Synthesis and mammalian cell compatibility of light-released glycan precursors for controlled metabolic engineering

Abstract

Sugar additions to biomolecules, or glycans, are some of the most abundant biomolecule modifications in biology because they enable cells to adapt to changing nutrient and stress conditions. An unmet challenge for the field of glycobiology is the study of glycan biosynthetic pathways with chemical control, especially in live cell settings. The objective of this study was to create biocompatible glycan precursors with controlled release properties. Here, we report eleven “caged” sugar probes that release glycan biosynthetic precursor molecules upon light exposure. The specific sugar pathways we target with our probes regulate the addition of the N-acetyl sugars GlcNAc, GalNAc, and sialic acid onto biomolecules in cells, each of which has the potential to alter glycan processes involved in cell morphology, signaling, and behavior. We hypothesized that our glycan precursor probes would remain biologically inert until light-initiated decaging conditions were met, avoiding biological activities including metabolism and cytotoxicity. The photocaged analogs of GlcNAc, GalNAc, and ManNAc (sialic acid precursor) sugars, which we call “photo-sugars,” were released within minutes of light exposure at their optimal wavelengths. During the course of the study, we characterized the cell compatibility of these sugars under their respective decaging conditions, and found highly cell compatible GlcNAc, GalNAc, and ManNAc photocaged precursors. Release of GlcNAc-1-phosphate precursors led to altered ATP levels in cells, demonstrating preliminary metabolic engineering. We envision these probes as useful additions to the chemical glycobiology field that will enable spatiotemporal control over glycosylation pathways in living mammalian cells.

1. Introduction

The ability to chemically activate biological systems through mild, abiological activation methods is one of the key promises of chemical biology.^1–2 Bioorthogonal chemistry, reactions that use motifs that do not inherently react with cellular biomolecules, can be used to manipulate glycans, proteins, and nucleic acids in and on cells in a selective, chemically-regulated manner.² Key methods for chemical activation of biomolecules include reaction-driven decaging of extracellular ligands for receptors³ as well as photochemical methods for internal compound release of signaling agents, drugs, or metabolites.^1,4,5 Among abiotic triggering methods greater challenges remain for activating bioorthogonal agents in live cells because reporter molecules must be cell permeable, non-toxic, and have resistance to intracellular reactivity including redox processes. Photochemistry using non-toxic wavelengths of light is a strategy to “trigger” chemical probe release with a critical benefit of spatiotemporal activation limited to those cells within the area and timeframe of light irradiation.^1,4,6.

Carbohydrates are well-studied substrates for biological engineering, both for intracellular activation with cyclic sugar nucleotides like cyclic adenosine monophosphate⁷ or as signaling sugars like trehalose-6-phosphate.⁸ A key sugar-driven pathway in cells is the glycosylation of biomolecules. In mammalian cells, glycosylation of proteins makes up a major class of post-translational modifications (PTMs) via N- or O-atom linkages to proteins.⁹ Glycolipids and glycoRNAs are also biomolecules modified by carbohydrates.¹⁰ Glycan precursors involved in biomolecule glycosylation include a relatively small number of acetylated sugar monosaccharides, N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), and sialic acids derived from N-acetylmannosamine (ManNAc).⁹ Each of these N-acetyl sugars is not used for metabolism, but rather become activated as nucleotide sugar phosphates for biomolecule glycosylation via a diverse class of glycosyltransferase enzymes (Fig. 1A).¹¹ In cells, the concentration of nucleotide sugar donors like uridine diphosphate-GlcNAc (UDP-GlcNAc) is critical for regulating glycan formation.¹² A key example is the reversible attachment of O-linked GlcNAc (O-GlcNAc) to thousands of proteins in mammalian cells.^13–14 The concentration of UDP-GlcNAc directly regulates O-GlcNAc dynamics.^12,15–18 Additionally, both cellular UDP-GalNAc and sialic acid glycans correlate with hyperglycemic conditions.^19–20

Fig. 1.

