O-GlcNAc cycling shows neuroprotective potential in C. elegans models of neurodegenerative disease

Abstract

C. elegans has proven to be an excellent organism in which to model human neurodegenerative disease.^1–7 The worm’s simple nervous system, lineage, and neural maps, easily scored movement phenotypes, and robust forward and reverse genetics make it optimal for studying age-dependent processes on a reasonable time scale. A popular approach has been the introduction of transgenes expressing GFP-tagged proteotoxic human proteins into neurons leading to visible aggregation or movement phenotypes.^2,4,6,8–13 In addition, the maintenance of proteostasis networks has been extensively studied using the power of worm genetics.^8–13 These networks include genes involved in insulin-like signaling, the heat shock response, the response to hypoxia, and mTOR and AMPK pathways linked to aging.¹⁴ Another pathway with suggestive links to neurodegeneration is the O-GlcNAc cycling pathway, a nutrient-dependent post-translational modification known to be altered in brains from patients with Alzheimer disease.^15–19 In this commentary, we summarize our recent findings showing that viable mutants of O-GlcNAc cycling in C. elegans dramatically alter the neurotoxicity of four distinct C. elegans models of neurodegenerative disease.⁷ Mutants in O-GlcNAc cycling alter the toxicity of mutant tau, polyglutamine expansion reporters, and amyloid β-peptide. The findings further suggest that O-GlcNAc cycling acts at many steps in the lifecycle of aggregation-prone targets. The C. elegans system is likely to continue to provide insights into this complex problem. The involvement of O-GlcNAc cycling in the maintenance of proteostasis raises the possibility of targeting the enzymes catalyzing this critical post-translational modification for therapeutic intervention.

Introduction

Hexosamine signaling and O-GlcNAc cycling

The highly conserved enzymes of nuclear and cytoplasmic O-GlcNAc cycling are the O-GlcNAc transferase (OGT; C. elegans ogt-1) and the O-GlcNAcase (OGA; C. elegans oga-1). These enzymes are found in all higher metazoans and are composed of highly conserved structural domains.^15,18,20-24 OGT is composed of a tetratricopeptide repeat region of varying length (9–14 repeats) followed by a catalytic domain capable of binding both protein substrates and the nucleotide sugar UDP-GlcNAc. UDP-GlcNAc is synthesized from precursors deriving from nutrient sources and is, therefore, responsive to nutrient levels. Thus, the synthesis of UDP-GlcNAc is influenced by the availability of nutrients making it a potential “sensor” of cellular nutrient status (see Fig. 1, top). The catalytic properties of OGT are also tailored for this proposed sensing function. UDP-GlcNAc binds to OGT prior to peptide binding; if peptide binds first, sugar nucleotide binding is blocked.²⁵

Figure 1. Nutrient-driven O-GlcNAcylation maintained by the enzymes of O-GlcNAc cycling tips the balance of the complex Proteostasis network. Like phosphorylation cascades, reversible O-GlcNAcylation impacts cellular signaling (top). The O-GlcNAcase (OGA-1) and O-GlcNAc transferase (OGT-1) alter cellular signaling cascades in a manner that is dependent upon cellular nutrient levels. Nutrient-derived precursors are converted to UDP-GlcNAc that is used for the addition of O-GlcNAc to proteins by OGT-1. O-GlcNAc is removed by OGA-1. Known targets for this modification include chromatin-associated transcription factors, components of the translation machinery and regulators of protein degradation such as the proteasome and autophagy. By tipping the balance of these varied pathways, O-GlcNAc cycling alters the proteotoxicity associated with several C. elegans models of Human neurodegenerative disease including tauopathy, amyloid β-peptide, and polyglutamine expansion. In the C. elegans models, ogt-1 loss of function was neuroprotective while oga-1 loss of function increased proteotoxicity. The increased proteotoxicity leads to neurodegeneration and apoptosis.

