Role of NAD⁺ and FAD in Ischemic Stroke Pathophysiology: An Epigenetic Nexus and Expanding Therapeutic Repertoire

Abstract

The redox coenzymes viz., oxidized β-nicotinamide adenine dinucleotide (NAD⁺) and flavin adenine dinucleotide (FAD) by way of generation of optimal reducing power and cellular energy currency (ATP), control a staggering array of metabolic reactions. The prominent cellular contenders for NAD⁺ utilization, inter alia, are sirtuins (SIRTs) and poly(ADP-ribose) polymerase (PARP-1), which have been significantly implicated in ischemic stroke (IS) pathogenesis. NAD⁺ and FAD are also two crucial epigenetic enzyme-required metabolites mediating histone deacetylation and poly(ADP-ribosyl)ation through SIRTs and PARP-1 respectively, and demethylation through FAD-mediated lysine specific demethylase activity. These enzymes and post-translational modifications impinge on the components of neurovascular unit, primarily neurons, and elicit diverse functional upshots in an ischemic brain. These could be circumstantially linked with attendant cognitive deficits and behavioral outcomes in post-stroke epoch. Parsing out the contribution of NAD⁺/FAD-synthesizing and utilizing enzymes towards epigenetic remodeling in IS setting, together with their cognitive and behavioral associations, combined with possible therapeutic implications will form the crux of this review.

Graphical Abstract

Introduction

The maintenance of cellular redox homeostasis is entrenched in integrating oxidative and reductive processes. The crucial components of cellular redox machinery are universal coenzymes β-nicotinamide adenine dinucleotide (NAD⁺) and flavin adenine nucleotide (FAD) (Cantó and Menzies 2015). As a cofactor, NAD⁺ is vital in generation of adenosine tri phosphate (ATP), the primary metabolic energy currency of cell, during glycolysis, tricarboxylic acid cycle and mitochondrial oxidative phosphorylation. The ubiquity of NAD⁺ and FAD in maintaining redox homeostasis is evidenced by a sufficiently large number of reduction reactions in central metabolism powered by their utilization. These biochemical reactions are integral to cellular bioenergetics, maintenance of mitochondrial homeostasis and genomic stability, adaptive stress responses and cell survival (Walsh and Wencewicz 2013; Cantó and Menzies 2015; Chini et al. 2021). Understandably, any metabolic derangement which limits the availability of these metabolites could adversely affect cellular metabolic efficiency. This feature is canonical to the metabolic crisis occurring during cerebral ischemia/reperfusion (CI/R) that culminates in cerebral infarction or ischemic stroke (IS) (Sims and Muyderman 2010; Barile et al. 2013; Pehar et al. 2018; Yang JL et al. 2018; Lautrup et al. 2019; Xie N et al. 2020; Berndt et al. 2020; Tolomeo et al. 2021; Zapata-Pérez et al. 2021).

CI/R is characterized by a blockade in the middle cerebral artery (MCA), consequent of which oxygen and glucose supply is affected. In this context, attendant perturbations in the cellular environment due to glutamate-induced excitotoxicity, involve a steep decline in ATP/ADP ratio due to electron transport chain (ETC) and ion-channels (like potassium (ATP)) dysfunction. Further, massive energy failure with a concomitant dysregulation of cellular redox homeostasis occurs (Sims and Muyderman 2010; Yang JL et al. 2018). The pathological sequelae arising from a precipitous decline in the levels of energy metabolites, ATP generation and deranged cofactor ratios include calcium dyshomeostasis, oxidative stress, mitochondrial dysfunction, impaired mitochondrial biogenesis and subsequent neuronal death (Sims and Muyderman 2010; Yang JL et al. 2018). Further, the deleterious effects of reperfusion on brain energy metabolism are increasingly manifested in post-stroke epoch, that adversely impact homeostatic signaling in glio-neurovascular unit (Pehar et al. 2018). CI/R-induced damage to neuronal networks affects sensation, movement and cognition (Calabresi et al. 2003). Disruption of synaptic plasticity processes and circuitry occurs, that potentiates abnormal neural dynamics in different brain networks (Calabresi et al. 2003; Das and Rajanikant 2018).

NAD⁺ and FAD also constitute two crucial epigenetic enzyme-required metabolites. CI/R is associated with significant alterations in the availability of these cofactors/substrates that linearly affect epigenetic landscape, gene expression patterns and varied cellular processes thereof. NAD⁺ acts as a substrate for various enzymes like class III histone deacetylases (HDACs) or sirtuins (SIRTs), poly(ADP-ribose)polymerases (PARPs), ADP ribosyl-cyclases (CD38/CD157), and a NAD-glycohydrolase viz., sterile alpha and TIR motif-containing 1 (SARM1) (Pehar et al. 2018; Xie N et al. 2020). The coordinate action of these enzymes in a physiological or a pathological setting like IS, increasingly relies on subcellular compartmentation and metabolic availability of NAD⁺ for their activation (Cantó and Menzies 2015; Lautrup et al. 2019; Zapata-Pérez et al. 2021). Similar to NAD⁺, FAD regulates the activity of lysine-specific histone demethylases (LSDs), that are implicated in intermediary metabolism including Krebs cycle and β-oxidation of fatty acids (Forneris et al. 2005; Hino et al. 2012). SIRTs, PARPs and LSD1 introduce chromatin modifications through histone post-translational modifications (PTMs) viz., deacetylation, ADP-ribosylation and demethylation which, in part, could be considered as adaptive regimens of stress response in an altered metabolic setting like IS (Kim et al. 2004; Forneris et al. 2005; Jing and Lin 2015; Narne et al. 2017a; She et al. 2017; Khoury et al. 2018). In addition, SIRTs and PARP-1 through their site-specific modifications of non-histone proteins, could profoundly influence their functioning in varied cellular metabolic scenarios (Jing and Lin 2015). These could possibly restore cellular homeostasis and aid in synaptic plasticity, long-term potentiation (LTP) and axonal sprouting in ischemic penumbra, and also facilitate neurogenesis in post stroke epoch.

A deeper understanding of the role of NAD⁺ and FAD in IS etiology would plausibly enable devising of therapeutic strategies beyond the usage of tissue plasminogen activator (tPA) and aid in ameliorating post-stroke sequelae. In the event of IS, tPA (alteplase, reteplase, and tenecteplase) administration within the critical window of therapeutic opportunity enables thrombolysis. This involves dissolution of fibrin meshes formed by aggregation of activated macrophages, as tPA functions as a serine (Ser) protease (Saver et al. 2013). tPA bound to fibrin activates the fibrin-bound zymogen plasminogen by cleavage at Arginine561-Valine562 peptide, to form endogenously fibrinolytic plasmin, that subsequently disrupts structural support (fibrin) of the clot. Nevertheless, usage of tPA is plagued by narrow treatment window (patient needs to be presented within 4.5 h of IS symptoms onset). Also, contraindications and risk of complications like hemorrhagic transformation, intracranial hemorrhage and mortality following delayed recanalization, limit their eligibility for tPA administration (3–5%) (Saver et al. 2013). While the utility of NAD⁺ and FAD in various metabolic scenarios is vast enough, in the current review we would limit our discussion to their pathophysiological relevance to IS and underlying epigenetic links. A particular emphasis will be laid on epigenetic aspects governing cognitive features and behavioral outcomes, together with pertinent therapeutic possibilities.

NAD⁺ Metabolic Machinery in Ischemic Stroke—A Bevy of Therapeutic Options

NAD⁺ biosynthesis involves two principal enzymes viz., nicotinamide phosphoribosyltransferase (NAMPT) and nicotinamide mononucleotide adenylyl transferase (NMNAT). NAMPT is the principal rate-limiting enzyme in the salvage pathway of mammalian NAD⁺ biosynthesis and catalyzes conversion of nicotinamide (NAM) or nicotinic acid (NA) to NAD⁺ precursor, nicotinamide mononucleotide (NMN) (Cantó and Menzies 2015; Chen et al. 2015; Garten et al. 2015). NAMPT, occurring intra (iNAMPT) and extracellularly (eNAMPT), is a systemic endogenous neuroprotective molecule that maintains the pace of NAD metabolism through NMN biosynthesis. Low levels of iNAMPT in neurons makes them critical ‘frailty points’, as they depend on circulating NMN for adequate NAD⁺ biosynthesis (Pehar et al. 2018; Lautrup et al. 2019). Also, perturbations in NAMPT-mediated NAD⁺ biosynthesis impacts metabolism. This becomes evident from a significant reduction in iNAMPT levels in neurological condition like amyotrophic lateral sclerosis, and in phenotypes associated with hyperlipidemia and aging (Imai and Yoshino 2013). Several studies advocated NAMPT as a therapeutic target for various debilitating neurologies (Chen et al. 2015; Verdin 2015). As regards IS, NAMPT confers cerebro-protection during acute phase and promotes vascular repair and neurogenesis in the chronic phase (Wang and Miao 2019) (Fig. 1a) NAMPT-mediated NAD⁺ generation restores adult neurogenesis by promoting neural stem cell (NSC)/neural progenitor cell (NPC) growth, differentiation, self-renewal and oligodendrocyte formation in vivo and in vitro (Zhao et al. 2015) (Table 1). In line with this, P7C3 family of pro-neurogenic compounds functioning as NAMPT activators are proposed to counteract IS-induced long-term disability resulting from cognitive and behavioral deficits, as they promote adult neurogenesis in hippocampus and olfactory lobes (Wang P et al. 2015; Wang SN et al. 2016; Wang and Miao 2019) (Fig. 1a) (Table 1). In particular, P7C3-A20 has been shown to decrease cortical and hippocampal atrophy, with a corresponding increase in net magnitude of neurogenesis in subventricular zone (SVZ) and hippocampal dentate gyrus (DG) subgranular zones (SGZ), through an increased cellular NAD flux (Loris et al. 2017) (Fig. 1a) (Table 1).

Fig. 1.

a, b Nicotinamide adenine dinucleotide (NAD⁺) metabolism-based pharmacological targeting of cerebral ischemia/reperfusion (CI/R): a Pathological sequelae of lowered activity of inducible nicotinamide phosphoribosyltransferase (iNAMPT) and extracellular nicotinamide phosphoribosyltransferase (eNAMPT) on cerebrovascular outcomes in cerebral ischemia/reperfusion (CI/R) milieu. The salutary effects of NAMPT-based pharmacological intervention on CI/R sequelae are also presented. b The beneficial effects of nicotinamide mononucleotide (NMN), NADH and NAD⁺ administration on CI/R pathology in cerebral milieu (Refer “NAD⁺ Metabolic Machinery in Ischemic Stroke—A Bevy of Therapeutic Options” section in text for more details) (tPA tissue plasminogen activator)

Table 1.

