Synthesis of Mucin O‐Glycans Associated with Attenuation of Pathogen Virulence

Abstract

With the concerning rise in antibiotic‐resistant infections, novel treatment options against pathogens are urgently sought. Several recent studies have identified mucin O‐glycan mixtures as potent down‐regulators of virulence‐related gene expression in diverse pathogens. As individual mucin glycans cannot be isolated in sufficient purity and quantity for biological evaluation of discrete structures, we have developed an optimized synthetic approach to generate a small library of mucin glycans which were identified as most likely to display activity. The glycans have been prepared in sufficient quantity to assess biological function, studies of which are currently ongoing.

Recent studies have reported that pooled mucin O‐glycans attenuate virulence in diverse pathogens, but the molecular mechanisms have not yet been established. In this work, an optimized synthetic method is described to generate a small library of mucin O‐glycans for biological evaluation of discrete structures.

Introduction

The rise of antibiotic‐resistant infectious pathogens is globally recognized as an imminent threat. Nearly 3 million antibiotic‐resistant infections are recorded annually in the US alone, with approximately 35 thousand resulting in morbidity. ^[1] Therefore, novel approaches to treating infections are urgently sought.

Recently, several studies have reported that treatment with mucin glycoproteins is capable of attenuating virulence in diverse, cross‐kingdom pathogens, including Gram‐negative bacteria (Escherichia coli), Gram‐positive bacteria (Pseudomonas aeruginosa, Streptococcus mutans), and fungi (Candida albicans). ^[2] Mucins impart mucosal secretions with their characteristic viscosity and are the main protein component of mucus. Mucus coats all epithelial surfaces of the body that are exposed to the external environment and is important for lubrication and protection from chemical and physical damage. Mucus hosts our diverse mucosal microbiota, and dysfunctional mucosal barriers are strongly correlated with disease and susceptibility to infection. ^[3] The association of atypical mucus with disease has traditionally been attributed to changes in its physicochemical properties, but these recent reports of mucins attenuating pathogen virulence suggest that discrete biochemical signaling also plays an important role.

O‐Glycans densely coat the mucin backbone and have been recently identified as the active component of mucin in attenuating pathogen virulence. The treatment of pathogens with pooled O‐glycans from different mucin serotypes were shown to disrupt biofilm formation and downregulate the expression of virulence‐associated genes.[ ^2a , ^2c , ^2d , ^2e , ^2f ] Importantly, treatment with mucin O‐glycans increased susceptibility to host immune defense but did not affect overall bacterial survival,[ ^2c , ^2d ] suggesting that mucin O‐glycans may be a promising therapeutic avenue for treating infection with a decreased chance of developing resistance mechanisms.

Mass spectrometry has been used to analyze glycan composition across various mucin samples, identifying several hundred distinct O‐glycan structures. ^[4] Individual glycans contain up to 20 monosaccharide units, and typically consist of a GalNAc residue α‐linked to Ser/Thr residues, with the central GalNAc further elaborated at the 3‐ and/or 6‐positions to afford a diverse array of mucin glycans. ^[5] The diversity of mucin O‐glycan structures likely results from tissue specific expression patterns of different glycosyltransferases, with glycan expression dynamic and altered in response to environmental stimuli and disease.

Although the virulence attenuating activity of crude mucin O‐glycan mixtures has been confirmed, the discrete glycan structures responsible for this novel gene regulation are not well characterized. Individual mucin O‐glycan structures are not commercially available, and given their overlapping physical and chemical properties are not readily isolated as pure compounds from natural sources. Syntheses of certain mucin glycans have been reported, ^[6] but are typically very limited in scalability and incorporate a peptide scaffold not relevant to the current focus. By correlating mass spectrometric data, we could identify the most abundant glycans present in active mucins and focused on developing a novel synthetic approach to streamline the synthesis of a library of mucin O‐glycans of interest. The synthetic platform described within has been optimized to afford individual glycan structures in sufficient quantity for biological evaluation (>20 mg) to enhance our mechanistic understanding of mucin O‐glycan‐regulation of virulence‐related gene expression (Figure 1).

Figure 1.

Recent studies have reported that pooled mucin O‐glycans attenuate virulence in diverse pathogens, but discrete molecular mechanisms are not well established. In this work, we describe an optimized synthetic method to generate a small library of mucin O‐glycans for biological evaluation.

