Sweet Control: MicroRNA Regulation of the Glycome

Abstract

Glycosylation is a sophisticated informational system that controls specific biological functions at both the cellular and organismal level. Dysregulation of glycosylation may underlie some of the most complex and common diseases of the modern era. In the past 5 years, microRNA have come to the forefront as a critical regulator of the glycome. Herein we review the current literature on miRNA regulation of glycosylation and how this work may point to a new way to identify the biological importance of glycosylation enzymes.

INTRODUCTION

Glycosylation is increasingly being revealed as a sophisticated informational system that underlies important biological functions at both the cellular and organismal level¹. Glycans, aka carbohydrates or oligosaccharides, are the products of multiple glycosyltransferases and glycosidases working in a coordinated manner to synthesize structures appended to proteins and/or lipids. The importance of glycosylation is perhaps most apparent from the ever increasing number of genetic disorders and genome-wide association studies (GWAS) that point to glycosylation enzymes as causative agents of disease. In recent work, Joshi et al found that glycosylation enzymes implicated in complex diseases by GWAS are highly regulated, arguing that precise control over specific glycans is necessary². However, how nature keeps these low abundance transcripts tightly controlled is unclear.

In the past 5 years, microRNAs have come to the forefront as critical regulators of the glycome^3–5. microRNAs (miRNAs, miRs) are small, non-coding RNAs that bind to messenger RNAs (mRNAs) and regulate mRNA translation into proteins. miRs exit the nucleus as hairpin pre-miRs, which are then cut into 2 mature miRs, the 5p miR (which comes from the 5’ end, miR-5p) and the 3p miR (which is derived from the 3’ end, miR-3p, Figure 1). These miRs can have distinct mRNA targets within the cell. A mature miR is then loaded into a RISC complex where it binds the mRNA. The binding is often, but not exclusively, to the 3’-untranslated region (3’-UTR) and leads to translational repression. In many cases, mRNA degradation also occurs. A single miR can have hundreds of targets. If a glycosylation enzyme or other glycan related protein (e.g. transporters, metabolic enzymes, etc.; with glycosylation enzymes, collectively known as glycogenes) is regulated by a miR, a loss of the concomitant sugar epitope is then observed.

Figure 1.

miRs are loaded into RISC complexes and inhibit protein expression through translational repression or mRNA degradation. This impacts glycosylation through repression of glycogenes such as FUT4. Lowered expression of the biosynthetic enzyme would shift the expression of the corresponding glycan epitope, as shown above.

Unlike transcriptional regulators, the role of miRs is not to turn a gene on or off but rather to tune protein expression. Thus, miRs provide a mechanism to maintain tight control of protein levels within a specific window⁶. This control is dependent on the precise mRNA transcript. Many genes have several transcripts that differ only in their 3’-UTRs and lead to identical proteins. Because networks of miRs act in concert, the biosynthesis of a glycan epitope may be regulated by multiple miRs acting to tune the expression of several glycogenes simultaneously (Figure 2). The predicted glycogene targets of miRs are unevenly distributed. An analysis of miR:mRNA interactions predicted by the miRANDA algorithm identified some glycogene transcripts as highly-regulated, with multiple miR target sites, while others had few predicted sites⁵. Several glycogenes known to be involved in complex diseases (e.g. FUT8^7–9, GALNT7^10–19, and GALNT1^20–22) were predicted to have highly-regulated transcripts, implying that miRs may play a direct role in dysregulation of these enzymes⁵. Enzymes previously believed to be “functionally redundant” such as the 20-member GALNT family, show large differences in potential miR regulation, implying that they control different biology. This argument against “redundancy” from the regulatory perspective, is borne out by new work showing that the GALNTs do indeed have distinct functions ^{23, 24}.

Figure 2.

miRs can regulate multiple glycogenes in a network, modulating glycan structures.

MICRORNAS ARE CRITICAL REGULATORS OF THE GLYCOME

A study in Caenorhabditis elegans (C. elegans) by Han and coworkers in 2009 was one of the earliest to show glycosylation as a major target of miRs²⁵. In this work, they characterized the miRs found in RISC complexes during worm development. They found that miR targets were enriched in signaling proteins, while housekeeping genes were under-represented. In addition, gene transcripts involved in glycosylation pathways were highly enriched in the pool of strong miR targets. They demonstrated that appending the 3’-UTR of one of the enriched glycogenes, sqv-3, was enough to repress expression of GFP in larval stages where this transcript was observed in the RISC complexes. To our knowledge, this work was the first example of glycogene regulation by a miR and supported a major role for miRs in the regulation of glycan biosynthesis.