Studies outlining the salutary effects of NAD⁺-restoration, SIRTs activation and PARP-1 inhibition in various pre-clinical models in cerebral ischemia/reperfusion setting

Therapeutic agent/modality	Experimental setting In vitro/in vivo	Proposed mechanisms and effects of intervention/supplementation in tested models	References
NAD+ restoration
NAMPT overexpression	In vivo: Neonatal hypoxia–ischemia (HI) brain injury in wild-type (WT) mice and those overexpressing Nmnat1 in cytoplasm (cytNmnat1-Tg mice)	↓Excitotoxicity-induced, caspase-independent injury to neuronal processes and cell bodies	Verghese et al. (2011)
Resveratrol or ischemic preconditioning (IPC)	In vivo: Intraperitoneal injection (i.p.) 0.5 mg/kg of Tat-conjugated ψɛRACK (PKCɛ agonist) into male Sprague Dawley (SD) rat cortex In vitro: oxygen–glucose deprivation (OGD) in primary neuronal-glial cortical cultures from SD rats	↑mitochondrial pools of Nampt and NAD ↑AMPK activation in vitro and in vivo ↑PKCɛ-mediated ischemic neuroprotection	Morris-Blanco et al. (2014)
NAD+ supplementation	In vitro: Glutamate excitotoxicity model of primary cultured cortical neurons	↓Apoptotic neuronal death, and apoptotic inducing factor translocation following glutamate-induced excitotoxicity ↓Glutamate-induced mitochondrial fragmentation ↓Impairment of mitochondrial biogenesis ↓Mitochondrial membrane potential depolarization and NADH redistribution	Wang et al. (2014)
NAMPT overexpression/ NMN and NAD supplementation	In vivo: Nampt-transgenic mice and ΔNampt (H247A mutant-Nampt transgenic mice) In vitro: Cultured neural stem cell (NSC) neurospheres	↑ Brain NAD+ levels ↑NSC/neural progenitor cell (NPC) growth (↑SIRT1, SIRT2), differentiation, self-renewal and oligodendrocyte formation ↑ Post-ischemic neurogenesis	Zhao et al. (2015)
NAMPT overexpression	In vivo: NAMPT‐transgenic and H247A dominant negative NAMPT‐transgenic mice subjected to middle cerebral artery occlusion/reperfusion (MCAO/R)	Improved cerebral blood flow (CBF) recovery by recruitment and incorporation of endothelial progenitor cells in cerebral arterioles ↑ Post-stroke neo-angiogenesis and neurogenesis	Wang P et al. (2015)
P7C3-A20 (NAMPT stimulator)	In vivo: Transient MCAO in rats followed by twice daily injection of P7C3-A20 or vehicle for 7 days	↓ Cortical and hippocampal atrophy ↑ Neurogenesis in subventricular zone and hippocampal dentate gyrus subgranular zone ↓ Neurodegeneration Chronic behavioral improvement	Loris et al. (2017)
FK866 (NAMPT inhibitor)	In vivo: MCAO/R reperfusion model of CI/R injury. Intracerebroventricular administration of FK866 30 min before MCAO In vitro: OGD/reperfusion (OGD/R) in cortical neuron-glia mix-cultures	↓Microgliosis and astrogliosis in vivo and in vitro → ↓microglial activation → ↓NF-κB → ↓TNF-α, NAMPT, IL-6 → ↓cellular inflammation ↓neurological dysfunction, infarct volume, neuronal loss 14 days after MCAO/R	Chen et al. (2017)
NMN administration	In vivo: tPA infusion and NMN administration (i.p. 300 mg·kg-1) following MCAO in CD1 mice	↓ tPA-induced molecular alterations ↓BBB permeability/↑tight-junction (TJ) proteins/ ↓MMP9 and MMP2	Wei et al. (2017)
Combined administration of NADPH and NAD+	In vivo: NAD+ (12.5–50 mg/kg) /NADPH (2.5 or 7.5 mg/kg) administration in MCAO mice 2 h after reperfusion In vitro: OGD/R in primary neuronal cultures. Treatment with NAD+ (15 mmol/L)/NADPH (10 μmol/L)	↑ ATP ↓ROS-induced oxidative damage ↓ Long-term mortality Improved functional recovery Protracted therapeutic window of NAD+	Huang et al. (2018)
SIRTUINS activation
Icarrin (ICA)	In vivo: MCAO mouse model treated with ICA/saline In vitro: OGD in primary cortical neurons in the presence of ICA or SIRT1 inhibitor III or PGC-1alpha siRNA	↑SIRT1 and PGC-1α in cortex → neuroprotection (improved neurological scores, ↓infarct size/ brain edema)	Zhu et al. (2010)
NAMPT overexpression and knockdown (lentivirus-mediated)	In vivo: AMP-activated kinase-α2 (AMPKα2) and SIRT1 knockout mice	↑Nampt → ↑NMN → SIRT-1-mediated serine/threonine kinase 11 (LKB1) deacetylation → ↑AMPK → ↑neuronal survival → ↓ischemia-induced cerebral injuries	Wang P et al. (2011)
Citicholine (CDP-choline)	In vivo: Photothrombotic stroke in Wistar rats. Administering citicoline 100 mg/kg or vehicle for 10 consecutive days (initiating 24 h following ischemia induction). Sensorimotor testing following a training period at days 1, 10, 21, and 28 post- stroke	↑SIRT1 → neuroprotection, neuro-regeneration ↑ neurogenesis in DG, SVZ, and peri-infarct area Excitation in perilesional cortex → improved neurological outcomes	Diederich et al. (2012)
Combination of HDAC 1–3 inhibitor entinostat (MS-275) and resveratrol	In vivo: Mouse model of transient MCAO; MS-275 (20 μg/kg and 200 μg/kg); resveratrol (6800 μg/kg) individually; MS-275 (2 μg/kg) and resveratrol (68 μg/kg) synergistically (7 h after stroke onset) In vitro: Primary cortical neurons exposed to OGD	Combined use of MS-275 and resveratrol →  ↓infarct volume and neurological deficits Recovery of optimal histone H3 acetylation MS-275 → ↑ total RelA acetylation; resveratrol → ↓RelA K310 acetylation → ↑AMPK-SIRT1 pathway Restoration of RelA acetylation → synergistic neuroprotection in OGD-exposed neurons	Lanzillotta et al. (2012)
NAMPT overexpression	In vivo: MCAO(2 h) in rats In vitro: OGD in cultured cortical neurons NAMPT overexpression or knock-down using lentivirus-mediated gene transfer	↑SIRT1 → regulation of TSC2-mTOR-S6K1 signaling pathway → ↑autophagy (LC3 puncta immunochemistry staining, LC3-II/beclin-1 expression and number of autophagosomes) → ↑ neuronal survival in ischemic brain	Wang P et al. (2012)
Activator 3 (SIRT1 activator)	In vivo: Permanent focal ischemia of WT and Sirt1−/− mice; i.p. activator-3 (10 mg/kg), or sirtinol (SIRT1 inhibitor, 10 mg/kg) for 10 min, 24 h, and 40 h after ischemia	↑SIRT1 in ipsilesional mouse brain cortical neurons ↓infarct volume (larger infarct volumes in Sirt1−/− mice) Sirtinol → ↑ischemic injury SIRT1 inhibition/deletion → ↑p53/NF- κB (p65) acetylation	Hernández-Jiménez et al. (2013)
Citicholine Resveratrol	In vivo: Permanent focal ischemia in Fischer rats and Sirt1−/− mice. Treatment with CDP-choline (0.2 or 2 g/kg), sirtinol (a SIRT1 inhibitor; 10 mg/kg), and resveratrol (SIRT1 activator; 2.5 mg/kg)	↑SIRT1 expression in brain → reduced infarct volume → neuroprotection Sirtinol/Sirt1−/− → blockade of citicholine-induced reduction in infarct volume Resveratrol → synergistic neuroprotective effect with citicholine	Hurtado et al. (2013)
Alpha-lipoic acid (ALA)	In vivo: Permanent MCAO model. Treatment with ALA (i.p. 50 mg/kg) 30 min prior to ischemia	↑ SIRT1, PGC-1α and SOD activity ↓ Neurological deficits, decreased infarct volume and brain edema	Fu et al. (2014)
SIRT1 overexpression	In vivo: Bilateral common carotid artery stenosis of Sirt1-overexpressing (Sirt1-Tg) mice and their WT littermates	Sirt1-Tg mice: Rescue from memory impairment and histological changes ensuing from cerebral hypoperfusion → preserved CBF ↓Acetylation of brain endothelial nitric oxide synthase post cerebral hypoperfusion → ↓irregularities in vascular endothelia and tight junction openings	Hattori et al. (2014)
Minocycline	In vivo: SD rats under hypobaric hypoxia administered minocycline for 1 h In vitro: Human brain microvascular endothelial cells (HBMECs) using Trans Epithelial Electric Resistance	↓HIF-1α/ MMP-2, MMP-9/ VEGF ↑TJs (ZO-1, claudin-5 and occludin) in HBMECs after hypoxia Reversal of hypoxia-induced PHD-2/SIRT3 reduction → ↑BBB integrity	Yang et al. (2015)
17β-estradiol	In vivo: CI rat model; to explore the effects of estrogen in CI damage, rats were subjected to subcutaneous injection of 17β-estradiol after ovariectomy (OVX) In vitro: OGD in neurons and treatment with estrogen-supplementation	Estrogen pretreatment: ↓CI-induced injury ↑SIRT1 → ↑AMPK activation through estrogen receptor α → ↓neuronal apoptosis → ↑neuroprotection in vitro	Guo et al. (2017)
Hyperbaric oxygen (HBO)	In vivo: MCAO (2-h) followed by hyperglycemia induction (i.p. 50% dextrose (6 mL/kg) at onset of reperfusion. Rats exposed to HBO at 2 atmospheres absolutes for 1 h. Interventions with ATP synthase inhibitor oligomycin A, NAMPT inhibitor FK866, SIRT1 siRNA	HBO treatment: ↑ATP, NAD+ → ↑SIRT1 → attenuation of hemorrhagic transformation, brain infarction; improvement of neurological function (↓hyperglycemia-induced BBB disruption and neurological defcits)	Hu et al. (2017)