Hexosamine biosynthetic pathways leading to donor sugars for glycosylation. A) Full biosynthetic pathways to each of the sugar nucleotides used in mammalian glycan synthesis. Green boxes indicate where photoactivatable sugar analogs can enter these pathways in this report. B) Examples of protein glycosylation motifs, O-linked GlcNAc (O-GlcNAc) or GalNAc-based O-glycosylation. Abbreviations: HK = hexokinase. GPI = glycose phosphate isomerase. GFAT = glutamine fructose-6-phosphate amidotransferase. GNPNAT = glycosamine 6-phosphate N-acetyltransferase. PGM3 = phosphoglucomutase 3. UAP/AGX1 = uridine diphosphate (UDP) N-acetylglucosamine pyrophosphorylase. GALE = UDP-galactose-4-epimerase. GK2 = glycerol kinase 2. MNK = ManNAc kinase. GNE = UDP-GlcNAc 2-epimerase. NANS = sialic acid synthase. NANP = N-acylneuraminate-9-phosphatase. CSS = cytidine 5′-monophosphate-N-acetylneuraminic acid synthetase. OGT = O-GlcNAc transferase.

Aberrant glycosylation patterns of proteins play critical roles in the progression of diseases such as diabetes²¹, cancer,²² rheumatoid arthritis²³, and neurodegenerative disorders¹²,²⁴ because sugar metabolism can be altered in these pathological cell states. However, there is limited knowledge regarding the spatiotemporal dynamics and regulatory mechanisms that are affected by these deviations from the normal glycosylation patterns in these diseases. Specifically, a key challenge for glycobiology studies is the timeframe immediately after glycan precursors are created in live cells.^25–27 To our knowledge, only a single unnatural sugar has been used for light-controlled release in mammalian cells, the azide-modified sialic acid precursor N-acetylmannosamine.^28–29 Here, we report a diverse series of sugars for photochemical release in cell-compatible conditions. We attached photo-caging groups to the anomeric position of sugars as well as synthesized the 1-phosphate-caged and 6-phosphate-caged sugars as light-releasable glycan precursors based on the glycan metabolic pathways in mammalian cells. The release of these sugars can be controlled with light under cell-compatible conditions, enabling a platform to track sugar-driven phenotypes with real-time chemical control.

2. Materials and methods

2.1. General information

Reagents used in chemical synthesis were sourced as follows: N-acetylglucosamine (GlcNAc) (Sigma-Aldrich), acetic anhydride (Ac₂O) (Sigma-Aldrich), pyridine (Acros Organics), trimethylsilyl trifluromethanesulfionate (TMSOTf) (Sigma-Aldrich), camphor sulfonic acid (CSA) (Acros Organics), m-chloro-peroxybenzoic acid (mCPBA) (Alfa Aesar), anhydrous N,N-dimethylformamide (DMF) (EMD Chemicals Inc.), phorphorous trichloride (Sigma-Aldrich), p-toluene sulfonylchloride (pTSCl) (Alfa Aesar), sodium azide (NaN₃) (Fisher Scientific), 5-thioethyl-1H-tetrazole (Sigma-Aldrich), 1-(2-nitrophenyl) ethanol (NPE) (Ambeed, Inc.), 1-(6-nitro-1,3-benzodioxol-5-yl)ethanol (NBDE) (Tokyo Chemical Industry). Dichloromethane (CH₂Cl₂) (Sigma Aldrich) was dried over molecular sieves. Silica gel G-60 F254 aluminium TLC plates were used to monitor reaction progress. Column chromatography was performed on silica gel 60–200 mesh. Deuterated chloroform (CDCl₃), deuterated methanol (CD₃OD) and deuterated dioxide (D₂O) were purchased from Cambridge Isotope Laboratories. ¹H, ¹³C, and ³¹P NMR spectra were collected on a Varian 400-MR or Agilent 600 MHz DD2 instruments, as indicated in the Supplemental Material. High resolution mass spectra were collected on an Orbitrap Exploris 120 (ThermoFisher) in electrospray ionization (ESI)-positive mode. High-performance liquid chromatography was performed on an Agilent 1260 Infinity II equipped with an autosampler and the traces were processed in LCOpenLAb. Plate reads for toxicity assays were measured using a BioTek Cytation One and the data was processed in GraphPad Prism 9.

