternating washings with methylene chloride and DMF (see the ESI,^† for details).

This completes the synthetic sequence needed for the synthesis of disaccharide 8. For the synthesis of pentasaccharide 9, the temporary Fmoc protecting group was removed using a 40% solution of piperidine in DMF following the glycosylation step. Three injections of the piperidine solution were performed, and the reaction was followed using the variable wavelength detector set at 301 nm. The sequence was completed by washings alternating CH₂Cl₂ and DMF. Last, the polystyrene resin was reacidified through injecting 100 μL of the TMSOTf solution previously used for glycosylation. The two reactions were repeated as needed to afford the target glycan 9. As previously described, the two-way split valve diverts the flow to a collection flask. The compounds are then cleaved using a 1 M solution of sodium methoxide in MeOH/MeOH/CH₂Cl₂ (1.5/1/1, v/v). Through a series a post-automation steps the target glycans 8 and 9 were isolated in 80% and 30% yields, respectively. Noticeably, both syntheses were conducted at room temperature and without utilizing strictly anhydrous reaction conditions or inert atmosphere.

In conclusion, PanzaGel, a new polystyrene-based resin for polymer supported glycan synthesis has been developed. The resin that was designed specifically for glycan synthesis in mind. PanzaGel incorporated best features of common polymeric supports for peptide synthesis: high stability of Merrifield resin and greater swelling JandaJel. PanzaGel has better swelling properties than used Merrifield resin and swells better in polar solvents than JandaJel. The glycan synthesis performed at the end of this study showcased how the improved polymeric support works in application to the HPLC-A. The target oligosaccharides were obtained using the HPLC-A experimental set-up containing the split valve, similar to that recently introduced by us.²⁷ The current application makes use of a standard HPLC analytical autosampler. New programming/injection sequences were developed specifically for this modified set-up. A new solid support for oligosaccharide synthesis will increase the attractiveness of all synthetic and automated technologies, setting the ground for further improvements, which could include grafting to improve swelling in polar solvents. The synthesis of the new resin is fairly straightforward, and the length of the cross-linker and its composition can easily be varied to achieve best results in terms of swelling, mechanical and chemical stability, flow-through applicability, and loading capacity. PanzaGel can be synthesized in many shapes and dimensions: we are currently pursuing the synthesis and application of monolithic resins in a cylinder shape that are more suitable for the in-column, flow-through applications in HPLC-A.