In 2014, work from our laboratory directly demonstrated a critical role for miRs in the regulation of glycosylation in human cells⁴. Using bioinformatic methods we integrated miR profiles of the NCI-60, a 59 human cancer cell line panel, with glycomic analysis obtained using our lectin microarray approach^{26, 27}. We identified multiple miRs that correlated with specific glycosylation patterns. These miRs directly targeted the transcripts of glycogenes underlying the observed glycans and were able to alter the glycosylation of cells. Our work underscored the important role of miRs in controlling the glycome. At the time of this publication in 2014, only 10 glycogenes were known targets of miRs. In the past 5 years, there has been an explosion of interest in miR regulation of glycogenes and over 80 glycogenes are currently known miR targets (Table 1). Herein, we discuss select examples of glycome regulation by miRs and the potential of miRs to identify the biological functions of specific glycosylation motifs.

Table 1.

List of known miR regulators for human glycogenes organized by pathway. The HUGO Genome Nomenclature Committee (HGNC) symbol is given for each gene along with the nomenclature used in the accompanying literature cited. For the miRs, Designations for -5p and- 3p are noted where specified in reference.

Pathway	Gene Symbol (HGNC)	Alternative symbols used in literature	miRNAs
O-GlcNAc	OGT	O-GLCNAC, HRNT1, MGC22921, FLJ23071, OGT1	hsa-miR-485-5p36, 39, hsa-miR-10135, hsa-miR-48337, hsa-miR-200a/200b-3p38, hsa-miR-24-140, hsa-miR-42442, hsa-miR-423-5p44, hsa-miR-746
	OGA	MGEA5, MEA5, NCOAT	hsa-miR-53947
N-linked pathway
Glycosyltransferases	RPN2	SWP1, RPNII, RIBIIR, RPN-II	hsa-miR-12876, hsa-miR-37877
	ALG3	NOT56L, Not56, CDGS4, D16Ertd36e	hsa-miR-34278
	ALG12	ECM39, CDG1G	hsa-miR-147a79
	ALG13	GLT28D1, CXorf45, CDG1S	hsa-miR-34a80
	FUT8		hsa-miR-1229, hsa-miR-34a9, hsa-miR-26a53, has-miR-26b53, hsa-miR-146a8, hsa-miR-198 69
	MGAT3	GNT-III	hsa-miR-23a 81
	MGAT4A	GnT-Iva, GnT-4a	hsa-miR-42442, hsa-let-7c50
Glycosidases	EDEM1	KIAA0212, EDEM	hsa-miR-211 82, hsa-miR-581 83, 84, hsa-miR-204 83, 84
	MAN1A2	MAN1B	hsa-miR-30c, hsa-miR-3614
	MAN1B1		hsa-miR-125b85, 86
	MANEA	FLJ12838	hsa-miR-120287
O-linked pathway
Initiation	GALNT1	GalNAc-T1	hsa-miR-216b21, hsa-miR-30b/30d10, hsa-miR-10a88, hsa-miR-12920
	GALNT2	GalNAc-T2	hsa-let-7b89
	GALNT3	GalNAc-T3, HHS, HFTC	hsa-miR-26a90, hsa-miR-17-3p and hsa-miR-22191
	GALNT4	GalNAc-T4	hsa-miR-426292, hsa-miR-993, hsa-miR-36594
	GALNT5	GalNAc-T5	hsa-miR-196b-5p95
	GALNT7	GalNAc-T7	hsa-miR-15418, hsa-miR-21411, 17, hsa-miR-30a-5p16, hsa-miR-49414, 15, hsa-miR-34a/c13, hsa-miR-17-3p/5p12, hsa-miR-21411, 17,hsa-miR-30b/30d10, hsa-miR-378 10–19
	GALNT10	GalNAc-T10	hsa-miR-12296
	GALNT13	GalNAc-T13, KIAA1918	hsa-miR-42442
	GALNT14	GalNAc-T14, FLJ12691	hsa-miR-125a97
	TMTC2	DKFZp762A217	hsa-miR-14298
	POGLUT1	KDELCL1, MDS010, MDS010, MGC32995, 9630046K23Rik, MDSRP, hCLP46, Rumi	hsa-miR-13499, hsa-miR-14299, 100
Elongation and Branching	B3GAT3	GlcAT-I	hsa-miR-23b101
	B3GLCT	B3GALTL	hsa-miR-200b, hsa-miR-200c, hsa-miR-4293
	B3GNT5	B3GN-T5, beta3Gn-T5	hsa-miR-203102
	C1GALT1	C1GALT, T-synthase	hsa-miR-148b103
	C1GALT1C1	COSMC, C1GALT2	hsa-miR-320104, hsa-miR-155105, hsa-miR-374b106
	GCNT2	NACGT1, II, GCNT5, CCAT, IGNT, NAGCT1, bA421M1.1, bA360O19.2, ULG3	hsa-miR-199a/b-5p 107
	GCNT3	C2GnT-M, C2/4GnT, C2GnT2	hsa-miR-302b-3p 108, hsa-miR-15b 109
	LFNG	SCDO3	hsa-miR-200f 110, hsa-miR-125a-5p 111, 112, hsa-miR-146a 113
Capping
PolyLacNAc	B3GALT5	beta3Gal-T5, B3GalT-V, GLCT5, B3T5	hsa-miR-203114