SIRT3 overexpression	In vivo: CI/R injury model- transient MCAO In vitro: Neuro-2A cells subjected HR injury Treatment with recombinant adenovirus-Sirt3	↑SIRT3 → ↓β-catenin phosphorylation → Wnt/β-catenin pathway → ↓mitochondrial fission and apoptosis → ↑pro-survival signals in CI/R-insulted neurons → ↓caspase-9-dependent cell death	Zhao et al. (2018)
Melatonin	In vivo: tMCAO model (male C57/BL6 mice) of CI/R injury Injecting melatonin (i.p. 20 mg/kg) after ischemia and before reperfusion	↑SIRT3 → ↓neurological dysfunction and cell apoptosis	Liu et al. (2019)
Genipin (UCP2-specific inhibitor)	In vitro: MCAO/R model of CI/R injury in male C57BL/6 mice with/without genipin	Genipin: Regulation of UCP2-SIRT3 signaling pathway to modulate energy metabolism and oxidative stress → ↓brain damage due to CI/R injury	Zhao et al. (2019)
C1q/tumor necrosis factor-related protein-3 (CTRP3)	In vitro: OGD/R in hippocampal neuronal cells (HPPNCs)	↑Mitochondrial biogenesis by activation of AMPK/SIRT1-PGC-1α pathway ↑ HPPNCs viability and ↓apoptosis	Gao et al. (2020)
Luteolin	In vivo: MCAO rat model of CI/R injury (i.p. luteolin 15/30/60 mg/kg after 1.5 h of ischemia)	↓infarct volume, cerebral edema, morphological changes in the cortex and hippocampus, neuronal apoptosis ↑SIRT3 → AMPK-mTOR pathway ↑Mn-SOD activity → ↓ROS → ↑mitochondrial function → ↑neuronal survival	Liu et al. (2020)
Notoginseng leaf triterpenes (PNGL)	In vitro: OGD/R in SH-SY5Y cells (HI cell model in vitro) (also treated with NAMPT inhibitor-FK866 to verify effect of PNGL)	PNGL regulates downstream SIRT1/2-Foxo3a and SIRT1/3-MnSOD/PGC-1α pathways ↑NAMPT in ischemic regions and OGD/R-induced SH-SY5Y cells Protective effects of PNGL plausibly mediated by NAMPT in vitro ↓Ischemia-induced injury ↑neuronal mitochondrial homeostasis → ↓ energy metabolism dysfunction ↓Neuronal loss and apoptosis ↑Neuronal survival under IH	Xie W et al. (2020)
Lin28a (RNA-binding protein) up-regulated expression	In vivo: CI/R injury model in mice	SIRT3 regulation → AMPK-mTOR pathway ↑LC3-II levels in nerve cells → ↑autophagy → ↓nerve cell apoptosis and nerve cell injury	Chen D et al. (2021)
Mangiferin (MGF), a natural C-glucosyl xanthone polyhydroxy polyphenol	In vitro: CI/R injury (HR)	↑SIRT1, PGC-1α, Nrf2, NQO1, HO-1, NRF1, UCP2, and Bcl2 ↑Activation of SIRT1/PGC-1α signaling ↓Decreased lactate dehydrogenase release and ROS generation ↑Neuroprotection against HR-induced injury ↑Cell viability and restored cell morphology	Chen M et al. (2021)
Circular RNA: circ_0000296 overexpression	In vivo: Hippocampus treated with chronic CI In vitro: OGD in HT22 cells	circ_0000296 specifically binds to miR-194-5p which is bound to 3'UTR region of Runx3 mRNA Runx3 binds to Sirt1 promoter region of promoting its transcription ↑circ_0000296 → ↓miR-194-5p/Runx3 → ↑Runx3 /Sirt1 → ↓neuronal apoptosis in CI brain	Huang K et al. (2021)
Ginsenoside Rc (SIRT1 activator)	In vitro: I/R model in PC12 cells	↑SIRT1 activity by interaction of Rc on SIRT1 residue sites ↑ETC complex II-IV levels (↑ATP, glucose uptake, hexokinase I/II, mitochondrial pyruvate carrier I/II) → ↑energy metabolism ↑SIRT1-↓PGC1α acetylation in I/R →  ↑Stimulation of PGC-1α pathway → ↑mitochondrial biogenesis in neurons	Huang Q et al. (2021;
Mild/intense exercise postconditioning (M/IPostE)	In vivo: Rats subjected to MCAO followed by either resting/ M/IPostE (24 h after reperfusion)	Both PostE groups: ↓ Brain infarct volumes and edema, neurological deficits, ROS production, and apoptotic cell death ↑SIRT1/↓Endoplasmic reticulum stress (ERS) proteins/ ↓ROS → ↑neuroprotection MPostE: ↑Bcl-2/BAX/caspase-3 and Bcl-2/BAX ratio /↓↓GRP78, caspase-12, IRE1α, CHOP (ERS proteins)	Li F et al. (2021)
Limb ischemic postconditioning	In vivo: MCAO/R model- CI/R injury in male SD rats	↑SIRT1 and PGC-1α; ↑NRF-1; ↓CytoC1 in cerebral ischemic tissue → improved mitochondrial function ↑SOD activity; ↓malondialdehyde (MDA) levels in brain tissue ↓Cerebral infarct, neuronal apoptosis, neurological deficits → ↓brain damage	Li L et al. (2021)
Melatonin	In vivo: MCAO model of CI/R injury in streptozotocin-induced diabetic (male C57BL/6 J) mice Melatonin injection (10 mg/kg i.p., immediately after induction of ischemia and at reperfusion onset) In vitro: Murine hippocampal neuron cell line HT22 pre-treated with melatonin (100 μM) for 4 h	Stimulation of Akt and SIRT3/SOD2 signaling → ↑mitochondrial biogenesis-related TFs ↓mitochondrial swelling, ROS generation, and cytoplasmic cytochrome C release ↑mitochondrial antioxidant enzymes, ATP and Δψm ↓neurological deficits, cerebral infarct volume, brain edema	Liu L et al. (2021)
Long non-coding RNA (lncRNA) MALAT1 overexpression	In vivo: MCAO model of CI/R injury In vitro: OGD/R-induced PC12 cells	↓OGD/R-induced cell necrosis and apoptosis → ↑cell proliferation ↓TNF-α, IL-6, IL-1β, ROS/MDA; ↑SOD and CAT in OGD/R-injured PC12 cells ↑SIRT1/ ↓miR-142-3p → ↓neurological impairment and cognitive dysfunction in CI/R mice	Meng et al. (2021)
Astragaloside IV (AS-IV)	In vivo: MCAO/R model- CI/R injury in male SD rats. Treatment with AS-IV (IP) (20 mg/(kgd))	↑SIRT1 and ↓hyperacetylation and hyperphosphorylation of microtubule-associated protein tau (MAPT)/SIRT1/MAPT pathway → Neuroprotection ↓ CI area/ neurological deficits	Shi et al. (2021)
lncRNA- Small nucleolar RNA host gene 8 (Snhg8) overexpression	In vivo: MCAO model of CI/R injury In vitro: OGD-treated primary microglia and BMECs	↑Snhg8 → miR-425-5p/SIRT1/NF-κB axis → ↓CI-induced microglial inflammation and BBB damage	Tian et al. (2021)
SIRT3 overexpression	In vivo: permanent MCAO model in WT and Sirt3 KO mice In vitro: OGD in cultured primary astrocytes	↑SIRT3 → ↓HIF-1α signaling post-ischemia → regulated VEGF expression → ↓neuroinflammation → ↑BBB integrity → ↓neuronal apoptosis and neurological deficits post-ischemia	Yang X et al. (2021)
Hyperbaric oxygen (HBO)	In vitro: OGD/R model using mice N2a cells to simulate in vitro brain I/R injury	HBO treatment: ↑SIRT1 → ↑ HMGB1 deacetylation → ↓neuroinflammation → ↓ischemic brain injury	Zhao PC et al. (2021)
Betulinic acid (BA)	In vivo: cerebral IS model;(1-h MCAO-24-h reperfusion) In vitro: OGD/R of PC12 cells	BA treatment: ↑SIRT1/FoxO1 pathway → ↓autophagy/↓oxidative stress → ↓cerebral injury	Zhao Y et al. (2021)
Notoginsenoside R1	In vivo: MCAO/R model- CI/R injury in male SD rats. R1 (i.p.) dministered immediately after ischemia induction In vitro: Pre-treatment of HBMECs with different concentrations of R1 for 12 h before OGD/R exposure	Improved mitochondrial energy metabolism Restoration of CBF Activation of NAMPT-NAD + -SIRT1 cascade/ modulation of Notch signaling and VEGFR-2 → ↑post-stroke angiogenesis (migration, proliferation and tube formation of HBMECs following incubation with 12.5–50 μM R1)	Zhu et al. (2021)
PARP-1 inhibition
Iduna	In vivo: MCAO model of CI/R injury In vitro: Cortical neuronal culture subjected to N-methyl D-aspartate -induced excitotoxicity	Iduna binds poly (ADP)-ribose polymer and prevents apoptosis-inducing factor (AIF) translocation from mitochondria to nucleus thereby promoting neuronal survival	Andrabi et al. (2011)
PJ34 (PARP-1 inhibitor)	In vivo: tMCAO model of CI/R in male and female C57BL/6 mice (i.p. PJ34 30 mg/kg, 48 h after MCAO) In vitro: Primary microglial culture treated with PARP-1 inhibitor DPQ (25 μM)	↓iNOS, MMP9 → ↓neuroinflammation → improved behavioral outcomes in male MCAO mice/ ↓microglial activation (males and females in vivo and in vitro)	Chen et al. (2020)
Baicalein	In vivo: MCAO model of CI/R injury (baicalein: 100 mg/kg for 7 days) In vitro: OGD in SH-SY5Y cells	↓pro-inflammatory cytokines ↓PARP-1 activation → ↓nuclear translocation of AIF and macrophage migration inhibitory factor → ↓cerebral infarct volume ↓apoptosis/ oxidative stress → ↑neuronal survival Restoration of mitochondrial function and mitochondrial membrane potential in OGD cells	Li et al. (2020)
PJ34	In vivo: MCAO model of CI/R injury (i.p. PJ34: 3, 6, or 12 mg/kg) injections of PJ34 before and after MCAO	↓Cyclooxygenase 2, inducible nitric oxide synthase in cerebral tissues → ↓neuroinflammation → ↓CI/R-induced injury (dose-dependent) → ↑neurological performance	Jiao and Li (2021)
14,15-EET	In vivo: MCAO model of CI/R injury (14,15-EET:100 nM) In vitro: OGD/R model of I/R injury in primary neuronal cultures (14,15-EET: 20 nM)	↑GSH-Px, SOD, heme oxygenase-1 in cortical neurons after ischemia and reperfusion → ↓ROS → ↓parthanatos	Zhao H et al. (2021)