2.2. Light-driven decaging

Decaging reactions were conducted in a Rayonet RMR-600 photoreactor (Rayonet, Connecticut) equipped with eight 350 nm ultraviolet bulbs (Cat No. RMR-3500-A). Samples were placed 7.5 cm from the light sources and reactions were conducted at 32 total watts, corresponding to approximately 176 μW/cm² intensity. For the 400–410 nm violet decaging reactions, True Violet light emitting diodes (LEDs) (Rapid LED, California) were powered by a 10 W power supply driver equipped with a dimmer switch, with samples placed 7.5 cm from the LED. All decaging experiments were conducted at room temperature in an enclosed box to prevent leakage from other sources of light.

2.3. Tracking decaging by NMR

Compounds 4, 5, 6, 7, or 8 (2 mg) were dissolved in DMSO-d₆ (562 μL) and diluted with D₂O (187.5 μL) to give a 25% v/v solution of D₂O in DMSO-d₆. The samples were analyzed via ³¹P nuclear magnetic resonance (NMR) to establish baseline spectra, then exposed to 350 nm light for the indicated time intervals and analyzed by ³¹P NMR. Exposures and measurements were repeated until the compounds had been fully decaged. The resulting spectra were processed in MestreNova.

2.4. Tracking decaging of compounds by UV–Vis

A solution of compound 4, 5, 6, 7, or 8 (400 μM, 0.1% DMSO) in 1X PBS buffer was sonicated and vortexed until the compound was fully dissolved. The solution was divided between two 35 mm culture dishes (2 mL per dish) for duplicate measurements. Samples were exposed to 350 nm light at the indicated time intervals and the UV spectra was taken. The data was exported from ChemStation (Agilent, California) and plotted in GraphPad Prism.

2.5. Tracking decaging of compounds by high-performance liquid chromatography (HPLC)

Caged sugars were dissolved in phosphate-buffered saline (400 μM, 0.1% DMSO). After the indicated UV exposure, samples were collected and analyzed by HPLC (Agilent Technologies Infinity II, equipped with vial sampler, quat pump, and Agilent 1260 Multiwave detector.) Compounds were separated on a Poroshell 120 C18 column (2.7 μm, 4.6 mm × 50 mm). Concentration and compound identification were performed by integrating the compound peaks at 200 nm, 265 nm (caged phosphate λ_max) and 310 nm (released 2-nitroacetophenone cage λ_max).

2.6. Cell culture conditions

HeLa cells (American Type Cell Culture #CCL-2) were cultured in Dulbecco′s Modified Eagle′s Medium – high glucose (4500 mg/L glucose), l-glutamine, sodium pyruvate, and sodium bicarbonate (Milipore-Sigma D6429) and incubated at 37 °C with 5% CO₂. To seed cells, cells were released from culture dish via trypsin (0.25%, Cytiva SH30031.01). Cell counting was preformed using a Countess™ II system and trypan blue dye following the manufacturer’s procedure (ThermoFisher, Massachusetts).

2.7. Cell viability and toxicity assays

HeLa cells were seeded in 96-well plates (5,000 cells/well) and incubated with the indicated concentrations of each compound or DMSO (1% v/v) for 24 h. After treatment, cell viability was measured using Cell Proliferation Kit 1 (MTT) (Roche, Switzerland) or CellTiter Glo-2.0 (Promega, Wisconsin). Toxicity was determined using CellTox-Green (Promega, Wisconsin). The data was exported from Gen5 software and imported into GraphPad Prism to generate curves.