	B4GALT1	GGTB2	hsa-miR-124-3p115
Sialylation	ST3GAL3	ST3Gal III, SIAT6, MRT12	hsa-miR-200a116
	ST3GAL4	STZ, SAT3, FLJ11867, CGS23, SIAT4, NANTA3, SIAT4C	hsa-miR-200a116, hsa-miR-370113
	ST3GAL5	SIAT9, ST3GalV, SIATGM3S	hsa-miR-26a117, hsa-miR-548l117, hsa-miR-34a117, hsa-miR-200b3, hsa-miR-200c3, hsa-miR-4293
	ST3GAL6	SIAT10, ST3GALVI	hsa-miR-26a118, 119
	ST6GAL1	SIAT1, ST6Gal I	hsa-miR-9120
	ST6GALNAC1	SIAT7A, ST6GalNAcI	hsa-miR-30d-5p121
	ST6GALNAC2	SIAT7, SIAT7B, SIATL1	hsa-miR-182122, 123, hsa-miR-135b122, 123
	ST6GALNAC4	SIAT7D, ST6GALNACIV, SIAT3C	hsa-miR-4299124
	ST6GALNAC5	SIAT7E, MGC3184, ST6GalNAcV	hsa-miR-200b, hsa-miR-200c, hsa-miR-4293
	ST8SIA1	SIAT8, SIAT8A	hsa-miR-33a,hsa- let-7e125
	ST8SIA2	SIAT8B, STX, ST8SIA-II, HsT19690	hsa-miR-3099126
	ST8SIA4	SIAT8D, ST8Sia IV	hsa-miR-26a/26b127, hsa-miR-146a/146b128, hsa-miR-181c128
Fucosylation	FUT1	H, HSC	hsa-miR-140-5p129, hsa-miR-149 129, hsa-miR-34a 130
	FUT2	SE, sej, Se2, SEC2	hsa-miR-15b 131
	FUT4	CD15, FUC-TIV, FCT3A, ELFT	hsa-miR-125a-5p 129, hsa-miR-26a/26b 53, 56, hsa-miR-200c 55, hsa-miR-200b 55, hsa-miR-493-5p 52, hsa-miR-224-3p 51
	FUT5	FUC-TV	hsa-miR-125a-3p 132
	FUT6	FT1A, FCT3A, FucT-VI, FLJ40754	hsa-miR-326 133, hsa-miR-125a-3p 132, hsa-miR-106b 133
	FUT8*	See above in N-linked pathway
GAG related enzymes
Chondroitin Sulfate Synthetases	CHSY1	KIAA0990, CSS1	has-miR-194, hsa-miR-515134
	CHPF	CSS2, CHSY2	has-miR-194, hsa-miR-515134
	CHSY3	CSS3, CHSY-2	has-miR-194, hsa-miR-515134
Glucuronyl acid epimerase	GLCE	KIAA0836, HSEPI	hsa-miR-218135
Sulfotransferases/sulfatases	CHST3	C6ST, C6ST1	hsa-miR-513a-5p 136
	HS3ST2	3OST2	hsa-miR-100 137
	HS6ST2		hsa-miR-141-3p, hsa-miR-145-5p 138
	NDST1	HSST, NST1	hsa-miR-149139, hsa-miR-24139, hsa-miR-191139
	SULF1	KIAA1077, SULF-1, hSulf-1	hsa-miR-21140
Hyaluronan synthetases	HAS1	HAS	hsa-miR-125a141, hsa-miR-214141
	HAS2		hsa-miR-410 (up-regulating) 142, hsa-miR-7143, hsa-miR-26b144, hsa-miR-378144, hsa-miR-23a-3p 145, hsa-miR-424/424*146, hsa-miR-23147, hsa-miR-574147, hsa-miR-101-3p148
	HAS3		hsa-miR-26a-5p149, hsa-miR-29a-3p150
Others
Glycosidases	FUCA2	MGC1314, dJ20N2.5	hsa-miR-145151, hsa-miR-200b, hsa-miR-200c, hsa-miR-429 4
	GALC		hsa-miR-140-5p152
	GBA	GLUC, GBA1	hsa-miR-22-3p153
	NEU1	NEU	hsa-miR-125b154
	HEXB		hsa-miR-207, hsa-miR-352 155
Nucleotide Sugar Metabolism	PMM2	CDG1, CDGS, CDG1a, PMI, PMI1	hsa-miR-451a156, 157
	TSTA3	FX, P35B, SDR4E1	hsa-miR-125a-5p, hsa-miR-125b158
	CMAHP	CMAH	hsa-miR-155-5p, hsa-miR-425-5p, hsa-miR-15a-5p, hsa-miR-503-5p, hsa-miR-16-5p, hsa-miR-29a-3p, and hsa-miR-29b-3p159
	UAP1	SPAG2, AGX1, AgX	hsa-miR-224-5p 160
Nucleotide sugar transporters	SLC35B2	PAPST1, UGTrel4	hsa-miR-22161
	SLC35B4	FLJ14697, YEA4	hsa-miR-1764, hsa-miR-1700162
	SLC35F5	FLJ22004	hsa-miR-369-3p 163
UDP-Glucuronyltransferases (involved in Drug Metabolism)	UGT2B15	UGT2B8	hsa-miR-331-5p164, 165, hsa-miR-376c 166, hsa-miR-770-5p 165, hsa-miR-103b 165, hsa-miR-3924 165, hsa-miR-376b-3p 165, hsa-miR-455-5p, 165 hsa-miR-605 165, hsa-miR-624-3p 165, hsa-miR-4712-5p 165, hsa-miR-3675-3p 165, hsa-miR-6500-5p 165, hsa-miR-548as-3p 165, hsa-miR-4292 165
	UGT2B17		hsa-miR-376c 166
	UGT2B7	UGT2B9	hsa-miR-1293, hsa-miR-3664-3p, hsa-miR-4317, hsa-miR-513c-3p, hsa-miR-4483, and hsa-miR-142-3p 164, 167
Other Glycosylation Related Proteins	COG6	COD2, KIAA1134	hsa-miR-1 168
	KL (Klotho)		hsa-miR-34a169, hsa-miR-199b-5p170, 171, hsa-miR-504172, hsa-miR-339172, hsa-miR-556173
	SPOCK1	TIC1, SPOCK, testican-1	hsa-miR-150-3p/5p 174, 175, hsa-miR-129-5p 174, hsa-miR-585 175
	SPOCK3	testican-3	hsa-miR-145176