Post-ischemic neurogenesis is a notable feature of NAMPT transgenic mice, nevertheless with no protection against acute brain infarction and neuronal deficits (Zhao et al. 2015). This in turn correlates with improved neural functional recovery, increased survival rate and accelerated body weight gain after middle cerebral artery occlusion (MCAO) (Table 1). The significance of NAMPT in preserving cognitive function further becomes evident from the histological sequelae viz., hippocampal and cortical atrophy, astrogliosis, microgliosis and abnormal CA1 dendritic morphology, in Ca²⁺/calmodulin-dependent protein kinase II (CAMKII)-NAMPT^−/− mice that lack NAMPT in forebrain excitatory neurons (Stein et al. 2014). These correlate with altered hippocampal connectivity and abnormal behavior including hyperactivity, motor skill defects and memory impairment with reduced anxiety. Nevertheless, sensory processes and LTP induction in Schaffer collateral synapses in hippocampal CA1 are not impaired. In a similar vein, inducible or conditional knock-out (CKO) of iNAMPT in adult projection neurons result in progressive loss of weight, hypothermia, motor neuron degeneration, motor function deficits, paralysis and death (Wang et al. 2017). Motor neurons are affected due to mitochondrial dysfunction, muscle fiber type conversion and atrophy with defective synaptic activity at neuromuscular junctions. Herein, therapeutic intervention with NMN is of considerable interest, as it alleviates motor function deficits and prolongs life span. In addition, resveratrol administration or ischemic preconditioning is also beneficial as an increase in mitochondrial NAMPT, and hence NAD⁺ levels can be achieved. This essentially involves protein kinase C (PKC)-ε-mediated AMP kinase (AMPK) activation that induces hypoxia inducible factor-1α (HIF-1α) and NAMPT upregulation (Morris-Blanco et al. 2014) (Fig. 1a) (Table 1).

While the role of iNAMPT in CI/R has been adequately documented, the lesser known eNAMPT has been shown to be neuroprotective as it is secreted following CI and protects brain white matter from ischemic injury (Jing et al. 2014). White matter constitutes about half the infarct volume, and ischemia-induced white matter injury propels cognitive dysfunction and sensorimotor impairments among others (Fern et al. 2014; Wang Y et al. 2016a). Therein, NAMPT-mediated fortification of white matter in addition to gray matter, could plausibly ameliorate cognitive domain performance and hence neurological outcomes in post-stroke epoch (Stebbins et al. 2008) (Fig. 1a). Contrastingly, NAMPT activation has been shown to exacerbate neuronal injury during acute phase of CI due to microglial activation, following nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) upregulation (Chen et al. 2017). Also, an increased synthesis and secretion of NAMPT, tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6) potentiates inflammatory damage. This observation is in line with recently proposed role of eNAMPT as an extracellular endogenous mediator of inflammation in the pathogenesis of several metabolic and inflammatory disorders (Audrito et al. 2020). Disruption with FK866, a potent NAMPT inhibitor alleviates harmful effects of NAMPT activity on microglial activation, as observed in Iba1-positive microglia or macrophages in ischemic core 14 days after MCAO/reperfusion (Chen et al. 2017) (Fig. 1a) (Table 1).

NAD⁺ is generated from NMN by NMNAT that exists in nuclear (NMNAT1), cytosolic (NMNAT2) and mitochondrial (NMNAT3) isoforms. NMNAT1 is the most abundant isoform and exhibits maximal specific activity for NAD⁺ synthesis and pyrophosphorylysis (Berger et al. 2005). In neonatal hypoxia-ischemic brain, NMNAT1 seemingly protects neuronal cell bodies and processes in cortex, hippocampus and striatum against excitotoxicity-inducing acute and chronic neurodegenerative insults, in caspase-3-independent manner (Verghese et al. 2011) (Table 1). Empirical evidence accords a general neuroprotective function to NMNATs and also a chaperoning function, owing to their structural similarity to known molecular chaperones like heat shock proteins. Another shared feature includes invoking a neuroprotective modality involving proteasome-mediated pathway in debilitating neurologies like spinocerebellar ataxia, and those involving accumulation of polyglutamine expanded tract-containing proteins (Zhai et al. 2008). Further, neuroprotective implications of a distinct chimeric protein with intact NMNAT1 activity against axon degeneration (a prominent feature of CI/R-induced damage), will be discussed in detail in the next few sections.

Given the subcellular compartmentation of NAD⁺ salvaging machinery, inclusive of NAMPT, NMNAT isoforms and degrading enzymes, it is reasonable to assume that the overall cellular NAD⁺/NADH (NAD + hydrogen) ratio will be affected by transient changes in their activities, in disparate compartments of cells during metabolic stress (Ryu et al. 2018; Cambronne and Kraus. 2020). As a corollary, the NAD flux is compartment-specific and does not summarily affect the ‘general pyridine nucleotide pool’ (Berger et al. 2005; Ryu et al. 2018). Given this consideration, global integration of NAD⁺ levels following a breach in compartmentalized pool integrity could plausibly occur under conditions of cellular stress and insults like CI/R, the consequences of which are yet to be explored (Pehar et al. 2018; Lautrup et al. 2019; Cambronne and Kraus. 2020). Further, the therapeutic applicability of NAD⁺ intermediates or precursors like nicotinamide riboside (NR), NA and NMN in alleviating various pathophysiologies is expanding enormously (Park et al. 2016; Pehar et al. 2018; Yoshino et al. 2018; Braidy et al. 2018). As regards IS, NMN reportedly precludes tPA-induced alterations following its administration after 5-h therapeutic window, thereby preventing transition into hemorrhagic transformation (Wei et al. 2017). NMN administration prevents blood brain barrier (BBB) breakdown resulting from reduced expression of tight junction proteins and increased expression of matrix metalloproteinases (MMPs) 9 and 2 respectively (Fig. 1b) (Table 1). Extension of therapeutic window in IS could also be achieved by concerted action of antioxidative NADPH (reduced form of NAD phosphate) and NAD⁺ that alleviates CI/R-induced metabolic stress (Huang et al. 2018) (Fig. 1b) (Table 1). The salutary effects include long term mortality reduction with improved functional recovery. This is attributed to energy sparing, which becomes evident from increased ATP levels and reduced oxidative stress, with a concurrent inhibition of apoptotic and necroptotic signaling pathways (Table 1). Further, the primacy of NAD⁺ in preserving mitochondrial biogenesis and integrity is illustrated by primary cortical neuron cultures treated with excitotoxic glutamate (Wang et al. 2014). Administration of exogenous NAD⁺ rescues them from apoptotic death by abrogating mitochondrial fragmentation, decreasing mitochondrial membrane potential (Δψm) depolarization, NADH redistribution and apoptosis-inducing factor (AIF) translocation (Wang et al. 2014) (Fig. 1b) (Table 1). Taken together, an array of regulatory processes is impacted by NAD⁺ and its precursors together with corresponding enzymes. Hence, they can be effectively targeted to alleviate the motor, cognitive and behavioral deficits in CI/R setting, which will be discussed further.

Axonal Degeneration and NMNATs: A Promising Therapeutic Premise in Attenuating CI/R Injury

Wallerian Degeneration

As regards NMNATs in neuroprotection, another aspect pertinent to IS is axon degeneration. The damage incurred by nerve cell fibers from acute insults inciting neurodegeneration like trauma and ischemia, triggers a series of stereotyped and synchronized molecular and cellular processes, that collectively describe an active auto-destruction program viz., Wallerian degeneration (Wld) (Fig. 2a). It involves an energetically detrimental end-point such as NAD⁺ and ATP depletion (Wang JT et al. 2012); Conforti et al. 2014). In post-stroke epoch, Wld of lesioned neurons or transsynaptic degeneration contributes to anterograde volume loss, that predictably links with an attendant tissue loss at discrete and widely separated locations along the neuraxis (Thomalla et al. 2005; Hinman 2014; van Niftrik et al. 2021; Zheng et al. 2021). Wld of descending fibre tracts occurs shortly after CI/R injury and involves disintegration of axonal structures, macrophage infiltration, myelin degradation, with eventual fibrosis and fiber tract atrophy (Thomalla et al. 2005; De Vetten et al. 2010). In line with this, motor deficits precisely correlate with Wld in corticospinal tract (CST), and CST injury correlates with cerebral peduncular atrophy in supratentorial unilateral infarction and stroke patients (Yu et al. 2009; Puig et al. 2010; Darwish et al. 2021).

Fig. 2.

a–c Molecular aspects of Wallerian degeneration (Wld) of axons outlining the role of nicotinamide mononucleotide adenylyl transferase (NMNAT) isoforms, PARP-1 and sirtuins. a Wld is an axon-related pathological attribute of neurodegenerative insults like CI/R. It involves axonal structure degeneration from distal regions, primed by NAD⁺ and ATP depletion and proteolytic cascade activation through intra-axonal release of Ca²⁺ from axoplasmic reticulum. The competence of mitochondrial handling of Ca²⁺ is markedly reduced. In this scenario, endogenous NMNATs preserve axonal integrity and prevent axonal loss. Wld slow (WldS), a chimeric protein composed of N-terminal 70 amino acids (N70) of UBE4B (ubiquitin fusion degradation protein 2a) fused to the coding sequence of NMNAT1 imposes a delayed axonal degeneration phenotype. This occurs by augmenting NAD⁺ production that counteracts axonal loss. In addition, NMNAT2 and NMNAT3 also preserve axonal integrity and maintenance through localized NAD⁺ generation. Axonal injury incites NMNAT2 loss and activation of sterile alpha and TIR motif containing 1 (SARM1) leading to NAD⁺ degradation and resultant accumulation of prodegenerative nicotinamide (NMN) that triggers Wld (Refer “Wallerian Degeneration”, “Wld and NMNATs” and “Wld and Mitochondria-Plausible Avenue for Therapeutic Intervention in CI/R Milieu” sections for more details). b As regards PARP-1 and PAR polymers, PARP-1 plausibly precipitates axon degeneration by NAD⁺ depletion through excessive enzymatic activation, auto-PARylation and PARylation of other substrates. Concurrently, ATP depletion also occurs through inhibition of glycolytic cascade and oxidative metabolism. This occurs by interaction of PARylated hexokinase (HK) with voltage-dependent anion channels (VDAC), resulting in its dissociation from mitochondrial membrane thereby impeding the attendant pathways. Overactivation of PARP-1 hinders axonogenesis and neurite outgrowth by impeding SIRT1 and SIRT3 activation. SIRT1 prevents axonal self-destruction in NAD⁺-dependent manner. It can also promote axonal regeneration by regulating phosphatase and tensin homolog (PTEN) activity that in turn impinges on phosphoinositide-3-kinase (PI3K)-Akt-mammalian target of rapamycin (mTOR) axis. This is ostensibly linked to axonal protein synthesis so as to support features of axonal regeneration (Refer “SIRT1 and Axonal Eegeneration—A Plausible Pharmacological Opportunity” section for more details). c PARP-1 and PAR polymers also inhibit axon regeneration by hampering growth cone motility and partaking in axonal growth-inhibitory cascades involving accumulation of growth-inhibiting substrates like myelin-associated glycoprotein (MAG), Nogo-A and chondroitin sulphate proteoglycans (CSPG) (Refer “PARP-1 and Axonal Damage—A Formidable Feature for Pharmacological Intervention” section for more details)