3. Results and discussion

3.1. Rationale

To enable our strategy of controlled cellular release of glycan precursors, we identified entry points within the biosynthetic pathways of GlcNAc, GalNAc, and sialic acid sugar donor substrates that could be photocaged for controlled chemical release (Fig 1A). We chose two points of photocage attachment: either direct glycosylation of photocleavable groups to the anomeric position of sugars or through the formation of sugar phosphate-based photocages. When activated by light, anomeric photocaged GlcNAc analogs were designed to release free sugars able to undergo biosynthetic transformations to uridine diphosphate-GlcNAc (UDP-GlcNAc) as a glycosyltransferase substrate.³⁰ In the first strategy, the anomeric caged GlcNAc sugars were accessed via glycosylation of photocleavable alcohols. Anomeric photocaged ManNAc analogs are reported to require extended decaging times for biological incorporation,²⁸ so we did not pursue direct caged ManNAc in this work.

Our second strategy created caged sugar-phosphates of GlcNAc, GalNAc, and ManNAc analogs that were designed to bypass the initial biosynthetic steps and release sugar phosphate-containing intermediates in each biosynthetic pathway (Fig. 1A).^30–31 Sugar phosphates were chosen due to downstream metabolic pathway placement, which we hypothesized could lead to greater specificity when used in cells. Additionally, photocaged sugar phosphates enable eukaryotic cell uptake.⁸ During conversion to UDP-GlcNAc, the monosaccharide GlcNAc goes through both 6-phosphate and 1-phosphate intermediates, GlcNAc-1-phosphate (GlcNAc-1P) and GlcNAc-6-phosphate (GlcNAc-6P). Intriguingly, both can be photocaged to allow further control over the biosynthetic steps involved in glycosylation precursor synthesis. For GalNAc, only the 1-phosphate photocage is used by the mammalian biosynthetic pathway to UDP-GalNAc.³⁰ Conversely, during conversion from ManNAc to CMP-Sialic Acid, the anomeric position is not phosphorylated, directing us to design only the 6-phosphate photocage for controlled release of the CMP-sialic acid precursor ManNAc-6-phosphate.³¹

The photocleavable groups were selected based on release with two different wavelengths of light, 350 nm ultraviolet (UV) and 410 nm visible light. Prior work proved successful for caging sugar phosphates with the UV-active nitrophenylethanol (NPE) based photocages.⁸ For higher quantum yield for UV-light decaging, we also synthesized 2-nitrobenzodioxyethanol (NBDE) photocages. We hypothesized that 410 nm decaging would offer the ability to release probes using visible light, potentially avoiding biological effects of UV exposure including DNA dimerization.³² For violet light decaging with 410 nm visible light, diethylaminocoumarin (DEAC) was used for the photocage in this work. Usefully, our modular synthetic routes enable other photocaging groups to be used, including high-quantum yielding (but bulky) nitrodibenzofuran (NBDF) or visible-light-responsive alcohol photocages.³² When exposed to decaging conditions, sugar analogs were envisioned to be released into their respective pathway as sugar alcohols or the corresponding sugar-1- or 6-phosphate. In cells, these probes could potentially release precursors to sugar nucleotides for glycosylation onto proteins (Fig. 1B).³⁰.

We synthesized each type of sugar as the peracetyl species to enhance cell permeability because conversion to glycan precursors and glycoproteins generally occurs inside of living cells.⁹ Acetyl groups are wellreported to be removed by cellular esterases once inside the cytoplasm.^33–34 To extend the utility of this approach, we synthesized probes bearing chemical tags for bioorthogonal labeling via bioorthogonal azide/alkyne cyclization reactions.^35–39 Previous literature reveals that the addition of an azide at the 6-position of GlcNAc is tolerated through the salvage pathway and gets glycosylated onto proteins, though efficiency is lower than the non-azido sugar analogs.³⁵ We incorporated a 6-azide tag on one of the caged GlcNAc analogs to allow tracking of the modifications made by these sugars with bioorthogonal “click chemistry” in subsequent studies. The structures of our chemical probes are shown in Figure 2.

Fig. 2.

Structures of light-released compounds reported in this paper.