The function of OGT is essential in vertebrates such as mice with gene knockouts dying at embryonic stage E5.5.^22,24 The OGT gene in Drosophila, Super Sex Combs (sxc) is also essential for embryonic development. Sxc(Ogt) is essential for polycomb repression in Drosophila where sxc(Ogt) loss-of-function alleles yield homeotic defects due to deregulated Hox gene expression.^21,24,26,27 In contrast, null alleles of the C. elegans homolog ogt-1 are viable and fertile.²⁸ Our previous work has provided genetic evidence that both ogt-1 and oga-1 are linked to insulin-like signaling in C. elegans influencing dauer entry and longevity, glucose stress, and UV stress.^23,24,29,30 We also showed that OGT-mediated O-GlcNAc addition occurs on proteins occupying the promoters of over 800 genes influencing stress, insulin signaling, and immunity.³¹ Like OGT, the O-GlcNAcase is also highly conserved^15,18,20-24 mutants, influence insulin-like signaling (dauer and longevity).^29-31 Thus, both enzymes of O-GlcNAc cycling are viable in C. elegans making it uniquely amenable for studying the role of O-GlcNAc addition and removal in a genetically tractable organism.

O-GlcNAc and neurodegeneration

The family of age-associated diseases includes Alzheimer disease, Huntington disease, Parkinson disease, and the Tauopathies.^1-7 These diseases share as a common pathological feature the accumulation of protein aggregates and are thus often termed proteopathies. They are widely thought to be due to a collapse of neuronal proteostasis mechanisms leading to loss of motor function or memory loss depending upon the affected neurons. These are also diseases of enormous importance; Alzheimer disease afflicts over 35 million people worldwide. Alzheimer is associated with decreased glucose uptake in affected neurons suggesting that metabolism may influence disease progression. One key nutrient-responsive pathway that has been linked to the pathophysiology of neurodegenerative disease is O-GlcNAc metabolism. Both OGT and OGA are highly expressed in brain; approximately 300 brain proteins have been identified as O-GlcNAc modified using proteomic approaches. In addition, the known neurodegenerative disease targets tau, and β-amyloid proteins are O-GlcNAcylated.⁷ We chose to study the role of O-GlcNAc cycling in several neurodegenerative diseases in the worm system because of the unique experimental advantages and extensive basic understanding of the C. elegans proteostasis network.

Mutants in O-GlcNAc Cycling Influence the Neurotoxicity of Diverse Worm Models of Neurodegeneration

Several well-characterized C. elegans strains have been developed by other groups to model a variety of neurodegenerative diseases. We used many of these models of neurodegenerative disease to explore the possible role of O-GlcNAc cycling in neuronal dysfunction. In strain CK10, (bkIs10[aex-3p::tau-V337M], myo-2p::GFP), the aex-3 “pan-neuronal” promoter drives the expression of the 4R1N isoform of human tau with a V337M mutation that has been identified in human frontotemporal dementia with parkinsonism chromosome 17 type (FTDP-17).^32,33 We also used strain HA659, rtIs11(osm-10p::Htn-Q150, osm-10p::OSM-10::GFP), in which the OSM-10::GFP fusion protein is co-expressed with another fusion protein containing the N-terminal 171 amino acids of human Huntington protein and 150 glutamine residues (Q150) under the osm-10 promoter in a small subset of neurons.³⁴ In strain Q40-YFP, another Huntington model, a protein consisting of 40 glutamine residues (Q40) is fused to the yellow fluorescent protein (Q40-YFP) allowing expression in muscles under the unc-54 promoter.³⁵ We also obtained strain CL2292 dvIs36(pAT2[myo-3::GFP::degron]; pRF4[rol-6{su250]}) for use as a general proteasome substrate. This strain expresses a fusion protein comprised of the 16-residue CL1 degron peptide fused to the C-terminus of GFP expressed in muscles under the control of the myo-3 promoter.³⁶ An amyloid model was also used: strain CL2006 dvIs(pCL12[unc-54::Aß1-42]; pRF4[rol-6{su1006}]) expresses the 42-amino acid β-amyloid peptide under the unc-54 promoter allowing muscle expression.³⁷