IS-induced axonal injury progresses in three distinct phases with distinct molecular underpinnings (Conforti et al. 2014; Hinman 2014). Consequent to the primary ischemic insult is a precipitous decline in ATP, increased sodium retention and intra-axonal release of Ca²⁺ from axoplasmic reticulum that triggers proteolytic cascades. Loss of axoglial contact and deranged glia-to-axon energy transmission characterize the second phase. A progressive degradation of injured axons that survived initial insult also occurs. Of particular interest, in connection with NAD⁺ metabolism and axon degeneration, is the third phase that involves anterograde Wld and axonal sprouting (that enhances recovery)/degeneration in response to permissive and repulsive cues. It is in this context that endogenous NMNATs act to preserve axonal integrity and prevent Wld-induced axonal loss (Coleman and Freeman 2010). Also, Wld is not singularly a consequence of physical injury of axon like neurite transection, and is also induced under conditions of non-lethal impairment of protein synthesis. As a tangential link, protein translation arrest and calcium dyshomeostasis resulting from escalating endoplasmic reticulum stress in an ischemic brain could plausibly induce Wld in IS, which could be investigated (Nakka et al. 2016).

Wld and NMNATs

In relation with axon rescuing, Wld slow phenotype (WldS), determined by a spontaneous autosomal dominant genetic alteration that generates a chimeric NMNAT1, governs the significant delay in axon degeneration. This variant is composed of N-terminal 70 amino acids (N70) of UBE4B (ubiquitin fusion degradation protein 2a), a ubiquitin-chain assembly factor, fused to tandemly triplicated complete coding sequence of Nmnat1 (Mack et al. 2001). UBE4B/NMNAT1 chimerism associates with an increased NAD⁺ generation, that effectively delays axonal degeneration in transected neurites (Fig. 2a). WldS-mediated axon protection phenotype necessitates convergence of N16-VCP interactions and NMNAT1 activity in vivo (Avery et al. 2009). Further, axonal protection does not seem to follow the norm of nuclear compartmentalization, as disruption of NMNAT1 nuclear localization or redistribution of paradoxically nucleus-localized WldS to cytoplasm does not attenuate axonal protection. Accordingly, misexpression of NMNAT1 in the form of an engineered mutant viz., cytNMNAT in cytoplasm, demonstrates its potential for axon protection by acting from one or more non-nuclear compartments (Beirowski et al. 2009; Sasaki et al. 2009). Structural and functional features of axons and synapses are better preserved in these mutants, as evident from the continuity of distal axon stumps and preserved motor nerve terminals, as compared with native mutant WldS. Corroborating this proposition, augmenting NMNAT1 expression by its viral transduction into injured axons or fusion with N-terminus of β-amyloid precursor protein of AD in putative destinations like axons and synapses produces similar effect (Sasaki et al. 2009; Babetto et al. 2010). These molecular features provide a reasonable premise for investigation of the role of NMNAT1 in axon degeneration in relation with IS pathology, which is open for question.

With respect to other NMNATs, NMNAT3 activity in mitochondrial matrix confers axonal protection phenotype in neurons of transgenic mice overexpressing NMNAT1, NMNAT3 and enzyme-inactive mutant WldS protein (bearing W258A mutation). Concurrent with this, is an increased ATP generation in isolated mitochondria with no alteration in ETC components (Yahata et al. 2009). As regards NMNAT2, it is a labile axonal maintenance factor and a pro-survival signal that associates robustly with axon degeneration (Gilley and Coleman 2010). Loss of NMNAT2 induces Wallerian-like degeneration of uninjured or established axons in peripheral and central nervous system. This precludes the extension of developing axons, an effect that cannot be compensated by NMNAT1 or NMNAT3 (Gilley et al. 2013). NMNAT2 is generated in soma or cell body and is transported along axonal tracts through anterograde transport by hitchhiking onto post-Golgi vesicles (Fig. 2a). Perhaps, the short life of NMNAT2 sets a threshold for NAD⁺ exhaustion to trigger Wld. It also explains the absolute requirement for continuous replenishment of axonal NAD⁺ for axon survival (Gilley and Coleman 2010). It also furthers the consideration that substituting for the loss of endogenous, axonal NMNAT2 is plausibly the central tenet of WldS-mediated axonal protection following axonal injury. Axonal injury or an impeded axonal transport rapidly degrades NMNAT2, consequent of which NAD⁺ levels decline. Following this, prodegenerative NMN that prevents neurite outgrowth, accumulates in axons that triggers axonal damage by Wld (Di Stefano et al. 2015) (Fig. 2a). The prominent player in NMNAT2 loss-dependent axon degeneration is SARM1, that promotes cell-autonomous molecular program of axon degeneration in response to injury and stress (Osterloh et al. 2012). SARM1 potentiates axon defects imposed by NMNAT2 loss, by initiating a local destruction program involving rapid NAD⁺ hydrolysis (Gerdts et al. 2015; Gerdts et al. 2016)]. SARM1 either acts downstream of NMNAT2 loss and NMN accumulation in a linear pathway, or in concert with NMNAT2 loss-dependent signals that converge at a critical step involving calpain activation (Fig. 2a). In view of major implications of NAD⁺ depletion in an ischemic brain and beneficial effects of SARM1 deletion in neuropathologies, it would be of considerable interest to investigate the role of SARM1 in potentiating CI/R-induced injury (Gilley et al. 2015; Henninger et al. 2016; Krauss et al. 2020; Loring and Thompson 2020).

Wld and Mitochondria-Plausible Avenue for Therapeutic Intervention in CI/R Milieu

Axonal mitochondrion is a point of convergence for WldS/NMNAT enzymatic activities, that are imperative for axonal protection and for processes regulating axon degeneration. WldS is expressed in axoplasm and axonal organelles like mitochondria and phagosomes, albeit its maximal expression in nucleus owing to endogenous NMNAT1. This indeed enables WldS to partially rescue the decline in mitochondrial respiration and glycolysis during latent phase of Wld through maintenance of NAD⁺ levels (Godzik and Coleman 2015). The axon-protective effects of mutant cytNMNAT1 against OGD and oxidative stress also depend on mitochondrial transport (Fang et al. 2014). In line with this, depletion of axonal mitochondria features prominently in axotomy and NMNAT downregulation. This could linearly affect synaptic plasticity, as mitochondrial integrity unequivocally links with neuroplasticity (Cheng and Hou 2010). Correspondingly, endogenous axonal trophic activity of WldS or all mammalian NMNATs invariably suppresses injury-induced mitochondrial loss, thereby protecting severed axons from degeneration (Fang et al. 2014). Corroborating this, mitochondrial localization of extra-nuclear WldS proteins is observed in synaptic preparations from mouse brain that augments basal mitochondrial motility in axons (Avery et al. 2012).

Axonal protection is supposedly linked to an increased number of motile mitochondria, as observed in distal segments of peripheral nerve axons subjected to laser axotomy in native WldS, compared to wild type injured axons. This is attributed to enzymatic activity of NMNAT isoforms that attenuate injury-induced Ca²⁺ spike by increasing the competence of mitochondrial handling of Ca²⁺. An increase in NAD⁺ levels correlates with decreased intra-axonal Ca²⁺ levels. This increasingly corresponds with an enhanced intrinsic capacitance for Ca²⁺ or Ca²⁺ buffering in WldS mitochondria (MacAskill and Kittler 2010; Avery et al. 2012) (Fig. 2a). Keeping the cytosolic Ca²⁺ concentration at bay prevents undocking of mitochondria from kinesin-dependent anterograde transport system (Wang and Schwarz 2009). Also, activation of Ca²⁺-dependent proteases that degrade axonal cytoskeletal proteins like spectrin is prevented, thereby preserving structural and functional integrity of axon (Wang JT et al. 2012). Conversely, high axonal Ca²⁺ arising from axoplasmic reticulum, mitochondria and reversal of Na²⁺/Ca²⁺ exchanger following NMNAT2-NAD⁺ breakdown (as in axonal injury) is associated with mitochondrial sequestration in soma, consequent of which mitochondrial motility into the axon is diminished (Fig. 2a). Interspersed with this is the possible role of Miro and Milton, the linker proteins required for mitochondrial anterograde transport (Stowers et al. 2002; Glater et al. 2006). Perturbation in the quantity or functioning of these proteins compromises WldS/NMNAT-mediated axon protection. NMNAT-mediated axon protection is also ascribed to ROS mitigation. NMNAT decreases axonal accumulation of and sensitivity to ROS (which accumulate profoundly in an ischemic brain) as observed in rotenone-treated dorsal root ganglion (DRG) neurons. This, however is not accompanied by neuronal ATP preservation (Press and Milbrandt 2008). In CI/R setting, a consensus on the central role of axonal mitochondria in initiation of axon-self destruction cascade triggered by various culpable factors is yet to be reached. Further, investigations on epigenetic underpinnings in terms of DNA and histone modifications in the paradigms and pathologies discussed above, in the direction of axonal regeneration and restoration of axonal integrity are only in nascency (Mahar and Cavalli 2018). More importantly, Ca²⁺ influx (which is increasingly pertinent to IS in terms of glutamate-induced excitotoxicity), is ostensibly associated with changes in histone acetylation in response to injury (Cho et al. 2013). Also, an array of histone modifiers acting as downstream effectors of signal transduction pathways activated by axon injury have been identified (Mahar and Cavalli 2018). As mitochondrial dysfunction, ROS accumulation and energy dyshomeostasis are canonical to CI/R pathology, it will be compelling to explore the epigenetic links between these cellular aspects, Wld, NMNATs and axonal rescue in CI/R setting.