3.2. Synthesis of Photo-Sugar probes

The synthetic schemes toward the photocaged sugars (Fig. 2) are collected in Scheme 1. Synthesis of photoactivatable anomeric caged GlcNAc analogs was carried out via activation of GlcNAc to its oxazoline intermediate and subsequent glycosylation with either diethylaminocoumarin (DEAC) or 2-nitrobenzodioxyethanol (NBDE) alcohol to afford Ac3GlcNAc-NBDE 1a, 6Az-Ac2GlcNAc-NBDE 2a, and 6Az-Ac2GlcNAc-DEAC 3a. To access the unprotected sugars, 1a, 2a, and 3a were subsequently deacetylated to give GlcNAc-NBDE 1b, 6Az-GlcNAc-NBDE 2b, and 6Az-GlcNAc-DEAC 3b in 82–90% yields (Scheme 1A).

Scheme 1.

Syntheses of A) anomeric-caged sugars 1–3 and B,C) phosphate-caged sugars 4–8. TMSOTf = trimethylsilyl triflate. DCE = dichloroethane. CSA = camphorsulfonic acid. NaOMe = sodium methoxide·THF = tetrahydrofuran. DCM = dichloromethane. mCPBA = meta-chloroperoxybenzoic acid. TrtCl = trityl chloride. Ac₂O = acetic anhydride. AcOH = acetic acid. Ac = acetate. NBDE = 2-nitrobenzodioxyethanol. DEAC = diethylaminocoumarin. NPE = 2-nitrophenylethanol.

For the caged phosphosugars, we observed low coupling efficiency of DEAC or NBDE to the phosphate possibly due to steric constraints. Switching to the smaller nitrophenylethanol (NPE) photocage enabled us to access both 1-phosphate and 6-phosphate caged compounds in moderate yield. The 1-phosphate light-releasable glycan precursors 1P-diNPE-GlcNAc 4, 6Az-1P-diNPE-GlcNAc 5, and 1P-diNPE-GalNAc 6 were synthesized following related sugar phosphate photocage synthesis conditions⁸ (Scheme 1B). Bis-(1-[2-nitrophenyl]-ethyl)-N,N-diisopro-pylphosphoramidite 9 was coupled to the anomeric alcohol of the desired sugar variants using 5-ethylthio-1H-tetrazole to give the desired compounds in 22–53% yields. This same procedure was used to form the 6-phosphate sugars, 6P-diNPE-GlcNAc 7 and 6P-diNPE-ManNAc 8, in 46% and 81% yields, respectively (Scheme 1C). Altered steric factors in the protection/deprotection of the 6-tritlyl group on mannose resulted in lower yields (8% and 18%, respectively) using conditions optimized for the β-2-OH sugars, though these conditions were not further optimized for mannose. Notably, the racemic chiral center on the NPE led to an equal mixture of 4 diastereomers for each diNPE-caged sugar phosphate (full characterization data found in Supplemental Material). Each compound resolved to a single chiral sugar-phosphate species following light-mediated release of the NPE cages, as determined below.

3.3. Characterization of sugar release

The caged sugars respond to 350 nm light (NPE, NBDE) or 410 nm light (DEAC).³² To track how each compound decaged, we performed nuclear magnetic resonance (NMR) and UV–visible light spectroscopy to characterize the decaging properties of our probes (Fig. 3). The phosphate probes’ decaging progress were selectively followed by ³¹P NMR between UV-light exposure times for the caged GlcNAc-phosphate probes 4, 5, and 7 (Fig. 3A and C). ¹H NMR was measured for nonphosphate molecules (Supp. Fig. 5–6). In brief, compounds were dissolved in NMR solvent and exposed to UV 350 nm UV light for the indicated timepoints. The ³¹P NMR spectrum were recorded postexposure to track decaging. Though the non-azide probes 4, 6, 7, and 8 fully dissolved in D₂O, low solubility of azide-containing 5 prompted the use of 25% (V/V) D₂O in DMSO-d₆ for its full solubility. We monitored the disappearance of the four diastereomeric caged phosphate peaks (blue boxes, Fig 3A and C) and formation of decaged species (red boxes, Fig 3A and 3C). After 10 total minutes of UV exposure the diastereomeric caged phosphate peaks collapsed to a single, upstream peak indicating the released sugar-phosphate species. Our results showed that these compounds were readily decaged in under 10 min in the presence of water. However, NMR has a concentration limitation that require experiments to be done well above biological concentrations, namely 3.7 mM.

