OGA Inhibition Alters Energetics and Nutrient Sensing in Alzheimer’s Disease Cytoplasmic Hybrids

Abstract

Background:

Alzheimer’s disease (AD) features reductions in key bioenergetic fluxes and perturbed mitochondrial function. Cytoplasmic hybrids (cybrids) generated through the transfer of AD subject mitochondria to mtDNA-depleted SH-SY5Y neuroblastoma cells recapitulate some of these features in an in vitro setting.

Objective:

For this study, we used the AD cybrid model to assess the impact of a nutrient-excess like-state via increasing O-GlcNAcylation on whole cell and mitochondrial homeostasis.

Methods:

We induced increased O-GlcNAc by treating AD and control cybrid cell lines with Thiamet G (TMG), an inhibitor of the O-GlcNAcase enzyme that mediates removal of the nutrient-dependent O-GlcNAc modification.

Results:

Relative to control cybrid cell lines, AD cybrid lines showed a blunted response to TMG-induced O-GlcNAcylation. At baseline, AD cybrid cell line mitochondria showed partial activation of several proteins that help maintain bioenergetic homeostasis such as AMP-Regulated Kinase suggesting that AD mitochondria initiate a state of nutrient stress promoting energetic compensation; however, this compensation reduces the capacity of cells to respond to additional nutrient-related stresses such as TMG treatment. Also, TMG caused disruptions in acetylation and Sirtuin 3 expression, while lowing total energetic output of the cell.

Conclusion:

Together, these findings suggest that modulation of O-GlcNAc is essential for proper energetic function of the mitochondria, and AD mitochondrial capacity to handle nutrient-excess is limited.

INTRODUCTION

Type 2 diabetes (DM2) accentuates cognitive decline and reduces insulin signaling in Alzheimer’s disease (AD) leading some to propose mechanistic overlap exists between these disorders. DM2-associated hyperglycemia at a superficial level mimics a nutrient-excess state, and nutrient-excess states trigger compensatory attempts to preserve energy homeostasis. By virtue of their role in cell respiration and their pivotal role in other carbon-utilizing bioenergetic fluxes, mitochondria are critical to energy homeostasis and multiple lines of investigation demonstrate perturbed mitochondrial function in AD subjects. Since mitochondrial function is sensitive to nutrient inputs, cells have evolved strategies for monitoring and responding to nutrient states and energy levels. One such strategy takes advantage of N-acetylglucosamine (GlcNAc), a hexosamine sugar whose levels uniquely reflect the cell’s carbohydrate, amino acid, and lipid fuel status.

O-GlcNAc is a ubiquitously expressed post-translational modification of serine or threonine residues in nuclear, cytoplasmic or mitochondrial proteins. The dynamic addition and removal of O-GlcNAc by O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) respectively is referred to as O-GlcNAc cycling [1]. UDP-GlcNAc, the donor sugar for OGT catalyzed reactions, is synthesized through the hexosamine biosynthetic pathway (HBP) [1]. Since multiple metabolic pathways feed into the HBP, O-GlcNAcylation acts as a nutrient sensor allowing cells to integrate flux through different metabolic pathways to respond to changes in nutrient levels [2]. Furthermore, O-GlcNAcylation is sensitive to inflammatory or stress signals [3, 4]. Thus, O-GlcNAc can integrate stress signals to modulate metabolic activity [5–7]. Because O-GlcNAcylation is sensitive to the cell’s ability to utilize metabolites, disruptions in HBP nutrient flux could contribute to disease pathogenesis and progression.

Changes in O-GlcNAc homeostasis are observed in AD. Measurements of O-GlcNAcylation, OGT, and OGA expression in 9 aged and sex matched control and AD human brains found an increase in O-GlcNAc levels and a decline in OGA expression [8], reflecting data from another independent study [9]. These data suggest a loss of OGA expression with AD progression with subsequent increased O-GlcNAcylation. Conversely, several studies propose a link between reduced tau O-GlcNAcylation and a gain in tau phosphorylation with AD progression [10]. Manipulation of O-GlcNAcylation therefore may offer a potential AD therapeutic target. Current efforts at this emphasize an OGA inhibitor, Thiamet-G (TMG), which enhances O-GlcNAcylation of tau and other proteins [11, 12]. However, TMG-induced increase in O-GlcNAcylation mimics a nutrient excess state such as seen with hyperglycemia. This “elevated” O-GlcNAc state influences mitochondrial energetics.