These strains were systematically crossed into three independent ogt-1-null alleles (ogt-1[ok430], ogt-1[ok1474], and ogt-1[tm1046]) as well as a strain harboring the oga-1 allele ok1207. In some cases, the models were also introduced into strains harboring the double mutant: ogt-1(ok430); oga-1(tm1207). Extensive phenotyping of the impact of the mutants on the neurodegenerative disease models revealed striking and somewhat surprising results. In each case, loss-of-function alleles of ogt-1 decreased aggregation and diminished proteotoxicity. The ogt-1 mutant restored near normal movement to worm strains harboring the tau V337M transgene. In wild-type worms, the human tau variant expressed as a transgene caused a severe trashing (swimming) phenotype. Interestingly, this severe movement defect was not corrected by oga-1 loss-of-function mutations. In the two different models of the polyQ toxicity associated with Huntington disease, ogt-1 mutants showed reduced protein aggregation whereas oga-1 mutants dramatically increased aggregate formation. Finally, in a model of β-amyloid peptide toxicity, oga-1 mutants showed a rapid, age-dependent loss of movement with dramatically increased severity compared with wild-type animals. To confirm that alterations in the proteotoxicity observed with the ogt-1 and oga-1-null mutations were due to the O-GlcNAc modification, we also compared the proteotoxicity phenotype of different transgenic models in the ogt-1l oga-1 double mutants to those of the oga-1 or ogt-1 single mutants. OGT functions upstream of OGA; the absence of O-GlcNAcylation of protein substrates in the OGT-null mutant makes the presence or absence of OGA activity irrelevant for O-GlcNAc cycling. Thus, if the effects of proteotoxicity was dependent upon O-GlcNAcylation, similar phenotypes will be observed for both the ogt-1 single mutant and the ogt-1; oga-1 double mutants. That was precisely what we observed for multiple neurodegenerative models.⁷

As a non-disease -ssociated control, we used the strain in which a toxic degron peptide fusion was expressed in C. elegans muscles inducing perinuclear aggregates and paralysis in the transgenic animals.³⁶ Intriguingly, we found that neither the stability of aggregates induced by the toxic degron fusion peptide nor the paralysis phenotype was influenced by either of the O-GlcNAc cycling mutants. One interpretation of this finding is that the aggregation of disease-causing proteins is distinct from the aggregates induced by a generic proteasomal substrate like the degron peptide. Taken together, these findings indicated that the enzymes of O-GlcNAc cycling OGT-1 and OGA-1 were key regulators of proteostasis and could influence the neurotoxicity of worm models of human tauopathy, Huntington disease, and Alzheimer disease. We sought to understand how these protective and toxic effects were mediated.

O-GlcNAcylation Regulates Proteostasis and Modulates Neurodegeneration by Altering Insulin-Like Signaling and Protein Degradation

The ability of O-GlcNAc cycling mutants to modulate proteotoxicity was explored in a number of ways. Although O-GlcNAc cycling can alter transcription, we quickly ruled out simple alterations in transgene expression. We also examined total ubiquinylated proteins and proteosomal function since O-GlcNAc has been shown to alter both of these activities.³⁸ There was a small, but significant, change in levels of ubiquitinylated proteins consistent with the activation of proteasomes in the ogt-1 mutants. We also examined autophagy, a process that has been shown to play a role in neurodegeneration and is extensively studied in C. elegans.⁵ Autophagy is a catabolic pathway activated by cellular stress, including starvation, pathogen infection, and proteotoxicity. It proceeds by the formation of double-membrane vesicles, called autophagosomes. In mutants lacking either of O-GlcNAc cycling, there was an accumulation of GFP::LGG-1 (the C. elegans homolog of Atg8 and LC3) and increased phosphatidylethanolamine (PE)-modified GFP::LGG-1 upon starvation. These are key triggers for the formation of autophagosomes that can surround organelles, glycogen granules, and protein aggregates. The autophagosomes fuse to lysosomes allowing for degradation of the entrapped content. Thus, we demonstrated that nutrient-driven O-GlcNAc cycling clearly modulates autophagy in C. elegans. We have subsequently expanded on these findings and speculate that O-GlcNAc cycling is a key nutrient-responsive regulator of autophagic flux acting at multiple levels, possibly including direct modification of Beclin-1 (BECN1) and Bcl-2 (BCL2).³⁹ In mammals, Beclin-1 and Bcl-2 expression levels are thought to regulate an autophagic-apoptotic “switch.”⁴⁰ Since O-GlcNAc cycling has been implicated in both mammalian apoptosis⁴¹ and neurodegeneration,^7,42 O-GlcNAcylation has the potential to influence the prosurvival-proapoptoticbalance in neurons. Upsetting this balance may be a key factor in the induction of age-dependent neurodegenerative disease (Fig. 1).