MICRORNA REGULATION OF O-GLCNAC

Unlike canonical O-glycosylation, O-GlcNAc is a dynamic glycan modification found on cytoplasmic, mitochondrial and nuclear proteins²⁸. This modification is controlled by two enzymes, O-GlcNAc transferase (OGT) and the glycosidase, OGA. O-GlcNAc modifies a diverse set of proteins including histones, transcription factors and signaling proteins and is involved in many biological processes from cell cycle to Alzheimer’s^29–31 to obesity^{32, 33}. OGT expression is tightly controlled at both protein and mRNA levels³⁴ and is predicted to be highly-regulated by miRs⁵. There are 11 known mRNA transcripts for OGT, and 3 protein isoforms. To date, work has focused on the most abundant OGT transcript (ID ENST00000373719.8) an mRNA of the nucleocytoplamic proteoform. In recent years, 10 miR:mRNA interactions have been identified for this transcript of OGT, regulating the enzyme in a variety of diseases from cancer to cardiovascular disease ^35–46. Several of the miRs identified to hit OGT directly impact cell proliferation and are involved in cancer progression. Examples include miR-483 in gastric cancer and miR-485-5p in esophageal and colorectal cancers^{36, 37}. Other identified functions of miRs targeting OGT include reducing tumor angiogenesis (miR-7⁴⁶), inhibiting cell invasion (miR-24⁴¹), and modulating glucose-induced inflammation (miR-200a/b³⁸).