Fig. 3.

Decaging of light-releasable glycan precursors characterization by NMR and UV–vis. A) ³¹P NMR shifts of compound 4 with 350 nm light exposure at the indicated time points (solvent: 25% D₂O in DMSO-d₆). Fully caged compound (blue), is mono-deprotected (purple), and rapidly di-deprotected (red) B) UV–vis data showing release of sugar phosphate + free photocage (λ_max = 325 nm) for compound 4 treatment with 350 nm light for the indicated time points (solvent: tissue culture-grade Dulbecco’s Modified Eagles Medium, DMEM). C) ³¹P NMR shifts of compound 7 with 350 nm light exposure at the indicated time points (solvent: 25% D₂O in DMSO-d₆). Fully caged compound (blue), is mono-deprotected (purple), and rapidly di-deprotected (red) D) UV–vis data showing release of sugar phosphate + free photocage for compound 7 treatment with 350 nm light for the indicated time points (solvent: DMEM). Small text on NMR indicates ppm shift values; full-size images and all compound data found in Supp. Figs. 5–11.

As a followup study, we used UV–vis spectroscopy to characterize decaging in fully aqueous conditions that used μM concenrations (vs. mM) that would be more appropriate for cell experiments. We chose both phosphate-buffered saline (PBS) and Dulbecco’s Modified Eagles Medium (DMEM) as common buffer and media types. Because PBS and DMEM each contain high concentrations of phosphates and/or glucose that would interfere with NMR, NMR was not suitable for decaging in these media. By UV–vis, we observed that the intact vs. released photocage provided a spectral shift from λ_max = 260 nm (caged phosphate) to 325 nm (free photocage), enabling UV–vis characterization of decaging (Fig. 3B/D and Supp. Figs. 1–4). For clarity, PBS gave the best UV–vis trace, but DMEM results matched the PBS data following background subtraction (data not shown). Irradiation with 350 nm light revealed that full decaging occurred within 5 min of UV exposure (vs. 10 min by NMR) for compounds 4 and 7. UV–vis de-caging profiles for sugars 6 and 8 followed the same trend (Supp. Figs. 1–4). The accelerated decaging in PBS/DMEM we observed may be due to a pH effect in these buffers, though this hypothesis was not confirmed. As we noted during the NMR experiment, azidosugar 5 did not fully dissolve in PBS or DMEM, so the UV–vis decaging experiment was not possible with 5. Gratifyingly, however, compounds 4, 6, 7, and 8 were each rapidly decaged in fully aqueous solvents suitable for mammalian culture and carried forward for biological evaluation.

To determine more exact conversions of decaging, HPLC traces were performed during decaging of compound 4 (Supplemental Fig. 12). Our results showed that photocaged starting material was ca. 95% depleted following 5 min of UV light exposure.