Recent studies emphasize a broad role for O-GlcNAc in regulating mitochondrial function. Numerous mitochondrial proteins are modified by O-GlcNAc [13], and both OGT and OGA localize to the mitochondria [14]. Acute OGT and OGA inhibition significantly affects mitochondrial respiration, ATP production, and membrane potential in neonatal rat cardiomyocytes [14]. Inducing OGT expression in mitochondria increases apoptosis, and suggests mitochondrial O-GlcNAcylation is important for proper function [15]. Recently, we showed that increased expression of OGT and OGA in SH-SY5Y neuroblastoma cells lowers mitochondrial respiration, disrupts morphology, and decreases expression of respiratory chain and the tricarboxylic acid cycle (TCA) proteins [16]. Furthermore, we demonstrated that inhibition of OGA reduces respiration, increases total NAD+/NADH ratio, and lowers ROS output [17]. The reduction in ROS levels is coupled to a decrease in the Nrf2 antioxidant response [17, 18].

In this study, we used TMG to artificially mimic a nutrient-excess state of elevated O-GlcNAc in AD cytoplasmic hybrid (cybrid) lines, which are generated through the transfer of AD subject mitochondria to mtDNA-depleted SY5Y (ρ0) cells that divide under special cell culture conditions. AD cybrid cell lines recapitulate mitochondrial functional parameters seen in direct studies of AD subject mitochondria, thereby allowing one to model AD mitochondrial function in vitro. These cells allowed us to assess the ability of cells with AD-associated mitochondrial dysfunction to respond to a TMG-induced nutrient-excess state. Importantly, we found that AD mitochondrial induce compensatory activation of several nutrient sensing pathways but TMG-induced nutrient-excess blunts compensation while altering mitochondrial acetylation and Sirtuin 3 expression (SIRT3).

MATERIAL AND METHODS

Antibodies

All antibodies were used at 1 : 2000 dilution for immunoblotting. The following antibodies were purchased from Abcam: NRF2 (ab62352), NADPH oxidase (ab79971), citrate synthase (ab96600), aconitase (ab110321), acetylation (ab80178), ACSS1 (ab101570), NDUFA9 (ab14713), ATP5F1 (ab117991), TFAM (ab47517), and ferredoxin reductase (ab16873). The following antibodies were purchased from Cell Signaling Technologies: AMPK (5831), ACC1 (4190), phosphorylated-ACC1 (11818), mTOR (2972), 4EBP1 (9644), phosphorylated-4EBP1 (2855), and Sirtuin 3 (C73E3). Sirtuin 5 antibody was from Millipore (ABE198). MnSOD was from BD Bioscience (611580). Actin was from Sigma (A2066). RL2 (anti-O-GlcNAc) was from ThermoFisher (MA1–072). OGT and OGA antibodies were a kind gift from Gerald Hart (University of Georgia Complex Carbohydrate Research Center).

Cell culture

As previously described [19], the cybrid lines used were generated via transfer of platelet mitochondria from aged-matched control (n = 4) and AD (n = 4) subjects to SH-SY5Y ρ0 cells. All cybrid lines were cultured in DMEM (Sigma), 44 mM sodium bicarbonate (Sigma), 5 mM glucose (Sigma), 15 mg/liter phenol red (Sigma), and supplemented with 10% fetal bovine serum (FBS; Gemini), 1% penicillin/streptomycin (Gibco), and 1% GlutaMAX (Gibco). Cells were adopted to normal glycemic condition for 2 weeks prior to 2-week adaption with Thiamet-G (TMG; SD Specialty Chemicals). Media was replaced daily.

Cell lysis

Cells were lysed as described previously [16]. Pellets were suspended in Igepal lysis buffer (20 mM Tris-HCl pH 7.4, Sigma; 150 mM NaCl, Fisher Scientific; 1 mM EDTA, Sigma; 1 mM DTT, Sigma; 40 mM GlcNAc, Sigma; and 1% Igepal, Sigma) and lysed on ice for 20 min with occasional vortexing.

Mitochondrial isolation

Mitochondrial were isolated using the modified nitrogen cavitation method [20]. A minimum of 2 × 10⁸ were used. Cells were digested of culture dishes with trypsin (Sigma), washed twice with PBS, and resuspended in 3 ml of ice-cold mitochondrial isolation buffer (225 mM Mannitol, Sigma; 75 mM sucrose, Sigma; 5 mM Hepes pH 7.4, Fisher Scientific; 1 mM EGTA, Sigma). The cell suspension was placed into the pre-chilled cavitation chamber (Nitrogen Bomb, Parr Instruments) and subjected to 900 p.s.i for 15 min. Cell suspension was centrifuged at 20,000 × g for 10 min. The pellet (crude mitochondrial/heavy membrane fraction) was washed three times with 500 μl of isolation buffer. The pellet was suspended in Igepal isolation buffer.