Results and Discussion

Since altered virulence‐related gene expression was observed in diverse mucin samples, ^[2d] we hypothesized that conserved glycan motifs may exist across mucin serotypes. We therefore focused our efforts on glycans that were prevalent across different mucin serotypes and in greatest abundance: core 1, core 1‐Fuc, core 2, core 2‐Fuc, and core 2‐Gal (types I and II) (Figure 2).

Figure 2.

Target library of mucin O‐glycans.

Most prior syntheses of mucin glycans have focused on generating multivalent displays, in efforts to mimic glycan display on the native mucin backbone.[ ^6a , ^6b ] In the aforementioned studies which reported attenuated pathogen virulence, glycans had been cleaved from the mucin backbone, thereby indicating that multivalency is not required for activity. Therefore, we opted to construct methyl α‐glycosides which mimic the stereochemistry of the natural Ser/Thr O‐linkage yet introduce minimal steric bulk. The methyl glycoside linkage is inherently more stable than glycosyl amino acids, which facilitates the construction of larger O‐glycan structures from purely chemical methods.

A synthetic approach using similar core building blocks was desired for streamlining the library assembly. A general strategy was envisaged where the GalNAc 3‐OH position could first be elaborated, followed by a regioselective 6‐O‐glycosylation of GalNAc using the appropriate mono‐ or oligosaccharide donor (Scheme 1). An appropriate orthogonal protecting group strategy would enable the selective removal of a 4,6‐O‐p‐methoxybenzylidene acetal from GalNAc under mild conditions in the presence of other protecting groups (esters, ethers, benzylidene acetal). A simple deprotection strategy could be performed for all glycan targets in few steps: (i) treatment with hydrazine to concomitantly remove the phthalimide and ester protecting groups; (ii) selective N‐acetylation; and (iii) hydrogenation.

Scheme 1.

Example of a retrosynthetic strategy used to obtain mucin O‐glycan 4. Glycans were prepared by first assembling the GalNAc 3‐OH chain, followed by a regioselective glycosylation at GalNAc 6‐OH using the appropriate mono‐ or disaccharide donor.

With the goal of synthesizing a library of glycans, we wanted to incorporate building blocks that could be synthesized in multi‐gram scale using a limited number of steps. Fortunately, key N‐acetyl‐galactosaminyl acceptor 12, present in all target glycans, was accessible from commercially available galactosamine hydrochloride 13 with only a single column chromatography purification after the final step (Scheme 2). This was accomplished by treatment of 13 with Ac₂O in MeOH to afforded an N‐acetylated intermediate. Heating with 2 % v/v HCl in MeOH generated the methyl glycoside and concomitantly hydrolyzed trace O‐acyl by‐products. Although both α‐ and β‐methyl glycosides were initially formed, extended reaction times facilitated anomerization to the desired α‐glycoside. Evaporation and then treatment with anisaldehyde dimethyl acetal catalyzed by the residual HCl afforded 4,6‐O‐acetal‐protected 12 in 70 % yield over 3 steps.

Scheme 2.

Synthesis of mucin O‐glycans 1 and 2. Conditions: (a) NaOMe, MeOH, 0.5 h; then Ac₂O, 0 °C, 2 h; (b) HCl, MeOH, 60 °C, 2 d; (c) pMPCH(OMe)₂, DMF, 16 h, 70 % over 3 steps; (d) AgOTf, tetramethylurea, CH₂Cl₂, 0 °C→rt, 16 h, 58 %; (e) AcOH/H₂O, rt, 4 h, 93 %; (f) NaOMe, MeOH, 3 h, 96 %; (g) NIS, TfOH, CH₂Cl₂/CH₃CN, −40 °C→0 °C, 4 h, 76 %; (h) NH₂NH₂ ⋅ H₂O, EtOH, 80 °C, 24 h; (i) Ac₂O, NaHCO₃, H₂O, 2 h, 77 % over 2 steps.