In contrast to OGT, OGA is not predicted to be a highly regulated gene. To date, only one miR:mRNA interaction has been identified for this glycogene. miR-539 is up-regulated in the failing heart, and targets OGA, increasing O-GlcNAcylation during heart failure⁴⁷. O-GlcNAcylation is highly dynamic in the heart, as are the transcript levels of both OGT and OGA. miR-24, which is involved in cardiovascular function has also been shown to modulate OGT⁴⁸, although not in the context of cardiovascular disease. Currently we still have a limited understanding of the regulation of OGT and OGA, but it is increasingly clear that miRs may play a strong role in the dynamic expression of these enzymes and the resulting O-GlcNAcylation levels. Given the critical importance of O-GlcNAcylation to a wide variety of diseases, miR regulation of these enzymes warrants a more thorough examination.

MICRORNA REGULATION OF N-LINKED GLYCOSYLATION PATHWAY

The N-linked biosynthetic pathway begins with the biosynthesis of Glc₃Man₉GlcNAc₂-dolichol primarily by the ALG genes. Bioinformatic analysis of these genes show 2, ALG13 and ALG9, are predicted to be highly-regulated. ALG13, a known epilepsy-related gene⁴⁹, is targeted by miR-34a in neuroblastoma cells. Currently there are no known miR targets for ALG9. The Glc₃Man₉GlcNAc₂ moiety is subsequently transferred onto asparagines in an Asn-X-Ser/Thr motif, where X is any amino acid except proline (Pro), the glucoses are removed by glucosidases as part of the protein quality control system to form the Man₉GlcNAc₂ structure (Man9). Mannosidases then trim the subsequent high mannose structures culminating in Man₅GlcNAc₂ (Man5). The mannosidases MANEA, MAN1A1 and MAN1A2 are highly regulated by miRs based on prediction⁵. Of these genes, miR regulation has been found for MANEA and MAN1A2. Interestingly, the high mannose epitope, levels of which are controlled by these genes, is emerging as a signaling molecule in infection and inflammation⁴⁹.

Man5 is further elaborated and trimmed by a variety of enzymes to make hybrid and complex N-glycans. Only select enzymes within this pathway are predicted to be highly-regulated by miRs, including MGAT2, MGAT4A and FUT8 (which controls core fucosylation). Currently miR:mRNA interactions have been validated for MGAT4A^{42, 50} and FUT8^51–57. MGAT4A catalyzes the transfer of GlcNAc to the biantennary (or tri-antennary) core structure of N-linked oligosaccharides to form a β1-4 linkage on the Manα1,3 arm (Figure 3). In recent work, miR-424 was shown to directly repress MGAT4A expression⁴². This miR inhibits cell cycle progression and was thought to mediate that effect through repression of CCDN1 and CDC25A, known cell cycle promoters. Downregulation of MGAT4A expression was also sufficient to inhibit cell cycle progression, indicating that this substructure is essential to intra- and inter-cellular interactions that maintain cell growth. Core fucosylation, which is controlled by FUT8, is dysregulated in many diseases including melanoma metastasis⁷, lung adenocarcinoma⁵⁸, liver cancer^{8, 9} and emphysema^59–63. Increasingly, this modification is found to have a direct role in promoting disease^{59, 60, 64–68}. Currently there are five miRs known to target FUT8, all of which have roles in tumor progression^{8, 9, 69}. Again, our knowledge of the miR regulation of the N-glycan pathway is woefully incomplete, but a picture is emerging in which glycogenes that are highly regulated are often involved in complex disease states, in line with the hypothesis of Joshi et al².

Figure 3.

Regulation of MGAT4A, and corresponding β-1,4 branching, by miR-424-5p as an example of how miR modulation of glycan structures can drive a biological state (cell growth) ⁴². Structures showing loss of β-1,4 branching are highlighted.