We also explored the insulin-signaling pathway that is a critical component of the network of genes regulating proteostasis in C. elegans.^43,44 Previously, we have shown direct genetic interactions between the insulin-like signaling pathway in C. elegans and both ogt-1 and oga-1.^28-31,39 Therefore, we examined the transcriptional response of strains harboring the tau V337M or Htn-PolyQ transgenes in both wild-type and ogt-1 mutant strains. We focused upon the impact on known regulators of proteostasis:, hsf-1 and hif-1, and daf-16. HSF-1, encoded by hsf-1, functions as a transcriptional regulator of stress-induced gene expression whose activity is required for heat-shock and proteotoxicity response, development, immunity, and regulation of adult lifespan. hif-1 encodes an ortholog of the mammalian hypoxia-induced factor HIF-1 required for lifespan and survival in hypoxic environments. Both hsf-1 and hif-1 genetically interact with the insulin-like signaling pathway. We found that the ogt-1 loss-of-function mutant induced only modest changes in hsf-1 and hif-1 levels in response to the proteotoxic transgenes. However, daf-16 and many of its downstream targets were significantly altered by tau V337M and Htn-PolyQ transgenes. Particularly striking were the changes observed in daf-16 isoforms; ogt-1 mutants elevated the expression of daf-16f relative to other isoforms. Intriguingly, this isoform has been linked to the regulation of longevity.⁴⁵ Thus, the altered toxicity of the neurodegenerative models observed in the O-GlcNAc mutants is linked to changes in daf-16-dependent transcription and the protein degradation machinery. As Figure 1 indicates, the enzymes of O-GlcNAc cycling may modulate proteostasis at many levels. Perhaps the primary impact is with cellular signaling pathways including insulin-like signaling. Downstream transcriptional cascades are also likely targets. In general terms, O-GlcNAcase has been linked to transcriptional activation and OGT to transcriptional repression.²¹ O-GlcNAc cycling may also contribute to mRNA stabilization since O-GlcNAc modification of the translational machinery is required for aggregation of untranslated messenger ribonucleoproteins into stress granules.⁴⁶ Finally, protein stability is influenced by O-GlcNAc cycling through direct modification of the proteasome and autophagy machinery, as previously described.^{38, 39}

OGA-1 and OGT-1, acting through nutrient-driven O-GlcNAcylation of a variety of substrates, may influence the balance of the proteostasis network leading to an increase in toxic aggregate formation and associated toxicity (Fig. 1). Loss of OGT-1 function tips the balance in favor of decreased aggregation and decreased toxicity. Loss of OGA-1 increases O-GlcNAc levels tipping the balance toward increased toxic aggregate formation. The cascade of maladaptive changes to this proteostasis network leads to proteotoxicity and neurodegeneration and apoptosis. Thus, nutrient-driven O-GlcNAc cycling contributes to maintaining proteostasis; when deregulated it can influence the severity of toxicity in worm models of human neurodegeneration at multiple levels.

Relationship to Human Neurodegeneration and Future Prospects

Our findings provide strong genetic evidence that O-GlcNAc cycling impacts the toxicity of human diseases modeled in C. elegans. The findings may appear surprising in light of reports suggesting that O-GlcNAcase inhibitors can decrease the toxicity of some forms of tauopathy in mice and tissue culture.^19,42 However, it should be noted that the genetic experiments we performed removed OGA-1 and OGT-1 activity throughout development while the pharmacologic manipulations were performed acutely. This is likely to be important, since many more factors are likely to come into play in a developmental context where compensation may occur (see Fig. 1). Since the highly conserved proteostasis machinery is complex, the proper balance of factors required may also differ somewhat between C. elegans and vertebrates. In addition, the multi-domain enzymes of O-GlcNAc cycling are likely to perform tethering and scaffolding functions distinct from their catalytic activities. The null alleles used in our C. elegans study removed all of these interaction domains, not just those blocked by inhibiting catalysis of the O-GlcNAcase. Although the precise mechanisms of proteotoxic modulation are not fully understood, the enzymes of O-GlcNAc cycling are clearly attractive targets for drug intervention. The C. elegans system described here could be used for the identification of small molecule modulators of O-GlcNAc metabolism influencing neurodegeneration. In addition, genetic screens aimed at identifying suppressors of the O-GlcNAc-dependent modulation of neurodegeneration may prove fruitful. As central players in the maintenance of proteostasis, the nutrient-driven enzymes of O-GlcNAc cycling may provide a key mechanistic link between brain nutrient mobilization and age-dependent neurodegenerative disease.

Wang P, Lazarus BD, Forsythe ME, Love DC, Krause MW, Hanover JA. O-GlcNAc cycling mutants modulate proteotoxicity in Caenorhabditis elegans models of human neurodegenerative diseases. Proc Natl Acad Sci U S A. 2012;109:17669–74. doi: 10.1073/pnas.1205748109.

10.4161/worm.27043