Electrophoresis and immunoblotting

SDS-PAGE was performed using 4–15% gradient polyacrylamide gels (Criterion Gels, Biorad). Cell lysates were mixed with protein solubility buffer (3 parts sample, 1 part mix) (100 mM Tris, pH 6.8, Sigma; 10 mM EDTA, Sigma; 8% SDS, Sigma; 50% sucrose, Sigma; 5% β-mercaptoethanol, Sigma; 0.08% pyronin-Y, Sigma) and separated at 130 volts. Samples were transferred onto PVDF membrane at 0.4 amps (Fisher Scientific). All antibodies were used at 1 : 2000 dilution for immunoblotting. All blots were blocked in TBST (Tris-HCl pH 7.6, Sigma; 150 mM NaCl, Sigma; 0.05% Tween, Sigma) with 3% BSA (Bovine Serum Albumin, Midwest Scientific). Blots were developed using HRP-conjugated secondary antibodies (anti-rabbit HRP and antimouse HRP, Biorad) and chemiluminescent substrate (Hyglo, Denville Scientific). Blots were stripped in 100 mM glycine (pH 2.5, Sigma) for 1 h, washed in TBST, and treated as before [16].

Cellular respiration and glycolysis assays

A XF96 analyzer (Agilent Seahorse XF technology) was used to measure cellular respiration and glycolysis. 25,000 control, or TMG-treated cells (control or AD cybrids) were seeded per well in a Seahorse 96-well cell culture plate 24 h prior to experimental assay. For respiration assays, cells were incubated in unbuffered DMEM (Sigma), 20 mM glucose (Sigma), phenol red (Sigma), 200 mM Gluta Max-1 (Gibco), NaCl (Sigma) at 37°C in a CO₂-free incubator for 1 h prior to loading. Oxygen consumption rate (OCR) was measured over a period of 100 min. Drugs oligomycin (0.5 μM, Sigma), FCCP (0.5 μM, Sigma), antimycin A (0.2 μM, Sigma), and rotenone (0.1 μM, Sigma) were added to each well sequentially at various time points during the assay [16]. For glycolysis assays, 20,000 cells were plated. Cells were serum starved for 24 h after plating prior to running the assay. Extra-cellular acidification rate (ECAR) was measured over a period of 100 min. Glucose (25 mM), oligomycin (1 μM), and 2-deoxy-Dglucose (100 mM, Sigma) were added at specific time points during the assay [16].

Statistical analysis

Densitometry analysis was done on immunoblots using Image J software. Statistical analysis was performed on all densitometry data using two-way mixed ANOVA, followed by pairwise tests, using paired tests where appropriate. Normality of distribution was assumed [21], as is standard for these data. For the data collected from the cellular respiration and glycolysis assays with respect to the different mitochondria lines, the values for technical replicates were averaged. Following that, the difference in means was compared using two-way mixed ANOVA, followed by pairwise tests.

RESULTS

AD cybrids have altered nutrient sensing

AD mitochondria have impaired capacity to respond to energetic demands and alterations in nutrient flux. Since alterations in O-GlcNAc levels affect mitochondrial function [17], we reasoned that manipulating O-GlcNAcylation would mimic a nutrient excess state impact AD mitochondrial function. Therefore, we used cytoplasmic hybrids (cybrids) to test our hypothesis. Mitochondria in AD cybrid cell lines recapitulate a variety of functional characteristics observed in mitochondria that reside in AD subject primary tissues. As previously described [19], the cybrids (C1 : 68-year-old male; C2 : 85-year-old male; C3 : 78-year-old female; and C4 : 73-year-old female) used were from aged-matched control and AD patients (AD1 : 81-year-old male; AD2 : 87-year-old female; AD3 : 74-year-old female; and AD4 : 76-year-old female), and the mitochondria were transferred into SH-SY5Y neuroblastoma cells lacking functional mitochondria. We would expect the AD participants who served as mtDNA donors would more likely be APOE4 carriers than control participants who served as mtDNA donors. As the cybrid technique expresses unique mtDNA sequences against a common nuclear background, we would expect this would mitigate the effects of an APOE4 allele. SH-SY5Y cells (female donor) are homozygous for the APOE3 allele therefore all cybrid lines were APOE3 homozygous. The 4 control cell lines were given the designation control 1–4, while the 4 AD cell lines were designated AD 1–4. Each cell line was adapted for 2 weeks in TMG to replicate a sustained elevation in O-GlcNAc mimicking how excess nutrients impacts O-GlcNAc. In cells from equal passage number, we measured O-GlcNAcylation, and OGT and OGA protein expression in these cells. At baseline, inter-group whole cell O-GlcNAcylation, OGT, and OGA levels were comparable, although OGT levels trended higher in the AD cybrids (Fig. 1A–D). Two weeks of TMG treatment as expected increased GlcNAc and OGA, and decreased OGT protein levels [17].