NAD⁺ Supplementation and Axon Rescue in Wld-Salutary Effects

The primacy of NAD⁺ network in axon rescue following degenerative insults is evident in delayed axonal degeneration following transection. This occurs by axonal NAD⁺ build-up through supplementation of exogenous NAD⁺ or extra copies of NMNAT (Sasaki et al. 2009; Wang JT et al. 2012). Substrates like nicotinic acid mononucleotide (NaMN), NR and NAM potentiate NMNAT activity. NMNAT expression consolidates axonal protective effects of augmented NAD⁺ or nicotinamide substrates (NAM and NMN) in the nucleus and mitochondria (Ali et al. 2013; Gerdts et al. 2016). NMNAT1-responsive steps in axon degeneration reportedly occur in initial phases after axon injury. Hence, it would be pragmatic to administer axon-protective agents like NMNAT1 after injury in a setting like IS. Alternatively, rectifying aberrant features such as rapid NMNAT2 turnover following axonal injury, with exogenous NMNAT2 supplementation or protease inhibition can rescue injured neurites from Wld (Di Stefano et al. 2015; Gerdts et al. 2016). While the efficacy of nucleosides and NAD⁺-synthesizing enzymes in conferring axonal protection is sufficiently advocated, it is as yet unclear as to how these non-drug phosphorylated compounds gain access to cell interior with savior purpose. One school of thought refers to priming of nucleoside riboside kinase 2 (Nrk2) signaling to protect DRG neurons. In line with this, NR vitamin supplementation increases Nrk2 mRNA 20 days after sciatic nerve transection (Tempel et al. 2007). This pathway assumes significance and therapeutic intervention could be conjectured in this direction, as stroke involves transient axonal stress. Alternatively, counteracting axonal buildup of prodegenerative NMN by pharmacological inhibition or genetic deletion of NMN delays Wld. Also, NAMPT inhibition by FK866 partially restores axon outgrowth in NMNAT2 deficient axons (Di Stefano et al. 2015). SARM1 depletion engenders an effect similar to NMN scavenging and is comparable with WldS in conferring axon protection (Gerdts et al. 2015; Sasaki et al. 2016). Steady state flux analysis of NAD⁺ metabolites indicates that over expression of cytNMNAT1 or NMN deamidase can impede axon-destructive action of SARM1. This occurs by preventing injury-induced NAD⁺ consumption, thereby obviating the need for NMN reduction. It also indicates that protection afforded by FK866 is considerably lesser than that of NAMPT (Sasaki et al. 2016). Arguably, SARM1 knockdown or knockout abrogates neuron degeneration. Hence, it is posited as a formidable therapeutic strategy in targeting peripheral neuropathy, traumatic brain injury, and neurodegenerative disease (Henninger et al. 2016; Sasaki et al. 2016; Krauss et al. 2020; Loring and Thompson 2020). With reference to metabolic energy preservation and Wld retarding phenotype, SARM1^−/− constitutes an axon protective mutation as it rescues basal extracellular acidification rate (glycolysis) instead of oxygen consumption rate (mitochondrial respiration) (Godzik and Coleman 2015). In view of this compelling evidence, a reasonable premise for investigation is presented by the conceivable links between NAD⁺ metabolism, SARM1, axon regeneration and CI/R/IS, with a particular emphasis on hitherto unknown epigenetic antecedents.

NAD⁺, Sirtuins and PARP-1—The Interplay in Devising Pharmacological Opportunities

The utilitarian aspects of NAD⁺ are diverse as it serves as a substrate for NAD⁺-metabolizing enzymes (Cantó and Menzies 2015; Jing and Lin 2015; Yang X et al. 2018; Lautrup et al. 2019; Zapata-Pérez et al. 2021; Chini et al. 2021). NAD⁺ signaling is tightly regulated by functional interplay between NAD⁺ synthesis and consumption as well as crosstalk between putative enzymes involved in biological processes. The prominent contenders for NAD⁺ consumption are SIRTs and PARP-1, which play diverse roles in modulating various cellular physiological processes in normal and altered/disease scenarios like that of CI/R. In the next few sections, the epigenetic basis of gene regulation as mediated by SIRTs and PARP-1, and their association with behavioral paradigms under normal and pathophysiological settings like CI/R will be discussed.

Sirtuins and Ischemic Brain

Sirtuins (1–7) are highly conserved NAD⁺ -dependent class III histone deacetylases (HDACs) and their metabolic implications are expansive, in terms of versatility of target molecules involved. SIRT1, 6 and 7 are predominantly nuclear in location and SIRT3,4 and 5 are localized in mitochondria (Jing and Lin 2015). The effective compartmentation of sirtuins conceivably aligns with that of NMNATs, which could explain the evolutionary significance of cofactor pool sequestration (Cantó and Menzies 2015; Jing and Lin 2015). This could be predicated on the dependence of nuclear sirtuins on NAD⁺ for the purpose of removal of acetyl groups in histone (H) tail lysine (K) residues. This subsequently modulates the expression of genes related to cellular metabolism (Jing and Lin 2015). The central tenet of neuroprotection by SIRTs is based on their ability to stimulate oxidative metabolism, mitochondrial biogenesis and antioxidant gene expression through deacetylation and ADP-ribosylation of metabolic enzymes, transcription factors (TFs) and their coactivators (Zhang et al. 2011; Langley and Sauve 2013; She et al. 2017; Ajami et al. 2017). Likewise, an increased activation of these enzymes in the initial stages of an ischemic assault is conceivable, given their ability to promote a ‘failsafe’ mechanism that serves to rescue the potentially salvageable ischemic penumbra from becoming necrotic as the core (She et al. 2017; Khoury et al. 2018).

Nuclear Sirtuins-Prominent Candidates in Therapeutic Arsenal Against CI/R Injury

SIRT1

SIRT1 acts as a pro-survival factor, which is consistent with its enrichment in cortical, striatal and hippocampal neurons that are typically assailed during CI/R. SIRT1 is essential for maintenance of neuronal plasticity, promoting axonal and dendritic growth and preventing cognitive decline (Michán et al.2010; Gao et al. 2010). In this regard, citicoline, a SIRT1 activator reportedly enhances neuronal plasticity and neuroregeneration in experimental stroke models (Diederich et al. 2012; Hurtado et al. 2013) (Table 1). As a vital modulator of energy metabolism, SIRT1 deacetylates and activates peroxisome proliferator-activated receptors gamma coactivator-1alpha (PGC-1α), which together with peroxisome proliferator-activated receptors (PPAR-γ), transcriptionally co-activates genes related to mitochondrial biogenesis, metabolism (gluconeogenesis, thermogenesis and fatty acid oxidation) and antioxidation (superoxide dismutase (SOD), catalase, glutathione peroxidase (GPx) etc.) (St-Pierre et al. 2006) (Fig. 3b). PGC-1α also attenuates oxidative stress, as observed in a transient global ischemia model through upregulation of uncoupling protein (UCP-2) and SOD genes, thereby preventing neuronal death in hippocampal CA1 region (Chen et al. 2010). Perhaps, this feature underlies the neuroprotective effect of icariin (a flavonoid) and α–lipoic acid in experimental stroke and focal ischemia (Zhu et al. 2010; Fu et al. 2014) (Table 1). SIRT1-PGC-1α signaling axis mediates the salutary effects of various neuroprotective agents against CI/R-induced neurovascular injury and cognitive dysfunction. These agents and paradigms modulate SIRT1-PGC-1α signaling axis (mangiferin, ginsenoside, limb postconditioning); AMPK/SIRT1-PGC-1α pathway (CTRP3); NAMPT-NAD⁺ and its downstream SIRT1/2/3-FOXO3A-MnSOD/PGC-1α signaling pathways (notoginseng leaf triterpenes, notoginsenoside R1); Sirt1/Mapt pathway (astragaloside IV) (Gao et al. 2020; Xie W et al. 2020; Chen M et al. 2021; Huang Q et al. 2021; Li L et al. 2021; Shi et al. 2021; Zhu et al. 2021) (Table 1). These findings expand the repertoire of neuroprotective agents for improvement of neurological outcomes and aid in neuro-restoration following IS. Contrastingly, SIRT1 inhibition with sirtinol following permanent focal ischemia promotes acetylation and activation of NF-κB (p65/RelA) and p53, both of which activate inflammatory and apoptotic pathways respectively, leading to an increased infarct volume. Counteracting with an activator 3-mediated increase in SIRT1 expression in ipsilesional mouse brain cortical neurons following MCAO prevents p53 and p65 activation, resulting in reduced infarct volume (Hernández-Jiménez et al. 2013) (Fig. 3b) (Table 1). Also, co-administration of MS-275 (histone deacetylase 1–3 inhibitor) and resveratrol reduces acetylation of RelA K310 residue and histones. It prevents apoptotic cell death with a consequent reduction in infarct volume and neurological deficits, as demonstrated in a transient MCAO (tMCAO) mouse model (Lanzillotta et al. 2012) (Table 1). Also, SIRT1 promoter harbors putative NF-κB binding sites and FOXO1 core binding repeat motifs and is regulated in a positive feedback loop by NF-κB and FOXO1 (Xiong et al. 2011; Katto et al. 2013). Perhaps, deacetylation of these transcriptional regulators promotes SIRT1 autotranscription.

