Research Area 2. Applying MGE to Treat Disease

This webpage describes the use of metabolic glycoengineering (MGE, a technique introduced on our Research Area 1 page) to study, diagnose, and treat human diseases such as cancer and osteoarthritis. Specifically, Section 2.1 provides and overview of current efforts to apply MGE to treat human disease; Section 2.2 describes the use of high flux “1,3,4” analogs as labeling, diagnostic, and glycoproteomic tools; Section 2.3 describes progress towards developing these compounds as cancer drug candidates; and Section 2.4 describes the potential for exploiting the anti-inflammatory properties of “3,4,6” analogs to treat osteoarthritis.

2.1 Medical Applications of MGE

2.1.1 Introduction and Background. Prospects for developing biomedical applications of MGE are discussed in detail in our review article “Exploiting metabolic glycoengineering to advance healthcare

Figure 2.1.1 Overview of medical applications of MGE (graphical abstract from Agatemor et al, 2019)

2.1.2 Examples of Therapeutic Applications of MGE. To date, biomedical applications of MGE have often focused on cancer; two illustrative examples are given in Figure 2.1.2. In Panel A, tumor infiltrating mesenchymal stem cells (MSCs) are labeled ex vivo using an MGE strategy (Layek et al) and then introduced into the body where they seek out tumors allowing them to be imaged. Panel B illustrates and immunotherapy approach reported by Qiu and coworkers that begins with inoculation of tumor-bearing animals with KLH conjugated with GM3NPHAc, a form of ganglioside GM3 containing the N-phenylacetido-form of sialic acid. Subsequent injection with ManNPhAc leads to selective metabolic incorporation of this analog into cancer cell-displayed GM3, sensitizing the tumor to antibodies generated to the vaccine leading to cancer cell targeting by host immunity via antibody-directed cell cytotoxity (ADCC), complement mediated cytotoxicity (CMC), or antibody-dependent cellular phagocytosis (ADCP).

Figure 2.1.2. MGE-based cancer imaging (Panel A) and immunotherapy (Panel B)

2.1.3. Improving Analog Design to Improve Efficiency and Facilitate Bioactivity. As shown above in Figure 2.1.2(A), most MGE experiments use peracetylated hexosamine analogs, exemplified by Ac4ManNAz. As described in Research Area 1, our group developed “1,3,4” and “3,4,6” tributanoylated analogs that have increased efficiency and unique biological activities compared to their peracetylated counterparts (Figure 2.1.3).

Figure 2.1.3. Structures of peracetylated and “1,3,4” and “3,4,6” tributanoylated ManNAc analogs. The acetate and butyrate groups are removed from the analogs by intracellular esterases, allowing the “core” hexosamine to enter its targeted metabolic pathways; for example, ManNAc analogs intercept the sialic acid biosynthetic pathway. The stereochemical placement of the butyrate groups results in characteristic “1,3,4” vs “3,4,6” biological activity regardless of the composition of the N-acyl group, which can be the natural acetyl group or a non-natural R group such as a ketone, azide, or alkyne.

Our research group is developing applications for tributanoylated hexosamine analogs, examples of these studies are summarized below.

Return to top

2.2 Exploiting “1,3,4” Analogs as Labeling, Glycoproteomic, and Diagnostic Tools

2.2.1 “1,3,4” Analogs Facilitate Efficient, High-flux Glycan Labeling. Peracetylated hexosamine analogs (e.g., Ac4ManNAz, Figure 1.1.3) have been widely used in MGE labeling experiments. Our discovery and development (Aich et al, 2008) of non-cytotoxic, high flux “1,3,4” tributanoylated analogs (Figure 2.1.3) provide a superior alternative for MGE labeling (Figure 2.2.1).

Figure 2.2.1. Comparison of glycoconjugate labeling with Ac4ManNAz and 1,3,4-O-Bu3ManNAz by bioorthogonal glycan blots. (a) Jurkat, (b) SW 1990, (c) MDA-MB-231, (d) CHO, or (e) PANC-1 cells were metabolically labeled with either analog for 48 h and cell surface azido groups were conjugated with biotin-alkyne, lysed, and equal quantities of protein were loaded for each sample, resolved by SDS–PAGE, and visualized by incubation with streptavidin–HRP. In all cases, samples were treated with (from left to right) 12.5, 25, 50, 100, 150 µM Ac4ManNAz or 1,3,4-O-Bu3ManNAz. Lanes indicated with (-) represent control samples from cells treated with neither analog. f: The intensity of the labeled glycoconjugates in each lane was measured using NIH ImageJ software and samples from a particular cell line were compared to the signal from cells incubated with 12.5 µM Ac4ManNAz. Error bars represent the SEM for n ≥ 3 replicates. From Almarz et al, 2012.

