HPLC-assisted automated oligosaccharide synthesis: the implementation of the two-way split valve as a mode of complete automation

Abstract

Reported herein is the development of a user-friendly platform for simple and transformative automation based on standard HPLC equipment. We showcase how the improved platform works in application to the completely automated, a “press of the button,” synthesis of various glycan sequences.

With improved understanding of the functions of glycans, the demand for robust methods to produce both natural glycans and their mimetics has increased.¹ Traditional glycan synthesis in solution involves reiteration of glycosylation–deprotection steps, with aqueous work-up to remove excess reagents after most steps, and requires chromatographic purification of most intermediates. Many advanced strategies that streamline oligosaccharide synthesis are based on either chemoselective or selective activation of building blocks.² Fraser-Reid’s armed-disarmed,³ Danishefsky’s,⁴ Roy’s⁵ and Boons’⁶ active-latent, Ogawa-Ito’s orthogonal,⁷ Ley’s tunable,⁸ Wong’s programmable,⁹ and Huang-Ye’s preactivation¹⁰ concepts are among the most effective strategies known for glycan synthesis in solution. Nevertheless, there is still no universal synthetic route to glycans, and this type of synthesis requires relevant training and qualifications, so it is practically impossible to implement these reactions outside specialized glycosynthetic labs. This significantly hampers the development in glycosciences, whereas other biopolymers, peptides¹¹ and oligonucleotides,¹² can now be produced by machines. Efforts to automate solution phase synthesis of glycans using parallel synthesizers have been reported by Takahashi,¹³ Pohl,^14,15 and Nokami.^16,17 Being still relatively unexplored, these approaches offer viable alternatives to automated enzymatic syntheses being developed by Wong,¹⁸ Nishimura,¹⁹ Chen,²⁰ Wang,²¹ and Boons.²²

Solid-phase synthesis also involves reiteration of glycosylation-deprotection steps, but it eliminates the need for conventional reaction work-up and purification of intermediates.²³ Another strength of the solid phase synthesis is its automation amenability. This was demonstrated in 2001 by Seeberger who adapted a fully automated peptide synthesizer to glycan synthesis.²⁴ In 2012, Seeberger reported “the first fully automated solid-phase oligosaccharide synthesizer”²⁵ that was then commercialized as Glyconeer 2.1.²⁶ Also in 2012, our labs reported HPLC-based automation (HPLC-A) of solid phase synthesis.²⁷ The underpinning idea for this technology is to use the standard lab equipment that will allow recording a successful synthetic sequence as a computer program. To advance our original HPLC-A set-up based on Tentagel,²⁷ we identified JandaJel resin as a more efficient support for HPLC-A.²⁸ We also implemented a standard analytical HPLC autosampler for delivering TMSOTf, the promoter for glycosylation.²⁸ This approach was recently applied to the synthesis of the N-glycan core pentasaccharide.²⁹

Even with the addition of the autosampler to deliver the promoter of glycosylation, the manual component of HPLC-A remained. The building block intake/delivery remained semi-manual because the reagent bottle swap required the operator. Also switching between the reaction and discharge/collection modes required the operator intervention. To enhance our automation capabilities, reported herein is our improved HPLC-A platform that incorporates a preparative autosampler to deliver all building blocks and all reagents for all steps of the synthesis. Also included is a standard two-way split valve, which was programmed to switch between the discharge and the product collection modes, 1 and 2, respectively (Scheme 1).

Scheme 1.

Modified HPLC-A set-up to achieve complete automation.

This new automated set-up completely eliminates all variabilities that could be caused by the operator. No operator intervention is necessary after the software start button, initiating the programmed synthetic sequence, has been pressed and until the final product has been delivered to the collection flask. This approach was showcased by the synthesis of regular 1→6 and 1→4-linked glucans 1 and 2 shown in Fig. 1. Also synthesized was glycan 3 with alternating 1→6 and 1→4-glycosidic linkages. The HPLC-A was programmed to perform the entire sequence of reactions consisting of a series of glycosylations with Fmoc deprotections in between. These steps were followed by the off-resin cleavage, and collection of the final product, all in 12 h in the completely automated “press of a button” mode. To perform the assembly of all oligosaccharide sequences described herein we chose building blocks 4–7 shown in Fig. 1. Glycosyl donor 4 and acceptors 6 and 7 were synthesized as previously reported,^27,28 while trichloroacetimidate donor 5 was derived from known thioglycoside precursor.^25,30

Fig. 1.

Glycan sequences 1–3 obtained in the fully automated mode and building blocks 4–7 needed for all syntheses.

In our approach, the first building block, glycosyl acceptor, is conjugated to the resin on a large scale (5–10 g of the resin) prior to the automation steps. The extent of loading is determined by cleavage/validation sequence on a small portion (50 mg) of the entire load. This approach helps to avoid overestimation of loading and to ensure reproducibility of multiple experiments. Resin functionalized with glycosyl acceptor 6 or 7 (typical scale 50 mg, typical loading 0.02 mmol) is then packed in an Omnifit glass column and the latter is integrated into the HPLC system. Glycosyl donors 4 or 5, promoter (TMSOTf), and other reagents such as piperidine for Fmoc removal are dissolved in separate vials that are placed into the autosampler tray. All subsequent steps are automated by programming the standard operation software without manual intervention (see the ESI^† for complete details).