Fig. 1.

AD cybrids treated with OGA Inhibitor TMG have increased O-GlcNAc. A) O-GlcNAc (RL2), OGT, and OGA levels in SH-SY5Y whole-cell lysates. Box plots represent quantitation of O-GlcNAc (B), OGT (C), and OGA (D) bands with actin used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

AD cybrids influence proteins and pathways that respond to changes in nutrient status. Importantly, elevated O-GlcNAc impact the function of other nutrient sensing pathways such as AMP-regulated kinase (AMPK) and mammalian target of rapamycin (mTOR) [2]. Therefore, we assessed activation of these pathways in the AD cybrids at baseline and following TMG treatment. From whole-cell lysates, we did not see baseline inter-group differences between control and AD cybrids in the mTOR pathway (mTOR, ACC1, ACC1 phosphorylation, 4EBP1 phosphorylation) (Fig. 2A–G); but at baseline the AD cybrids showed decreased AMPK and increased 4EBP1 levels. Following TMG treatment, mTOR decreased in the AD but not in the control cybrid group. Although 4EBP1 phosphorylation trended with mTOR levels, the change in phosphorylation was not statically significant. Although TMG did not alter whole cell intra-group levels of ACC1, pACC1, AMPK, 4EBP1, and p4EBP1. pACC1 was elevated in the TMG treated AD cybrids suggesting increases in AMPK activation.

Fig. 2.

AD cybrids show decreased AMPK expression. A) AMPK and mTOR nutrient sensing pathway levels in SH-SY5Y whole-cell lysates. Box plots represent quantitation of mTOR (B), phosphorylated ACC1 (C), ACC1 (D), AMPK (E), phosphorylated 4EBP1 (F), and 4EBP1 (G) bands with citrate synthase (CS) used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

Previously, sustained TMG treatment lowered ROS levels and expression of nuclear factor (erythroid-derived)-like 2 (NRF2), the transcription factor that regulates antioxidant response. We measured no difference between NRF2 levels from control and AD cybrids. TMG treatment had no effect on AD cybrids, but NRF2 levels trended lower in the TMG-treated controls (Fig. 3A–C). We also measured NADPH oxidase expression in the cybrids. NADPH oxidase is the main source of cytoplasmic ROS and contributes to cellular damage in neurodegeneration [22]. NADPH oxidase levels were significantly higher in the AD cybrids compared to control while TMG had no effect on NADPH oxidase levels (Fig. 3C).

Fig. 3.

AD cybrids have higher expression of NADPH Oxidase. A) NRF2 and NADPH Oxidase levels in SH-SY5Y whole-cell lysates. Box plots represent quantitation of NRF2 (B) and NADPH oxidase (C) bands with CS (Fig. 2) used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

Altered O-GlcNAc homeostasis impacts the mitochondrial proteome and acetylome

We similarly assessed O-GlcNAc status in mitochondria isolated from the AD and control cybrid lines (Fig. 4A–D). Although baseline mitochondrial inter-group O-GlcNAc and OGA levels were comparable, the OGT level was higher in the AD cybrids. TMG treatment again broadly increased O-GlcNAc and OGA, and decreased OGT within the mitochondria. Previously, we demonstrated that overexpression of OGT and OGA affects the composition of the mitochondrial proteome, therefore, we probed for mitochondrial proteins known to be regulated by O-GlcNAcylation [13, 16, 17]. TMG treatment altered several parameters exclusively in the control cybrids, including increases in NDUFA9, ATP5F, TFAM, Ferredoxin, and MnSOD (Fig. 5A–F). AD cybrids lacked the ability to compensate for these TMG induced changes. We found that AD and TMG treated AD cybrids had lower MnSOD levels but higher Ferredoxin levels.