MICRORNA PROXY HYPOTHESIS AND APPLICATION TO GLYCOSYLATION

In one of the earliest examples of glycan regulation by miRs, Hernando and coworkers identified the GALNT7 as a target for miR-30d, a microRNA that promoted melanoma metastasis in patients and mouse models. Downregulation of GALNT7 was found to phenocopy miR-30d, increasing metastasis as a result of inhibiting O-glycosylation¹⁰. This showcases a common theme in miR biology, namely that downregulation of the targets of a miR phenocopies the effects of miR expression. This observation led us to propose the microRNA proxy hypothesis. Our hypothesis states that the regulation of protein expression by changes in the expression levels of miRs identifies proteins holding a privileged position in driving the underlying biology. In other words, if a miR drives a specific biological phenotype, such as migration or metastasis, the targets of that miR will drive the same biological phenotype. Thus, miRs can be used to identify (by proxy) the biological functions of specific glycosylation enzymes (or other proteins). We first formulated and tested this powerful hypothesis in a publication in 2015³. In that work, we examined the targets of miR-200b-3p, a miR that controls epithelial to mesenchymal transition (EMT). This miR is high in epithelial cells and low in mesenchymal cells. We identified 5 targets of miR-200b-3p and tested 3 of them to see whether inhibiting the expression of these enzymes would phenocopy overexpression of the miR. In all 3 cases (B3GLCT, ST3GAL5 and ST6GALNAC5), mesenchymal cells reverted to an epithelial state upon repression of these glycosylation enzymes. This phenotype was not transduced by repression of the transcription factor ZEB1, another target of miR-200b-3p commonly thought to be responsible for the EMT phenotype. Instead knockdown of all 3 glycogenes caused increases in ZEB1 levels, arguing that inhibiting glycosylation can alter EMT independent of the transcription factor. This provides evidence that miRs target key hubs driving the biological phenotypes that they regulate, in line with our hypothesis. A further example of using a known miR phenotype to identify glycan function comes from the aforementioned MGAT4A work⁴². Here, a role in cell cycle was found.

To date, all of the work examining our hypothesis has focused on single miRs. This is because currently the prediction algorithms for miR:mRNA interactions are inaccurate^{70, 71}. This inaccuracy may stem from the use of transcriptional data in creating the algorithms. Although commonly used as a metric of protein abundance, transcriptional data is not accurate to the proteome ⁷². A recent analysis of the agreement between protein expression and mRNA expression levels using data from the Human Proteome Map and Genotype-Tissue expression project found strong concordance in the expression levels for only 6.1% of genes⁷³. For low abundance proteins, such as glycogenes, it is known that transcription levels are not accurate to protein abundance ⁷⁴. All miR interactions impact translation, and at best only ~80% of interactions impact the transcriptome⁷⁵. This may be lower for low abundance genes, where transcriptional data is inherently more noisy. Thus, reliance on the transcriptome may bias current algorithms. At present, studies into miR:mRNA interactions requires that each interaction be validated by luciferase assay. If one were to study multiple miRs that co-regulate a biological phenotype, this would then require tens to hundreds of luciferase assays to validate interactions and identify a common target set. Improving the prediction algorithms for miR regulation of glycogenes, and other gene sets, is crucial for testing our hypothesis and truly understanding miR regulation of protein expression.

CONCLUSIONS

Glycosylation enzymes that are more tightly regulated appear to be more prevalent in controlling underlying complex disease states. microRNA is an emerging regulator of glycosylation, helping to provide fine tuning for the low abundance glycan biosynthetic enzymes. Given the emerging importance of miRs in disease and their potential to identify genes that underlie specific biological function, it is clear that more attention should be paid to miR:glycogene interactions.

ABBREVIATIONS

EMT

epithelial-mesesnchymal transition

Fuc

fucose

Gal

galactose

GWAS

genome-wide association study

Glc

glucose

GFP

green fluorescent protein

Man

mannose

miRNA, miR

microRNA

GalNAc

N-acetylgalactosamine

GlcNAc

N-acetylglucosamine

OGA

O-GlcNAcase

OGT

O-GlcNAcyltransferase

RISC

RNA induced silencing complex

Neu5Ac

sialic acid

UTR

un-translated region