3.4. Photo-Sugar cell compatibility

Utility of these probes requires cell compatibility for the probe molecules themselves as well as the light exposure used for photodecaging. Cell treatment with sugar probes normally causes an increase in glycosylation pathways,^35,40–42 sometimes including cytotoxicity at high concentrations.⁴³ However, we hypothesized that the caged nature of our sugar probes could prevent bioutilization as well as toxicity, at least until decaging conditions were initiated. To evaluate effects of photocaged sugar analogs, we performed cell viability studies to determine the toxicity effects of these compounds on cells. We chose HeLa cells, a well known human cervical cancer model cell line, to investigate the toxicity effects of photo-releasable glycan precursors 4, 6, 7, and 8 prior to exposure with UV light. HeLa cells were treated with increasing concentrations (12.5, 25, 50, 100, 200, and 400 μM) of 4, 6, 7, and 8 for 24 h (Fig. 4A). The photo-releasable glycan precursors to GlcNAc, GalNAc, and ManNAc (4, 6, 8) demonstrated low cytotoxicity when compared to the DMSO treated control. UDP-GlcNAc precursor 7 showed modest toxicities above 50 μM. These results indicate that our analogs can be incubated with HeLa cells for extended periods of time prior to decaging without disrupting the viability of the system. We note, however, that not all compounds were soluble above 500–800 μM. Therefore, we capped the experiments at 400 μM for the direct comparison. Accordingly, accurate IC₅₀ values for toxicity could not be measured for all compounds because the toxicity was far less than the maximum soluble concentrations for compounds 4, 5, 6, and 8.

Fig. 4.

Evaluating cell viability of caged sugars and decaging conditions in HeLa cells. A) MTT viability assay of compounds 4, 6, 7, and 8 at concentrations of 12.5, 25, 50, 100, 200, and 400 μM for 24 h compared to the DMSO control. B) MTT assay of 350 nm and 410 nm light compared to a non-light-exposed control. C-D) Decaging conditions study: compound 4, 6, 7, or 8 was incubated with cells for 16 h, the media was exchanged, and the cells were exposed to 350 nm light for 5 min. Cell viability was then determined after 1 h (panel C) or 24 h (panel D). E-F) Multiplexed cell viability (black) and toxicity (green) data for independent measures of cellular effects of compound caged 1-phospho-GlcNAc, 4 (panel E) and caged 6-phospho-GlcNAc, 7 (panel F). All assays were performed in triplicate with a representative biological replicate shown of at least two biological replicates per assay. Error bars represent standard error of the mean. Full data for all compounds shown in Supp. Figs. 13–16.

As an additional control, we evaluated the toxicity of 350 nm and 410 nm light on HeLa cells to ensure that decaging of photo-releasable glycan precursors in vitro would not affect the viability of the cells. HeLa cells were irradiated with 350 nm (6 W) or 410 nm (10 W) light for 5–60 min followed by a 24-hour incubation. While 350 nm light demonstrated low toxicity to the cells, we found that 410 nm light was toxic to cells with viability below 30% after just 2 min of irradiation (Fig. 4B). This toxicity is possibly due to generation of reactive oxygen species (ROS) in mammalian cells,⁴⁴ though we did not pursue mechanistic studies. Toxicity of the 410 nm light prevented the use of coumarin-caged sugars 3a and 3b in vitro. However, 350 nm light was found to be minimally cytotoxic as we maintained greater than 80% viability for at least 10 min of exposure. Notably, a 5-minute exposure to 350 nm allowed full decaging of these compounds in mammalian cell culture media (Fig. 3, above), suggesting that these conditions are compatible for cellular studies.

To fully determine the effects of caged glycan probe treatment followed by UV light-release conditions on cells, a multiplexed viability and cell toxicity system was used to study both cell proliferation (viability) and cell toxicity. Cells were treated with the indicated compound 4, 6, 7, or 8 overnight to enable cellular uptake. The media was exchanged to fresh DMEM and the cells were exposed to 5 min of 350 nm light. Viability and toxicity assays were conducted 1 h and 24 h following the release of glycan precursors (Fig. 4C–D, just viability data shown for clarity). Compounds 4, 6, and 8 (caged GlcNAc-1-phosphate, GalNAc-1-phosphate, and ManNAc-6-phosphate) showed little impact on cell viability in the 100–400 μM treatment range. Intriguingly, we observed some decrease in cell viability at the lower concentrations of the caged compounds, which was replicated by an increase in cell toxicity observed by a CellTox-Green assay (Supp. Figs 13–16). Toxicity and loss of viability at low concentrations may be due to inefficient blocking of UV-light effects during the decaging reaction, but this observation bears further mechanistic study. The caged GlcNAc-6-phosphate 7 displayed significant dose-dependent cytotoxicity and loss in cell viability at both time points (Supp. Fig. 14). To confirm these results, we performed multiplexed cell viability and toxicity samples using the CellTiter-Glo and CellTox-Green systems (Fig. 4E–F). The strong toxicity response in Fig. 4F indicated that this compound was toxic to cells. Intriguingly, the regioisomeric compound 4 showed no toxicity at the 24 h timepoint up to 400 μM and a mild decrease in viability at 400 μM, which may represent a modest reduction in signal for proliferation. However, the cell viability assay we used, CellTiter-Glo 2.0, measures cellular ATP levels as its output. We interpreted the CellTiter-Glo drop in signal but lack of CellTox-Green toxicity as an indication that ATP levels were consumed by release of GlcNAc-1-phosphate in these cells but that toxicity did not result. These preliminary findings demonstrated controlled cellular metabolic engineering of ATP consumption following photo-release.