Fig. 3.

a NAD⁺ serves as a co-substrate in the deacetylation reaction catalyzed by sirtuins (SIRT 1–3,5,7). It entails an acetylated substrate becoming deacetylated, and generating deacetylated substrate, nicotinamide (NAM) and 2′-O-acetyl-ADP-ribose. ADP-ribosyl transferase activity of SIRT4 and SIRT6 catalyzes ADP-ribosylation of target proteins using NAD⁺ as a donor of ADP-ribose group. b, c Schematic representation of specified cellular targets and pathways of nuclear SIRTs, of relevance to cerebral ischemia/reperfusion (CI/R) pathology. SIRT1 deacetylates and activates peroxisome proliferator-activated receptor gamma coactivator-1alpha (PGC-1α), which together with peroxisome proliferator-activated receptor (PPAR-γ) transcriptionally co-activates the genes related to mitochondrial biogenesis, metabolism and antioxidation (superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GPx). SIRT1 also upregulates uncoupling protein (UCP-2) and manganese-SOD (Mn-SOD) (mitochondrial) expression, that attenuates oxidative stress and precludes neuronal death. SIRT1-mediated deacetylation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) (p65/RelA) and p53 inhibits neuronal apoptosis and inflammation. SIRT1 mediates NAMPT-induced LKB1 (a serine/threonine kinase) deacetylation and AMP-activated protein kinase (AMPK) activation. This associates with autophagy signaling cascade viz., tuberous sclerosis complex 2 (TSC2)-mammalian target of rapamycin (mTOR)- ribosomal protein S6 kinase beta-1 (S6K1) axis. Phosphorylation of TSC2 at Ser1387 by AMPK negatively regulates mTOR and induces autophagy. This explains SIRT1-activating and inhibitory effects of estrogen and nicotinamide respectively. Both SIRT1 and SIRT6 negatively regulate PARP-1 to prevent neuronal apoptosis (parthanatos) by limiting apoptosis inducing factor (AIF) translocation. SIRT6 shares SIRT1 histone targets H3K9 and H3K56 in the promoters of NF-κB (p65/RelA)-dependent genes and abrogate NF-κB-dependent apoptosis and inflammation. This potentiates mitochondrial metabolism and prevents neuronal death. SIRT1 and SIRT6 modulate glycolysis by deacetylating H3K9 at hypoxia inducible factor-1 alpha (HIF-1α)-target glycolytic gene promoters. SIRT1 and SIRT6-mediated deacetylation of high mobility group box 1 (HMGB1) prevents inflammatory response. A cocktail of micro (miR), long non-coding(lnc), circular(circ) RNAs mediate cellular responses in SIRT1-dependent manner, that avert neuro-damaging events and accords neuroprotection, through neuronal survival and improved neurological outcomes. SIRT7 confers genomic stability by being recruited to stress-inflicted double stranded DNA breaks in PARP-1 dependent manner. Therein, SIRT7 promotes chromosome condensation (increased H3K122 desuccinylation), non-homologous end joining (NHEJ) of DNA (increased H4K18 acetylation that directs 53BP1 recruitment) and precludes apoptosis by p53 deacetylation. Also, by modulating mitochondrial protein folding stress (PFS^mit), unfolded protein response (UPR^mit), endoplasmic reticulum (ER) stress response (X-box binding protein (XBP1)-induced SIRT7 expression) and oxidative stress (by deacetylation of p53 and HIF-1α), it affords cellular protection under stress conditions. (BBB: blood brain barrier) (Refer “Nuclear Sirtuins-Prominent Candidates in Therapeutic Arsenal Against CI/R Injury” section for more details)

Further, SIRT1 can mediate hormetic effects of oxidative stress that is linked to its substrate specificity. Accordingly, phosphorylation of SIRT1 N-terminal Ser27 and Ser47 residues by JNK1 during oxidative stress promotes histone H3K deacetylation that triggers a stress protective pathway (Nasrin et al. 2009). This presumably underlies the neuroprotective action of tetrahydroxystilbene glucoside against CI that involves antioxidative, anti-inflammatory and anti-apoptotic effects (Wang et al. 2009). Also, the neuroprotective effects of exercise postconditioning following stroke are attributed to SIRT1-mediated regulation of ROS/endoplasmic reticulum stress (ERS) pathways to an extent (Li F et al. 2021) (Table 1). Another salutary effect of SIRT1 against cerebral hypoperfusion injury is preservation of cerebral blood flow (CBF) and hence cerebrovascular reserve through endothelial nitric oxide synthase (eNOS) deacetylation (Hattori et al. 2014) (Table 1). Also, hyperbaric oxygen (HBO) treatment activates ATP/NAD⁺/SIRT1 pathway that attenuates hemorrhagic transformation and brain infarction. This is substantiated by improved neurological scores in a hyperglycemic MCAO rat model (Hu et al. 2017) (Table 1). SIRT1-induced deacetylation of hyperacetylated nuclear protein viz., high mobility group box 1 (HMGB1), with a concomitant decrease in MMP-9 expression has been observed following HBO treatment thereby ameliorating neuroinflammation in CI/R setting (Zhao PC et al. 2021) (Fig. 3b) (Table 1).

NAMPT and SIRT1 constitute integral components of a systemic regulatory network for preserving the robustness of metabolic processes. This occurs through orchestration of physiological responses in response to internal and external metabolic cues (Imai and Yoshino 2013). Modulation of AMPK activity by SIRT1 underlies the neuroprotective effect of NAMPT upregulation against cerebral ischemic damage, through an increase in NMN availability in brain gray matter (Zhang et al. 2009; Wang P et al. 2012). NAMPT-induced LKB1 deacetylation and AMPK activation is SIRT1-dependent, as demonstrated in SIRT1^−/−, SIRT1^+/− and AMPKα2^−/− neurons (Zhang et al. 2009; Wang P et al. 2011) (Fig. 3b) (Table 1). This modality links with cell-protective/adaptive regimen as autophagy. Neuronal survival upon MCAO or oxygen glucose deprivation (OGD) stress is effected by NAMPT-mediated autophagy induction following NAD⁺ depletion (Wang P et al. 2012). Arguably, SIRT1 activation regulates autophagy when glucose levels are limiting. This involves AMPK-induced glyceraldehyde-3-phosphate dehydrogenase (GAPDH) activation by Ser112 phosphorylation, causing it to redistribute to nucleus, bind to SIRT1 and induce autophagy (Chang et al. 2015). SIRT1 also modulates autophagy signaling cascade viz., tuberous sclerosis complex 2 (TSC2)- mammalian target of rapamycin (mTOR)- ribosomal protein S6 kinase beta-1 (S6K1) axis in protecting neurons from necrosis during CI. This involves phosphorylation of TSC2 at Ser1387 by AMPK that negatively regulates mTOR phosphorylation (Wang P et al. 2012) (Fig. 3b) (Table 1). SIRT1 also directly interacts with TSC2 and antagonizes mTOR signaling thereby triggering autophagy in ischemic brain. Estrogen-mediated cerebroprotection against CI-induced injury also aligns with the modality of an activated SIRT1-AMPK axis, that seemingly confers resistance to neuronal apoptosis (Guo et al. 2017) (Fig. 3b) (Table 1). Contrastingly, betulinic acid-induced suppression of autophagy through activation of SIRT1/FoxO1 pathway seems to ameliorate cerebral injury in CI/R setting (Zhao Y et al. 2021) (Table 1). Such findings will be crucial in resolving equivocal relationship between autophagy and CI/R, as the former functions as a double-edged sword (Wang P et al. 2018; Mo et al. 2020). Despite an apparent consensus on neuroprotective effect of SIRT1, few reports made it confounding as they proposed a detrimental effect of SIRT1 activation during acute neuronal damage and death (Ng and Tang 2013). This relates to its ability to accelerate NAD⁺ depletion in energetically compromised neurons (Liu et al. 2009). Added to this, SIRT1 can sensitize neurons to oxidative stress by deacetylating insulin/IGF-1 signaling adaptor IRS-2 and promoting downstream activation of pro-oxidative Ras/ extracellular signal-regulated kinases (ERK)1/2 pathway (Li et al. 2008). It is anticipated that ongoing studies can resolve the contested role of SIRT1 in neurodegeneration and protection, in view of its potential as a therapeutic target in human stroke (Teertam and Prakash Babu 2021).

Expanding the arsenal of epigenetic modalities, a gamut of non-coding RNAs targeted by SIRT1 has been implicated in modulating neurological outcomes in CI/R. In this order, elevation in c-myc levels directed by miR-200b-5p-targeted SIRT1 expression has been reported to lower neuroinflammation, neuronal apoptosis and enhance neuronal survival in MCAO model (Liu X et al. 2021) (Table 1). In yet another study, miR-19a/b-3p has been shown to exacerbate inflammatory responses by targeting SIRT1/FoxO3/SPHK1 axis during CI/R (Zhou et al. 2021). LncRNA MALAT1 was shown to attenuate CI/R injury and cognitive dysfunction by regulating miR-142-3p/SIRT1 axis (Meng et al. 2021). Similarly, LncRNA Snhg8 supposedly regulates miR-425-5p-mediated SIRT1/NF-κB signaling, thereby lessening microglial inflammation response and BBB damage in IS (Tian et al. 2021) (Table 1). Further, with circular RNAs, mmu_circ_0000296 was reported to regulate neuronal apoptosis induced by chronic CI through miR-194-5p/Runx3/SIRT1 pathway (Huang K et al. 2021) (Fig. 3b) (Table 1).

SIRT6

The neuroprotective ability of SIRT6 to an extent overlaps with the salutary effects of SIRT1 activation in IS. This is due to its sharing of SIRT1 histone targets H3K9 and H3K56 in the promoters of NF-κB p65(RelA)-dependent genes related to glycolysis and lipid metabolism (Kawahara et al. 2009) (Fig. 3b). This implies target gene expression similarities between SIRT1 and 6 and their concerted effect on stimulation of mitochondrial respiration. Notwithstanding this, SIRT6 has been shown to independently modulate glycolysis by deacetylating H3K9 at HIF-1α target glycolytic gene promoters (Kugel and Mostoslavsky 2014). SIRT6 also functions as a transcriptional corepressor by physically interacting with p65/RelA subunit and deacetylating H3K9 at the promoters of NF-κB target genes that abrogates NF-κB-dependent apoptosis and inflammation (Zhong et al. 2010). In an ischemic milieu, down-regulated SIRT6 expression, as observed in ischemic areas of rat brain promotes cytoplasmic translocation of HMGB1 and an inflammatory response (Lee et al. 2013) (Fig. 3b). This is primarily attributed to increased NF-κB activation in the absence of SIRT6-induced H3K9 deacetylation. SIRT6 expression declines following CI/R and its overexpression in the brain employing in vivo gene transfer stimulates antioxidant NRF2 signaling, that in turn abates oxidative stress (Zhang et al. 2017). The same is recapitulated in OGD/reoxygenation -stimulated neuro-2a neuroblastoma cells in vitro, mimicking the effects of antioxidant N-acetyl cysteine. Concurrent with this is a reduction in brain tissue damage and attendant neurological deficits. In addition, endothelial SIRT6 is instrumental in blunting CI/R-mediated BBB damage, thus appearing as a promising therapeutic target for IS (Liberale et al. 2020). Given the strong K deacetylation activity of SIRT6, it would be interesting to decipher the components of brain acetylome that get impacted by SIRT6 during CI/R.