Next, we turned our attention to finding an appropriate galactosyl donor. Different combinations of per‐O‐acetylated and per‐O‐benzoylated trichloroacetimidates and thioglycosides were evaluated, but in our hands using various activators and reaction conditions, all were hindered by either significant amounts of orthoester formation which complicated purification, or overall low yields. We hypothesized that these low yields may arise from the high acid sensitivity of the p‐methoxybenzylidene acetal protecting group, and therefore turned our attention to milder activation conditions. Fortunately, AgOTf‐activation of 2,3,4,6‐tetra‐O‐acetyl‐galactosyl bromide 14 ^[7] in the presence of tetramethylurea to concomitantly enhance acceptor nucleophilicity and reduce acidity afforded disaccharide 15 in a moderate but acceptable yield (58 %). Selective deprotection of the p‐methoxybenzylidene acetal proceeded cleanly in aqueous AcOH at ambient temperature to afford 16, a key intermediate for synthesis of the target core 2 glycans. Alternatively, global deprotection of 16 under Zemplén conditions afforded the core 1 glycoside (1) in 96 % yield.

By this stage, challenging features of the protected GalNAc‐containing structures had begun to emerge. They displayed a distinctly high polarity, which eventually led us to avoid liquid‐liquid extractions to significantly improve product yields. Further complicating characterization, NMR spectra in CDCl₃ frequently indicated the presence of multiple structures, while measuring NMR at elevated temperatures or in alternative solvents (CD₃OD or CD₃CN) with identical samples often afforded a single structure (Supporting Information). We hypothesized the presence of conformational isomers with restricted rotation, which would also contribute to the poor solubilities observed. This structural rigidity was supported by the significant broadening of NMR signals associated with atoms surrounding the Gal‐GalNAc glycosidic linkage, caused by short transverse relaxation (T ₂) times due to rigid nuclei. ^[8]

To continue with assembly of the series of core 2 glycans, a versatile method to support the incorporation of different GlcNAc or Gal‐GlcNAc building blocks at GalNAc 6‐OH was required. A phthalimide‐protected thioglycoside glucosaminyl donor was selected, which would promote the desired β‐stereoselectivity during glycosylation, and as well the stable thioglycoside would enable assembly of different disaccharide donors (Scheme 2). For the simplest core 2 glycan (2), 3,4,6‐tri‐O‐acetyl‐2‐deoxy‐2‐phthalimido‐β‐d‐glucosamine 7 ^[9] was activated with NIS/TfOH and coupled to disaccharide acceptor 16 to produce 17 in 76 % yield. Global deprotection with NH₂NH₂ ⋅ H₂O at 80 °C, followed by selective N‐acetylation using Ac₂O and NaHCO₃ afforded 2 in 77 % yield over 2 steps.

We then turned our attention to installation of the orthogonally protected 3‐O‐β‐galactosyl linkage required for the fucosylated glycan targets (Scheme 3). Initially, we pursued glycosylation using per‐O‐benzylated 1,2‐anhydro‐galactose, which based on literature precedent ^[10] we predicted would concomitantly generate the desired β‐linkage and a free 2‐OH for subsequent fucosylation within a single step. Unfortunately, in our system, stereocontrol of the epoxide opening proved challenging under various conditions, and always afforded significant amounts of the undesired α‐anomer which led to difficult separations. Therefore, we instead explored an orthogonal protecting group strategy with 4,6‐O‐benzylidene‐protected galactosyl thioglycoside donor 18, ^[11] using selective benzylation at the 3‐OH position. For our particular thioglycoside, both NaH‐ and NiCl₂‐mediated regioselective alkylation ^[12] showed excellent selectivity for the 3‐OH position, but afforded significant amounts of over‐alkylated 2,3‐di‐O‐benzylated by‐product. Iron‐catalyzed alkylation using Fe(dibm)₃ ^[13] was also selective for the 3‐position, but very sluggish to proceed and difficult to scale up. In the end, we opted for conventional tin acetal‐mediated alkylation which proved most reproducible at larger scale to obtain the 3‐O‐benzylated intermediate (→19, 61 %). After benzoylation to obtain 11 (91 %), glycosylation with acceptor 12 and subsequent 2‐O‐benzoyl cleavage provided disaccharide 9 in 75 % yield over 2 steps. Fucosylation at −78 °C using per‐O‐benzylated imidate donor 10 ^[14] afforded selectively the α‐configured product 20 in 69 % yield. Global deprotection via Pd(OH)₂‐catalyzed hydrogenation provided the target core 1‐Fuc glycoside (3) in 81 % yield.

Scheme 3.