Fig. 4.

OGA inhibition increases O-GlcNAc and decreases OGT expression in mitochondria. A) O-GlcNAc, OGT, and OGA levels in mitochondrial extract. Box plots represent quantitation of O-GlcNAc (B), OGT (C), and OGA (D) bands with Aconitase (Ac) used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

Fig. 5.

OGA inhibition influences the mitochondrial proteome. A) NDUFA9, ATP5F1, TFAM, MnSOD, and Ferridoxin levels in mitochondrial extract. Box plots represent quantitation of NDUFA9 (B), ATP5F1 (C), TFAM (D), MnSOD (E), and Ferridoxin (F) bands with Ac (Fig. 4) used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

Since O-GlcNAc is sensitive to levels of acetyl-CoA [23], we investigated the interplay between altered TMG-induced nutrient-excess and mitochondrial acetylation (Fig. 6A–D). Acetylation is a highly abundant mitochondrial post-translational modification and acetylation levels are linked to electron transport chain function [24]. TMG led to increased mitochondrial protein acetylation in both groups. Interestingly, SIRT3, the mitochondrial deacetylase, levels increased with TMG treatment but were lower in the AD groups including TMG treated AD cybrids. ACSS1 (Acetyl-CoA synthetase 1), which is involved in converting acetate into acetyl-CoA during ketogenesis [25], was increased after TMG treatment in both the control and AD cybrids.

Fig. 6.

OGA inhibition increases mitochondrial acetylation and SIRT3 expression except in AD cybrids. A) Acetylation, ACSS1, and SirT3 levels in mitochondrial extract. Box plots represent quantitation of Acetylation (B), ACSS1 (C), and SirT3 (D) bands with Ac (Fig. 4) used as a loading control. The colored region is the inter-quartile range while the bars represent points falling within 1.5 times the IQR. *p < 0.05. **p < 0.01. ***p < 0.001. WB, western blotting; Con, control; AD, Alzheimer’s disease; TMG, Thiamet G.

Sustained TMG treatment lowers metabolic function

Since we saw similar adaptive changes with TMG treatment as we previously described [17], we hypothesized that both control and AD cybrids treated with TMG would have lower metabolic rates. First, we performed a respiration stress test in the cybrid lines and measured OCR over time. We measured no significant difference between control and AD cybrid OCR (Fig. 7). However, TMG treatment did substantially suppress OCR in both control and AD cybrids (Fig. 7). TMG treatment lowered basal respiration (Fig. 7A); maximum respiration (Fig. 7B); proton leak rate; and ATP production (Fig. 7C, D). Next, we performed glycolytic stress test on cybrids and measured ECAR (Fig. 8). TMG treated cybrids had substantially lower glycolytic reserve (Fig. 8A); basal glycolytic rate (Fig. 8B); maximum glycolytic capacity (Fig. 8C), and non-glycolytic acidification rate (Fig. 8). These changes were consistent with previous results and demonstrate that the AD cybrid’s mitochondrial function changes in a manner similar to control cybrids.

Fig. 7.

Prolonged TMG treatment decreases mitochondrial energetic demands. Basal respiration (A), maximal respiration (B), proton leak rate (C), and ATP Production rate (D) in TMG treated SH-SY5Y cells was determined using a XF-96 analyzer (two-way, repeated measures ANOVA, experimental replicates n = 3; technical replicates n = 8, technical replicates n = 15 for Control 2 and AD 1). *p < 0.05. **p < 0.01. ***p < 0.001.

Fig. 8.

Prolonged TMG treatment alters glycolytic rate. Glycolytic reserve (A), basal glycolytic rate (B), glycolytic capacity (C), and non-glycolytic acidification (D) in TMG treated SH-SY5Y cells was determined using a XF-96 analyzer (two-way, repeated measures ANOVA, experimental replicates n = 3, technical replicates n = 15). *p < 0.05. **p < 0.01. ***p < 0.001.