All glycosylation sequences follow the same basic blueprint. The resin is washed/swelled with the HPLC grade methylene chloride. Then, the flow rate is lowered, and the donor and promoter are delivered as follows. The robotic arm of the autosampler brings the vial containing the donor from the sample tray to the needle that draws the programmed amount of the donor solution. The robotic arm of the autosampler returns the donor vial to the sample tray and then brings the vial containing the promoter solution to the needle. The needle draws the required amount of the promoter, and the donor and promoter are premixed in the “needle seat,” a capillary that connects the needle to the valve of the autosampler. The resulting solution of the pre-activated donor is then injected into the system. After that, the autosampler needle is washed using a dedicated vial in the autosampler tray to avoid cross contamination with vials used in subsequent steps. In our current configuration, recirculation previously performed for each glycosylation, has been substituted by multiple injections. This ensures that the system is supplemented with small portions of fresh preactivated donor until the glycosylation completed. The reaction completion is accessed by the detector, wherein the donor consumption is monitored at 254 nm. The number of injections, and the amount of the donor utilized for every injection can be tuned to ensure that only the desired amount of the donor has been delivered. For instance, 7.5 equiv. of donor 4 were utilized for each glycosylation step for the synthesis of 1→6 linked pentasaccharide 1, and 10 equiv. of the respective donors were used to obtain 1→4 linkages in glycans 2 and 3.

The glycosylations were followed by the Fmoc group removal, and this operation is performed by delivering a solution of piperidine in DMF/CH₂Cl₂. This step is monitored at 254 and 301 nm, with the second wavelength being specific for monitoring dibenzofulvene-piperidine adduct. Once the desired oligosaccharide sequence is achieved via the reiteration of the glycosylation and deprotection steps, the reagent for cleavage, sodium methoxide in methanol-DCM, is delivered to affect the off-resin cleavage. At this stage, the split valve is programmed to direct the flow towards the collection flask (Scheme 1, mode 2).

A representative protocol, general for the synthesis of all target glycans, is exemplified by the synthesis of glycan 1 as depicted in Scheme 2. Compound 8 was obtained with the iterations of glycosylations and Fmoc removal steps. The complete sequence of synthetic steps is built as lists of multiple operations including draw, mix, and inject commands (see the ESI^† for details). These operations are then modulated as needed for the synthesis of a particular target. This approach allows for flexible adaptation to specific targets to maximize the efficiency and yield. Compound 9, resulting from the off-resin cleavage, is then manipulated through a series of post-automation steps. These steps include acetylation with acetic anhydride in pyridine and purification by column chromatography to afford the final pentasaccharide 1 in 33% yield. The result accounts for an average yield of 86% per synthetic step, which compares favorably to single-step conversions observed in previous generations of the HPLC-A synthesizer.²⁸ Additionally to the desired pentasaccharide 1, the deletion sequences were also observed, and their identity was proven by HR-mass spectrometry (see the ESI^† for details). Thus, the corresponding tetrasaccharide and trisaccharide were isolated in 22% and 9% yield, respectively.

Scheme 2.

Automated synthesis of glycan 1.

The final yields for (1→4) linked glycan 2 and tetrasaccharide 3 with alternating (1→6) and (1→4) linkages were 20% and 30%, respectively. A somewhat lower isolated yields are due to the low reactivity of the secondary 4-OH group in glycosyl acceptors. Differently from glycan 1, post-glycosylation manipulations after the off-resin cleavage step involved benzoylation with benzoyl chloride in pyridine. Additionally to the desired pentasaccharide 2, the respective tetrasaccharide deletion sequence was isolated in 6.5%. For tetrasaccharide 3, no deletion sequences were observed, but the monosaccharide derived from the unreacted glycosyl acceptor 6 was isolated in 38% yield.

A new experimental set-up for the HPLC-based automated synthesizer has been developed. The sequences were written to achieve complete automation of all steps, starting from delivering sugar building blocks to obtaining glycan targets in the collection flask. This was made possible by implementing the new two-way switch valve used to direct the flow and by programming the preparative autosampler to deliver all reagents for all steps of the synthesis. The reaction recirculation in this experimental set-up is substituted by multiple and repetitive injections of fresh glycosyl donors and other reagents. The reaction completion is monitored by a standard UV detector. The modular character of HPLC-A allows for varying accessories and can still be used for its intended purpose: purification. Combined use of HPLC-A and chromatography separation with or without mass spectrometry-based analysis can also be a possibility. Machine-assisted synthesis also helps to eliminate variabilities, and to accurately reproduce experiments multiple times. This in turn will simplify the implementation of our protocols in other interested labs, all pre-recorded computer program sequences for all steps of the synthesis are available as a part of the ESI^† or from the authors. However, poor accessibility to sugar building blocks hampers development of all synthetic approaches and the automation platforms suffer the most. With the fact that large excess of glycosyl donors is needed for solid-phase synthesis, researchers experience significant setbacks because they have to continue to remake building blocks.³¹ As Seeberger notes “differentially protected monosaccharide building blocks is currently the bottleneck for chemical synthesis”.³² Things are further complicated because “unlike the synthesis of peptides and oligonucleotides, there are no universal building blocks or methods for the synthesis of all glycans”.³³ In spite of general improvements in application of protecting groups in the mainstream carbohydrate research,³⁴ building blocks for the introduction of uncommon (rare) sugars remain largely underdeveloped.³⁵