Synthesis of mucin O‐glycans 3 and 4. Conditions: (a) n‐Bu₂SnO, toluene, reflux, 16 h; then n‐Bu₄NBr, BnBr, 60 °C, 3.5 h, 61 %; (b) BzCl, Py/CH₂Cl₂, 6 h, 91 %; (c) 12, NIS, TfOH, CH₂Cl₂/CH₃CN, −20 °C, 3 h, 87 %; (d) NaOMe, MeOH/CH₂Cl₂, 60 °C, 48 h, 86 %; (e) TMSOTf, CH₂Cl₂, −78 °C, 6 h, 69 %; (f) H₂, Pd(OH)₂, H₂O/MeOH, 24 h, 81 %; (g) AcOH, H₂O, rt, 3.5 h, 82 %; (h) 7, NIS, TfOH, CH₂Cl₂/CH₃CN, −20 °C, 3 h, 75 %; (i) NH₂NH₂ ⋅ H₂O, EtOH, 80 °C, 24 h; (j) Ac₂O, NaHCO₃, MeOH, 2 h, 79 % (2 steps); (k) H₂, Pd(OH)₂, H₂O/MeOH, 24 h, 82 %.

Core 2‐Fuc (4) was synthesized from glucosamine donor 7 and trisaccharide intermediate 20 (Scheme 3). Selective deprotection of the 4,6‐O‐p‐methoxybenzylidene acetal in 20 using aqueous AcOH was achieved with short reaction times and ambient temperature to afford key intermediate 8 in 82 % yield. Regioselective glycosylation of GalNAc 6‐OH using donor 7 proceeded well (→21, 75 %), which was then subjected to NH₂NH₂ ⋅ H₂O treatment and selective N‐acetylation to afford the partially protected tetrasaccharide 22 in 79 % yield over 2 steps. A final hydrogenation step using Pearlman's catalyst afforded the desired core 2‐Fuc glycoside (4) in 82 % yield.

To prepare the core 2‐Gal glycosides which are further extended at GalNAc 6‐OH, different orthogonally protected GlcNAc building blocks were required. Both type I (β‐1,3) and type II (β‐1,4) Gal‐GlcNAc linkages have been reported in mucin O‐glycans[ ^4b , ¹⁵ ] and therefore were of interest for biological evaluation. In screening different glycosyl donors and activating conditions with acceptors 24 and 26 (Schemes 4 and 5; see Supporting Information for preparation of 24 and 26), orthoester formation was found to be a significant problem, particularly in 4,6‐O‐benylidene‐protected acceptor 24. In the end, per‐O‐acetylated trichloroacetimidate donor 23 ^[16] afforded the greatest yields, to produce 25 and 27 in 82 % and 77 % yields, respectively. Surprisingly, in this system, glycosylation using related per‐O‐benzoylated donors was observed to generate greater amounts of orthoester by‐product than their per‐O‐acetylated counterparts, suggesting that steric effects caused by the bulky 2‐phthalimido group may hinder formation of the desired products.

Scheme 4.

Synthesis of mucin O‐glycan 5. Conditions: (a) TMSOTf, CH₂Cl₂, 0 °C, 2 h, 82 %; (b) 16, NIS, TfOH, CH₂Cl₂/CH₃CN, 0 °C, 5 h, 81 %; (c) NH₂NH₂ ⋅ H₂O, EtOH, 80 °C, 16 h; (d) Ac₂O, NaHCO₃, H₂O, 3 h; then AcOH, 70 °C, 14 h, 62 % over 3 steps.

Scheme 5.

Synthesis of mucin O‐glycan 6. Conditions: (a) TMSOTf, CH₂Cl₂, 0 °C, 6 h, 77 %; (b) 16, NIS, TfOH, CH₂Cl₂/CH₃CN, 0 °C, 6 h, 70 %; (c) NH₂NH₂ ⋅ H₂O, EtOH, 80 °C, 16 h; (d) Ac₂O, NaHCO₃, MeOH, 2 h, 73 % over 2 steps; (e) H₂, Pd(OH)₂, H₂O, 18 h, 88 %.