Conversely, compound 7 showed dose-dependent toxicity (green curve) and loss of viability (black curve) at all concentrations tested (data from all compounds found in Supp. Figs. 13–16). We are currently investigating the nature of the unexpected toxicity of caged GlcNAc-6-phosphate, but it was notable that toxicity occurred independently of UV release, suggesting the particular NPE-based structure was itself toxic. Altering the photocaging groups may reduce toxicity in future caged GlcNAc-6-phosphate compound series. Despite the unexpected toxicity of caged GlcNAc-6-phosphate, our results indicate that photocaged GlcNAc-1-phosphate, GalNAc-1-phosphate, and ManNAc-6-phosphate probes will be suitable for cellular decaging experiments.

4. Conclusions

In summary, we synthesized eleven photo-releasable glycan precursors and characterized their release properties and cell compatibility for use in biological studies. These compounds cover three major glycan precursors that are typically used by cells for protein post-translational modifications: UDP-GlcNAc, UDP-GalNAc, and CMP-sialic acid. Phosphate-caged biosynthetic precursors toward these sugar nucleotide donors were light-released in standard tissue culture media within 5 min. Furthermore, the phosphate caged GlcNAc-1-phosphate, GalNAc-1-phosphate, and ManNAc-6-phosphate glycan precursors showed minimal cytotoxicity. Some limitations of this work include poor aqueous solubility of azide-labeled probes, which will necessitate further analog designs for successful photoreleased chemical glycan labeling. We observed significant cell death upon caged GlcNAc-6-phosphate sugar concentrations above 50 μM, a lower threshold than the other probes, and studies to confirm the nature of this unexpected toxicity are underway. However, our data revealed that photo-release of GlcNAc-1-phosphate was nontoxic and resulted in increased ATP consumption in cells, demonstrating preliminary metabolic engineering control. We envision that our strategy, including the probes synthesized here, will be useful as chemical biology tools to characterize effects of glycan biosynthetic pathways with spatiotemporal control. Applications of these “photo-sugars” for glycan engineering experiments to control glycoprotein phenotypes in cells are ongoing in our laboratory.

Abbreviations:

CMP-Sialic Acid

cytidine monophosphate sialic acid

DEAC

diethylaminocoumarin

DMEM

Dulbecco’s Modified Eagles Medium

DMSO

dimethylsulfoxide

GalNAc

N-acetylgalactosamine

GlcNAc

N-acetylglucosamine

GlcNAc-1P

GlcNAc-1-phosphate

GlcNAc-6P

GlcNAc-6-phosphate

LED

light emitting diode

ManNAc

N-acetylmannosamine

NBDE

2-nitrobenzodioxyethanol

NBDF

nitrodibenzofuran

NMR

nuclear magnetic resonance

NPE

nitrophenylethanol

O-GlcNAc

O-linked GlcNAc

PBS

phosphate buffered saline

PTM

post-translational modification

ROS

reactive oxygen species

UDP-GalNAc

uridine diphosphate GalNAc

UDP-GlcNAc

uridine diphosphate GlcNAc

UV

ultraviolet