2.2.2 “1,3,4” Analogs as Glycoproteomic Tools. Metabolic glycoengineering has been used to label newly-made glycoproteins in living cells and animals; again, Ac4ManNAz is a widely used analog for glycoproteomics experiments where the “tagged” proteins are analyzed by mass spectrometry. Our team has shown that 1,3,4-O-Bu3ManNAz is a superior alternative for these experiments, which are summarized in Figure 2.2.2.

Figure 2.2.2 Strategy for analyzing azide-modified sialoglycoproteins. The strategy used to analyze the samples includes multiple steps as follows: 1) Cells were metabolically labeled using 1,3,4-O-Bu3ManNAz. 2) Proteins were extracted using RIPA buffer at which point samples were divided with one set of aliquots used for steps 4 and 5 and another set of aliquots used for steps 6 through 9. 3) Azide-labeled proteins were biotinylated using through the Staudinger reaction using biotin-PEG3-phosphine and excess reagent was removed by protein precipitation. 4) Biotinlabeled, azide-modified proteins were purified using monomeric avidin agarose. 5) Glycan profiles of biotin-labeled, azide-modified proteins were determined by lectin microarray analysis. 6) Proteins were trypsin digested after biotinylation. 7) Biotin-labeled peptides were coupled to streptavidin agarose. 8) PNGase F was used to release the formerly N-glycosylated peptide from the agarose beads. 9) The released peptides were analyzed by LC-MS. From Tian et al, 2015.

2.2.3 “1,3,4” Analogs Have Diagnostic and Biomarker Potential. The MGE labeling strategy just described was combined with mass spectrometry to analyze sialoglycoproteins in the SW1990 human pancreatic cancer line (Tian et al, 2015). This method identified 75 unique N-glycosite-containing peptides from 55 different metabolically labeled sialoglycoproteins of which 42 were previously linked to cancer in the literature. A comparison of two of these glycoproteins, LAMP1 and ORP150, in histological tumor samples showed overexpression of these proteins in the cancerous tissue (Figure 2.2.3) demonstrating that our approach constitutes a viable strategy to identify and discover sialoglycoproteins associated with cancer, which can serve as biomarkers for cancer diagnosis or targets for therapy.

Figure 2.2.3 Verification of protein expression in pancreatic cancer and matched nontumorous pancreas tissues using IHC. (A) Increased expression of LAMP1 in pancreatic tumor: (i) H&E staining of nontumorous pancreas duct, (ii) low expression of LAMP in matched nontumorous pancreas ductal cells, (iii) H&E staining of pancreatic tumors, and (iv) overexpression of LAMP in pancreatic tumor. (B) Increased expression of ORP150 in pancreatic cancer tissue as compared to matched nontumorous pancreas duct: (i) H&E staining of pancreas tumor and adjacent nontumorous pancreas duct, (ii) IHC of nontumorous pancreas duct versus pancreatic tumor, (iii) a high power view of adjacent nontumorous pancreas duct, and (iv) A high power view of a pancreatic tumor. The blue arrows indicate the nontumorous pancreas ductal cells, and red arrows indicate pancreatic tumor cells. From Tian et al, 2015.

Return to top

2.3 Developing “1,3,4” Analogs as Cancer Drug and Theranostic Candidates

2.3.1 Synergy of 1,3,4-O-Bu3ManNAc and EGFR-targeting TKI Drugs. Higher than normal levels of sialic acid are often associated with cancer suggesting that 1,3,4-O-Bu3ManNAc, which increases flux through the sialic acid biosynthetic pathway, would be oncogenic. Accordingly, we found it surprising that this analog showed inhibitory activity against pancreatic cancer cell lines (Mathew et al, 2016). At first, we attributed this activity to the butyrate groups, which are liberated by non-specific intracellular esterases (Mathew et al, 2012) and can function as HDAC inhibitors (Sampathkumar et al, 2006). For context, when we made this discovery (ca., 2006), the first clinically successful HDAC inhibitor (a hydroxamic acid derivative SAHA, also known as vorinostat [Zolinza®] had just been approved by the FDA to treat cutaneous T-cell lymphoma. Therefore we reasoned that the HDAC inhibitory properties of butyrate generated from 1,3,4-O-Bu3ManNAc had potential anti-cancer potential despite any confounding pro-oncogenic activity resulting from increased sialylation resulting from this analog.