For assembly of the core 2‐Gal glycosides, coupling of disaccharide donors 25 and 27 with disaccharide acceptor 16 was performed. Activation of 25 with NIS/TfOH afforded the protected tetrasaccharide intermediate in 81 % yield after column chromatography (Scheme 4). Global deprotection using NH₂NH₂ ⋅ H₂O at 80 °C, selective N‐acetylation with Ac₂O and NaHCO₃, and acetal cleavage afforded the desired type I core 2‐Gal tetrasaccharide (5) in 62 % yield over 3 steps. Alternatively, using the same conditions, donor 27 was coupled to disaccharide 16 to afford the protected core II tetrasaccharide in 70 % yield after column chromatography (Scheme 5). Similar tandem NH₂NH₂ ⋅ H₂O treatment–N‐acetylation afforded partially protected tetrasaccharide 28 in 73 % yield over 2 steps, which upon Pd(OH)₂‐catalyzed hydrogenation afforded the target type II core 2‐Gal glycoside (6) in 88 % yield.

Conclusion

Opportunistic pathogens are known to cause disease in unhealthy mucus, but the molecular mechanisms by which healthy mucus protects hosts from pathogens are not well understood. Several studies have recently demonstrated that mixtures of mucin O‐glycans effectively down‐regulate virulence gene expression in diverse pathogens. In efforts to investigate these mechanisms, we have developed an optimized synthetic approach to obtain a targeted library of mucin O‐glycans that were found to be of greatest abundance in virulence‐attenuating mucin samples. The described method affords glycan structures in sufficient quantity for biological evaluation and can enhance our mechanistic understanding of mucin O‐glycan‐regulation of virulence‐related gene expression. Due to increasing concerns over antibiotic‐resistance in chronic infections, novel therapeutics targeting mechanisms of virulence and regulation of biofilm are of immense interest. By elucidating bioactive glycan epitopes, the development of more ‘drug‐like’ and synthetically accessible glycomimetic targets can be explored.

Experimental Section

General methods

All commercial reagents were used as supplied unless stated otherwise. Anhydrous solvents were either commercially acquired or dried using standard techniques and stored over molecular sieves. Thin layer chromatography was performed on silica‐coated glass plates (TLC Silica Gel 60 F₂₅₄, Merck) with detection by fluorescence, charring with 5 % aqueous H₂SO₄, and/or staining with a ceric ammonium molybdate solution. Organic solutions were concentrated and/or evaporated to dry under vacuum in a water bath (<40 °C). Molecular sieves were dried at 400 °C under vacuum for 30 minutes prior to use. Amberlite IR‐120H resin was washed extensively with MeOH and dried under vacuum prior to use. Medium‐pressure liquid chromatography (MPLC) was performed using CombiFlash Companion instruments equipped with either RediSep normal‐phase flash columns or self‐packed reverse‐phase C18 columns, and solvent gradients refer to sloped gradients with concentrations reported as % v/v. NMR spectra were recorded on a Bruker Avance DMX‐500 (500 MHz) spectrometer, and assignments achieved with the assistance of 2D gCOSY, 2D gTOCSY, 2D gHSQC, and 2D gHMBC. Chemical shifts are expressed in ppm and referenced to either Si(CH₃)₄ (for CDCl₃), residual CHD₂OD (for CD₃OD), residual CHD₂CN (for CD₃CN), or a MeOH internal standard (for D₂O). Low resolution electron‐spray ionization mass spectrometry (ESI‐MS) was performed using a Waters Micromass ZQ. High resolution mass spectrometry was performed using an Agilent 1100 LC equipped with a photodiode array detector and a Micromass QTOF I equipped with a 4 GHz digital‐time converter, or alternatively a ThermoFisher Scientific LTQ Orbitrap XL hybrid ion trap mass spectrometer. Optical rotation was determined in a 10 cm cell at 20 °C using a Perkin‐Elmer Model 341 polarimeter. HPLC analysis was performed using an Agilent 1100 LC equipped with an Atlantis T3 (3 μm, 2.1×100 mm) C18 column and detection via an evaporative light scattering detector (ELSD).

Additional information on experimental details, physical data, and ¹H, ¹³C, ¹H‐¹H COSY, and ¹H‐¹³C HSQC NMR spectra of all compounds can be found in the Supporting Information.

Conflict of interest

R.H. is listed as an inventor on a patent application related to the therapeutic use of mucin O‐glycans.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Minzer G., Hevey R., ChemistryOpen 2023, 12, e202200134.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.