DISCUSSION

AD presents with increased cerebral glucose, reduced neuronal glucose uptake, and impaired energetics demonstrating that AD is a disease of altered neuronal glucose utilization; thus, AD has been characterized as type 3 diabetes [26]. Therefore, we sought to characterize AD cytoplasmic hybrids as a model system to test how nutrient excess induced by OGA inhibition affected mitochondrial function [27]. Subsequently, we found that at baseline, some nutrient-sensitive parameters differed between the AD and control cybrid groups, which likely reflects mtDNA-determined differences in respiratory chain function. However, some baseline differences were exacerbated with TMG treatment including increased AMPK activity in the AD cybrids. Furthermore, compared to control cybrid mitochondria, AD cybrid mitochondria had increased OGT expression, and TMG treatment impacted mitochondrial acetylation though decreased SIRT3 expression. These results demonstrate the utility of using AD cybrids to understand how nutrient-excess as interpreted by O-GlcNAc contribute to mitochondrial malfunction.

The blunted AD cybrid TMG response could reflect a disconnect between the O-GlcNAc signal and the nutrient sensing apparatus. If so, this disconnect ultimately arises because of mtDNA-determined differences in respiratory chain function. Under this scenario it is not possible to determine whether a disconnect directly reflects an altered respiratory chain flux, or a downstream change such as oxidative stress. Alternatively, it is possible AD cybrids at baseline have already undergone adaptation to some type of mitochondrial or nutrient stress and cannot further compensate in the wake of additional nutrient stress induced by a TMG-dependent perturbation in O-GlcNAc.

With either scenario, it is important to consider how cells might perceive short-term nutrient-generated changes in O-GlcNAc versus long-term changes. We predict cells would perceive short-term nutrient-generated changes as adaptive and would serve to relieve stress. On the other hand, we predict cells would perceive long-term chronic nutrient changes, i.e., TMG treatment as maladaptive, with subsequent stress introduction. Our data are certainly consistent with this latter point. It is important to consider this possibility, since chronic high glucose or nutrient excess would elevate O-GlcNAc levels like TMG treatment.

OGA inhibition strongly impacts respiration. Previously, we demonstrated that TMG treatment lowered metabolic output of cells and tissues (ETC and glycolytic function), and reduced ROS production [17]. Importantly, TMG had the same impact on AD cybrids suggesting that AD mitochondria will reduce ETC function and ROS production with TMG treatment. AD neurons already suffer energy deficits [28], will OGA inhibition exacerbate these deficits or would the neurons become more efficient? Although reduced ROS production would be beneficial to the AD neuron, the chronic decline in energy production caused by elevated O-GlcNAc would exacerbate the AD neuron energy deficit. Importantly, O-GlcNAc induced declines in ATP production did influence other nutrient sensing pathways. Both AMPK and mTOR levels were altered in AD cybrids after TMG treatment. Thus, these data argue for a complex crosstalk between nutrient sensing pathways in AD, and would suggest that as long as these pathways can remain functional pathological decline would slow [29–31].

Excessive mitochondrial acetylation impairs electron transport chain function promoting metabolic syndrome [32, 33]. The lysine deacetylase Sirtuin 3 (SIRT3) modulates mitochondria acetylation, is essential for proper metabolic function, and impaired SIRT3 function contributes to metabolic syndrome [34–36]. TMG treatment elevated levels of SIRT3 in control cybrids; however, AD cybrids had lower SIRT3, and TMG treatment did not rescue AD SIRT3 expression. Healthy, robust mitochondria found in the control cybrids show acetylome reprogramming in response to OGA inhibition but AD mitochondria do not adapt properly to the changes in O-GlcNAcylation. AD mitochondria appear to lack the ability to promote SIRT3 expression after TMG treatment. Likely, AD mitochondria have reduced ability to modulate SIRT3 expression in general and this is amplified with OGA inhibition. Thus, the reduced ability to modulate SIRT3 expression in AD mitochondria would likely exacerbate metabolic defects within the AD brain [32]. Any ability to offset the loss of SIRT3 expression in AD would likely be advantageous since overexpression of SIRT3 improves mitochondrial energetics in mouse models of parkinsonism [37].

Finally, with therapeutic manipulation of O-GlcNAc moving into clinical development for AD treatment, our study points at the benefits and risks of manipulating O-GlcNAc in patients. The robust ability of cells and tissue to modulate changes to O-GlcNAc [38] allow for the targeting of OGA in AD treatment with minimal side effects to healthy cells, but how AD affected neurons respond to O-GlcNAc challenge is unknown. We argue that OGA inhibition would reduce energetic demand activating AMPK pathways while also reducing ROS production, but the effects in mitochondrial acetylation and AD SIRT3 expression could reduce efficacy over time. Certainly, more studies exploring how O-GlcNAc homeostasis affects AD is warranted.