An in-depth exploration of the mechanism of action (MOA) of 1,3,4-O-Bu3ManNAc, however, revealed that this compound had anti-cancer potential because of (instead of despite) its ability to enhance sialylation. Specifically, increased sialylation in analog-treated cells disrupted the galectin lattice, a stabilizing force for oncoproteins such as the epidermal growth factor receptor (EGFR, Figure 2.3.1).

 Figure. 2.3.1 Proposed galectin lattice-mediated mechanism for modulation of EGFR signaling through 1,3,4-O-Bu3ManNAc treatment (from Mathew et al, 2016) The combined modeling and early experimental results were consistent with the depicted biochemical mechanism where destabilization of the galectin lattice occurs due to the masking of galectin-binding epitopes by analog-driven increased sialylation of N-linked glycans. (A) Cancer cells often have highly organized, less-mobile surface receptors (e.g., the green structure represents EGFR while the black structures represent any other cell surface glycoprotein) in part because of a highly formed galectin lattice. (B) One result of a strong galectin lattice is lengthened residence times for EGFR on the cell surface (Lajoie et al, 2007), resulting in enhanced phosphorylation that lead to increased downstream EGFR signaling, which contributes to cancer progression. (C) 1,3,4-O-Bu3ManNAc treatment leads to an increase in sialylation of N-linked glycans bound to EGFR, based on lectin-staining data (specifically SNA binding as shown in the representative FACS plot depicts globally increased expression of α2,6-linked sialic upon 1,3,4-O-Bu3ManNAc treatment, which is consistent with both the data shown in Figure 1 (from Mathew et al, 2016) and our previous results (Almaraz et al, 2012); glycans terminated with α2,6-linked sialic acids mask galactose residues and negatively regulate galectin binding (Zhuo et al, 2011). (D) In turn, reduced galectin binding decreases lattice strength, thereby increasing the surface mobility of EGFR and enhancing its removal from the cell surface (Lajoie et al, 2007). (E) Ultimately – over time periods longer (e.g., 30 to 90 min) than the 2 min time frame investigated in our previous work (1) – this increased rate of internalization predicts faster inactivation of EGFR, which is computationally and experimentally demonstrated by Mathew et al, 2016.

2.3.2 Theranostic Possibilities of “1,3,4” Analogs. The term theranostics is a combination of the terms therapeutics and diagnostics. First generation theranostics were exemplified by radioisotopes that allowed tumor cells to be imaged and, at the same type, delivered toxic doses of radiation to the inherent cancer cells. Related to MGE, 1,3,4-O-Bu3ManNAz allows both the imaging of tumors (the use of “azido” analogs for this purpose is reviewed by Agatemor et al, 2019) and reproduces the synergistic ability of 1,3,4-O-Bu3ManNAc to sensitize pancreatic cancer cells to TKI cancer drugs (Mathew et al, 2015). In another example of theranostic use of MGE, we describe the harnessing of cancer cell metabolism for diagnosing and treating breast cancer (Badr et al, 2017).

Return to top

2.4 Exploring Glucosamine and “3,4,6” Analogs to Treat Arthritis

2.4.1 Background: Glucosamine is a (Controversial) Treatment for Osteoarthritis (OA). Background and Introduction (adapted from Varghese et al, 2007). Osteoarthritis is a degenerative disease characterized by unbalanced synthesis of articular cartilage matrix and associated growth factors. The use of putatively chondro-protective agents such as glucosamine (GlcN) has been explored to treat OA. Glucosamine feeds production of amino-sugars used as building blocks of glycosaminoglycans (GAGs) leading to the hypothesis that dietary supplementation of GlcN stimulates cartilage regeneration, thereby alleviating symptoms of OA. The clinical impact of GlcN remains controversial, however, insofar as poor bioavailability of this compound limits in situ level in synovial fluid to submicromolar levels, which are too low to unambiguously increase synthesis of GAG building blocks (e.g., nucleotide sugars such as UDP-GlcNAc and UDP-GlcA). This led us to investigate alternative biochemical mechanisms to explain the potential therapeutic benefits of GlcN to treat OA, as described next. Our in vivo Results (adapted from Gibson et al, 2014). Optimum responses to GlcN require concentrations impractical with oral dosing. Accordingly we used intra-articular delivery of a bolus dose of GlcN to overcome this obstacle by locally administering GlcN to treat a meniscal transection model of rat osteoarthritis (OA). The knees of male rats underwent medial meniscal transection leading to arthritic changes over 4 weeks. Treatment groups were then given 100 μL injections of 35 μg, 350 μg, 1.8 mg, or 3.5 mg of GlcN dissolved in normal saline thrice weekly. Gross images, modified Mankin scores, and histomorphometric measurements were used as outcome measures (Figure The 350 μg dosage of GlcN had the most significant positive impact on all components of the modified Mankin score. Together, these findings suggest the local delivery of concentrations of GlcN much higher than can be achieved through oral dosing are well tolerated and can suppress experimentally-induced OA through influences on both bone and cartilage. Our in vitro Results (adapted from Varghese et al, 2007). We investigated dose-dependent effects of GlcN on cell morphology, proliferation, cartilage matrix production, and gene expression in primary bovine chondrocytes in monolayer (2D) and hydrogel (3D) in vitro cell culture experiments. Results from the 3D hydrogel experiments indicated a narrow window of GlcN concentrations promotes matrix production (dose-dependent improvement occurred up to 2 mM but was lost by 15 mM where the cells experienced substantial cytotoxicity). The narrow therapeutic window is consistent with previous studies that variously have reported beneficial or detrimental effects of GlcN. Mechanistically, GlcN modulated TGF-β1 signaling in chondrocytes, helping to provide a biochemical explanation for the impact of this sugar at low concentrations where flux into biosynthetic pathways is low.

Figure Histological analysis of arthritic changes and quantitative assessment of cartilage thickness in the rat meniscal transection model of osteoarthritis. Representative micrographs of the medial knee compartment stained with Safranin O. Sham operated knees showed no arthritic changes evident at low or high magnification (A and D, respectively). Osteophytes and chondrophytes are commonly observed in the periarticular region and increase in size and severity from 4 to 8 weeks post-operatively (B and C, respectively). Cellular changes consistent with OA are depicted at higher magnification (E and F). The semi-quantitative modified Mankin scores indicates that OA severity increases with time (G). Consistent with OA progression, increases in calcified cartilage thickness (H) and decreases in un-calcified cartilage thickness were also observed. Statistical significance is indicated as *p < 0.05 compared to sham and **p < 0.05 compared to 4-week time point. From Gibson et al, 2014.

2.4.2 Exploiting GlcN-derived “3,4,6” Analogs to treat OA Hexosamines Provide a Template for Drug Development. As discussed elsewhere (e.g., in Section 1.xx), the regioisomeric placement of three butyrate groups on the four hydroxyls of mammalian hexosamines (i.e., GalNAc, GlcNAc, and ManNAc) tunes bioactivity. Indeed, the core monosaccharide provides a versatile template for drug development (Figure

Figure The hexosamine template – a platform for drug discovery (from Elmouelhi et al, 2009) (A) The three common mammalian hexosamines (e.g., N-acetyl-d-mannosamine, ManNAc; N-acetyl-d-glucosamine GlcNAc, and N-acetyl-d-galactosamine GalNAc) are shown (R1 = CH3 and R2, R3, R4, and R5 = H for the natural sugars). These hexosamines can be derivatized with the ~25 ‘R1’ groups used in metabolic engineering (a sample of these are shown in Panel B with names given based on a ManNAc ‘core’) and the R2, R3, R4, and R5 positions can be derivatized with any of the SCFA shown in Panel C (longer-chain acyl groups render the hybrid molecules insoluble in aqueous medium). Together, this platform can supply tens of thousands of compounds (e.g., [3 hexosamines] × [2 anomers (α/β)] × [25 R1 groups] × [5 R2 groups] × [5 R3 groups] × [5 R4 groups] × [5 R5 groups] = 93,750 different molecular species). The limited subset of these molecules tested by Elmouelhi et al, 2009 is given in Panel D. “3,4,6” Hexosamine Analogs have Anti-inflammatory Properties. We found that, of the large number of possible analogs that can be built using the hexosamine template (Figure, “3,4,6” tributanoylated hexosamines (Figure have anti-inflammatory properties exemplified by their ability to reduce NF-κB signaling.

Figure “3,4,6” Tributanoylated Hexosamine Analogs. Derivation with ester-linked butyrate groups at the C3, C4, and C6-OH groups while leaving the C1-OH group unmodified provides mammalian hexosamines with anti-inflammatory and NF-κB inhibitory properties.

The “3,4,6” analogs shown in Figure were initially considered as anti-cancer drugs due to the pro-oncogenic nature of NF-κB signaling. Overall, however, the prospects of anti-inflammatory agents as cancer drugs are uncertain based on the current paradigm that dampening the immune system contributes to cancer development. Accordingly, we directed this class of analogs to treat osteoarthritis (OA), as described next.

Return to top Cell-based Experiments Show Efficacy of 3,4,6-O-Bu3GlcNAc and 3,4,6-O-Bu3GalNAc (adapted from Coburn et al, 2013a). We investigated the anti-inflammatory and tissue production capacity of 3,4,6-O-Bu3GlcNAc, an shown to inhibit the nuclear factor κB (NF-κB) activity, a key transcription factor regulating inflammation. To mimic an inflammatory environment, chondrocytes were stimulated with interleukin-1β (IL-1β), a potent inflammatory cytokine. 3,4,6-O-Bu3GlcNAc exposure decreased the expression of NF-κB target genes relevant to OA by IL-1β-stimulated chondrocytes after 24 h of exposure. The capacity of 3,4,6-O-Bu3GlcNAc to stimulate extracellular matrix (ECM) accumulation by IL-1β-treated chondrocytes was evaluated in vitro utilizing a three-dimensional hydrogel culturing system. After 21 days, 3,4,6-O-Bu3GlcNAc exposure induced quantifiable increases in both sulfated glycosaminoglycan and total collagen. Histological staining for proteoglycans and type II collagen confirmed these findings. The increased ECM accumulation was not due to the hydrolysis products of the small molecule, n-butyrate and N-acetylglucosamine (GlcNAc), as the isomeric 1,3,4-O-tributanoylated N-acetylglucosamine (1,3,4-O-Bu3GlcNAc) did not elicit a similar response.

Figure Effect of 3,4,6-O-Bu3GlcNAc exposure on the biochemical content of IL-1β-stimulated chondrocytes in three-dimensional (3D) hydrogels after 21 days of exposure. (A) Schematic of a 3D culture system and (B) experimental time line ( (C) DNA normalized to dry weight, sulfated glycosaminoglycan (sGAG) normalized to the DNA content and total collagen normalized to the DNA content of the cell-laden hydrogels after 24 days of culture (n=3; * p<0.05, ** p<0.01 vs. 0 μM 3,4,6-O-Bu3GlcNAc exposure). (D) Histological staining for Safranin O and immunostaining for type II collagen of the cell-laden hydrogels (scale bars: 50 μm). An increase in both stains can be observed at as low as 25 μM 3,4,6-O-Bu3GlcNAc exposure, but is more evident at 50 μM and 100 μM 3,4,6-O-Bu3GlcNAc exposure. From Coburn et al, 2013a.

These findings demonstrate that a novel butanoylated GlcNAc derivative, 3,4,6-O-Bu3GlcNAc, has the potential to stimulate new tissue production and reduce inflammation in IL-1β-induced chondrocytes with utility for OA and other forms of inflammatory arthritis. In a follow-up study we found similar results with 3,4,6-O-Bu3GalNAc with stronger responses compared to 3,4,6-O-Bu3GlcNAc depending on the cell type, e.g., chondrocytes vs. MSCs (Coburn et al, 2013b). In vivo Efficacy of 3,4,6-O-Bu3GalNAc for Treating OA (adapted from Kim et al, 2016a).  Based on the superior effectiveness of 3,4,6-O-Bu3GalNAc in our cell-based assays (Section, which induces cartilage tissue production by human mesenchymal stem cells (hMSCs) and human OA chondrocytes in part by modulating Wnt/β-catenin signaling activity, we selected this analog for pilot animal testing. Specifically, we evaluated therapeutic effect of 3,4,6-O-Bu3GalNAc on the rat model of posttraumatic OA when delivered via local intra-articular sustained-release delivery using microparticles and found this method to be efficacious in preventing OA progression (Figure

Figure Prevention of cartilage degradation in OA induced rats by intra-articular injection of 3,4,6-O-Bu3GalNAc. A. Representative safranin-O stained histological images of the tibial plateau to evaluate the pathological changes 4 weeks after medial meniscal transection (scale bar: 200 μm). The Multi-GalNAc group was given IA injections of 3,4,6-O-Bu3GalNAc on days 7, 14, and 21 after the OA surgery. The Single-GalNAc, PLGA, and PLGA-GalNAc groups were given IA injections on days 7 after the surgery. B. Comparison of the tibial plateau joint based on the OARSI scoring system (The number of animals: 3~6, ** P<0.01, *** P<0.001, ANOVA). C. The size distribution of PLGA-GalNAc and scanning electron microscopic image of fabricated PLGA-GalNAc microspheres (scale bar: 5 μm). D. The cumulative in vitro release profile showed sustained release of 3,4,6-O-Bu3GalNAc from PLGA-GalNAc microspheres for over a month. From Kim et al, 2016a.

These results show that 3,4,6-O-Bu3GalNAc, a disease modifying OA drug candidate, has promising therapeutic potential for articular cartilage repair. We are developing additional delivery options, including controlled release of 3,4,6-O-Bu3GalNAc and 3,4,6-O-Bu3GlcNAc from electrospun microfiber scaffolds (Kim et al, 2016b).