SIRT7

As regards SIRT7, it is a weaker and substrate-specific deacetylase. This is reflected in absence of discernible changes in global acetylation levels of nuclear and nucleolar proteome following SIRT7 knockout. Nevertheless, SIRT7 functions as a dynamic nuclear regulator of mitochondrial homeostasis, as its deficiency induces multisystemic mitochondrial dysfunction. This involves deacetylation of specific K residues in homo and heterodimerization domain of GABPβ1, a master regulator of nuclear-encoded mitochondrial genes (Ryu et al. 2014). Diverse aspects of its functionality as a deacetylase and desuccinylase have been reported in association with metabolic regulation (glucose, lipid and mitochondrial); genomic stability, ribosome biogenesis and protein translation; cellular stress (oxidative and endoplasmic reticulum stress pathways) which are reviewed in detail in Wu et al. 2018 (Fig. 3c). As regards chromatin, SIRT7 exhibits DNA-activated deacetylase and desuccinylase activities and targets histones in chromatin context in PARP-1 dependent manner. SIRT7 thereby modulates chromatin compaction and genomic stability during DNA-damage response. This is accomplished by an increased H3K122 desuccinylation and increased H4K18 acetylation and 53BP1 recruitment to facilitate non-homologous-end-joining (NHEJ) of double-stranded DNA breaks induced by stress, that eventually affect cell survival (Wu et al. 2018). (Fig. 3c). Nevertheless, its relevance in CI/R paradigm is yet to be understood.

Mitochondrial Sirtuins

SIRT3

SIRT3,4 and 5 are localized in the mitochondria and influence the processes controlling energy production, metabolism, apoptosis and intracellular signaling (Verdin et al. 2010). SIRT3, a bona fide mitochondrial sirtuin stimulates oxidative phosphorylation (OXPHOS) and hence ATP production through deacetylation and activation of Krebs cycle enzymes and complex I,II and V of ETC (Cimen 2010; Yu et al. 2012; Ozden et al. 2014) (Fig. 4). A decrease in pH and concurrent loss of Δψm promotes the release of SIRT3 (that is otherwise sequestered by ATP synthase so as to maintain mitochondrial homeostasis) into the matrix to deacetylate the substrates (Yang et al. 2016). SIRT3 can also avert apoptosis and resultant neurodegeneration by deacetylating cyclophilin D that prevents its association with adenine nucleotide translocase (ANT), and hence mitochondrial permeability transition pore (MPTP) formation (Hafner et al. 2010). OXPHOS potentiation could also be achieved by this maneuver, as it stalls glycolysis through hexokinase II (HKII) dissociation from voltage dependent anion channel (VDAC) and its redistribution from the mitochondria to the cytosol (Shulga et al. 2010) (Fig. 4). This modality serves to explain the beneficial effects of SIRT3 against ischemia-induced MPTP opening and apoptosis (Yang Y et al. 2021). SIRT3 also attenuates mitochondrial oxidative stress. Exposure of cortical neurons to H₂O₂ upregulates SIRT3, that attenuates mitochondrial Ca²⁺ overload and promotes expression of mitochondrial biogenesis-related TFs (Dai et al. 2014). In an ischemic paradigm, SIRT3 can avert Ca²⁺ transients by upregulating PGC-1α and manganese (Mn)-SOD (mitochondrial) (Wang Q et al. 2015). This protective strategy is also adopted by microglia through FOXO3-mediated antioxidant gene expression when challenged with ischemic affronts (Rangarajan et al. 2015). Along this line, SIRT3-mediated deacetylation of FOXO3A at K271,290 residues upregulates a repertoire of genes related to mitochondrial biogenesis, fission/fusion, and mitophagy (Tseng et al. 2013) (Fig. 4). These are essential for preserving mitochondrial reserve for combating cellular oxidative stressors by enhancing mitochondrial mass, ATP generation and removing defective mitochondria. Following CI/R, SIRT3 expression declines resulting in exacerbated oxidative stress and increased mitochondrial fission (Zhao et al. 2018). SIRT3 overexpression seemingly reduces brain infarct volume and rescues neurons from caspase-9 dependent cell death. This is tangentially linked to decreased mitochondrial fission, that increasingly relies on Wnt/β-catenin pathway activation, which is otherwise attenuated during CI/R (Zhao et al. 2018) (Table 1). SIRT3 can also promote neurovascular recovery following chronic IS by dampening vascular endothelial growth factor (VEGF), protein kinase B (Akt) and extracellular signal-regulated kinase (ERK) signaling pathways that impact angiogenesis and neurogenesis, as demonstrated in a transient MCAO mouse model (Yang X et al. 2018) (Fig. 4). In line with this, regulation of HIF-1α/VEGF signaling in astrocytes and BBB integrity constitutes the modality of SIRT3-mediated neuroprotection against IS injury (Yang X et al. 2021) (Fig. 4). Mechanistically, following OGD, SIRT3 modulates VEGF expression by inhibiting HIF-1α signaling (Table 1). Further, SIRT1-SIRT3 axis has been proposed to be of therapeutic benefit, as it significantly modulates BBB physiology by regulating mitochondrial ROS generation (Chen et al. 2018).

Fig. 4.

Schematic representation of specified cellular targets and pathways of mitochondrial SIRTs, of relevance to cerebral ischemia/reperfusion (CI/R) pathology. SIRT3 stimulates oxidative phosphorylation (OXPHOS) and hence ATP production through deacetylation and activation of Krebs cycle enzymes (2-oxoglutarate dehydrogenase (2-OGDH, pyruvate dehydrogenase (PDH), succinate dehydrogenase (SDH) and complex I, II and V of electron transport chain (ETC). SIRT3-mediated deacetylation of cyclophilin D (CypD) prevents its association with adenine nucleotide translocase (ANT) and hence mitochondrial permeability transition pore (MPTP) formation, thereby inhibiting neuronal apoptosis and neurodegeneration. This has the collateral effect of OXPHOS activation and glycolysis inhibition, through hexokinase II (HKII) dissociation from voltage dependent anion channel (VDAC). SIRT3 can also diminish Ca²⁺ transients by upregulating PGC-1α and manganese (Mn)-superoxide dismutase (SOD) (Mn-SOD). Also, SIRT3-mediated deacetylation of transcriptional factor FOXO3A at K271,290 residues upregulates a repertoire of genes related to mitochondrial biogenesis, fission/fusion and mitophagy. SIRT3-mediated augmented expression of Mn-SOD mediates neuroprotective effects of melatonin in CI/R. By dampening vascular endothelial growth factor (VEGF)/ hypoxia inducible factor (HIF-1α) signaling, it promotes neurovascular recovery and preserves blood brain barrier (BBB) integrity. SIRT3 mediates neuroprotective effects of nicotinamide mono-nucleotide (NMN) administration in CI/R setting through improved mitochondrial functioning. By normalizing hippocampal NAD⁺ levels that diminish following an ischemic insult, NMN administration reduces Mn-SOD acetylation thereby activating it, and reducing the associated ROS production in SIRT3-dependent manner. By modulating AMP Kinase (AMPK)-mammalian target of rapamycin (mTOR) axis and ensuing autophagic response, SIRT3 mediates neuroprotective ability of Lin28a in CI/R setting. SIRT4 promotes mitochondrial function and biogenesis by deacetylating cypD and imminent sequelae. SIRT4 upregulates enzymes associated with fatty acid metabolism in peroxisome proliferator-activated receptor gamma coactivator-1alpha (PGC-1α)-peroxisome proliferator-activated receptor (PPAR-γ)-dependent manner. ADP-ribosylating activity of SIRT4 inhibits glutamate dehydrogenase (GDH) activity that affects glutamine metabolism. SIRT5 negatively regulates pyruvate PDH and SDH by desuccinylation, thereby impacting mitochondrial respiration and BBB permeability. Removal of SIRT5 fosters protein kinase C-epsilon (PKC-ε)-mediated reduction of ETC perturbations and cell damage that effectively counteracts cortical degeneration and cell death (Refer to “Mitochondrial Sirtuins” section for more details)

In yet another study, UCP2 that regulates ATP and ROS production has been shown to modulate SIRT3 activation through energy sensing. This, in turn preserves mitochondrial stability through deacetylation of specified targets (Su et al. 2017). However, refuting this observation, genipin an UCP2 inhibitor has been shown to protect against CI/R injury by intercepting UCP2-SIRT3 axis (Zhao et al. 2019). Melatonin also reportedly promotes SIRT3 expression and activates SIRT3 signalling pathway following tMCAO, so as to attenuate CI/R injury (Liu et al. 2019). Activation of Akt-SIRT3-SOD2 signalling axis, with consequential reduction of mitochondrial damage has been proposed to protect against focal CI/R injury in diabetic mice (Liu et al. 2019; Liu L et al. 2021) (Fig. 4) (Table 1). Another modality for conferring neuroprotection in CI/R setting involves Lin28a (highly conserved RNA-binding protein) that markedly promotes autophagy through AMPK-mTOR pathway activation in SIRT3-dependent manner, as evidenced by increased LC3-II levels in nerve cells (Chen D et al. 2021). Herein, SIRT3-induced LKB1 activation, with successive phosphorylation-dependent activation of AMPK, and concurrent inhibition of mTOR phosphorylation, conceivably leads to autophagic level acceleration (Table 1). Of particular interest, is a potent link between mitochondrial NAD⁺ metabolism, ROS production, and mitochondrial fragmentation, wherein NMN administration normalizes hippocampal mitochondrial NAD⁺ pools, protein acetylation, and ROS levels in SIRT3-dependent manner (Klimova et al. 2020) (Fig. 4). Herein, a decline in ischemia-induced SOD2 acetylation reduces ROS generation. It also prevents mitochondrial fragmentation, by precluding interaction of mitochondrial fission protein pDrp1(S616) with neuronal mitochondria. In a similar vein, upregulated SIRT3 expression and preserving mitochondrial function and integrity appeared to be the underlying mechanism for beneficial effects of luteolin in CI/R (Liu et al. 2020) (Table 1).

Further, SIRT3 upregulation can coordinate an adaptive response with epigenetic underpinnings. This is predicated on Krebs cycle intermediate viz., succinate being a negative regulator of 2-oxo-glutarate (2-OG),iron, citrate, molecular oxygen-dependent DNA and histone-demethylating enzymes viz., ten eleven translocation (TET) methylcytosine dioxygenases (TETs-1/2/3) and jumonji domain-containing histone demethylases (JHDMs) respectively (Xiao et al. 2012). Succinate accumulation through reductive carboxylation in a dysregulated Krebs cycle, is the striking metabolic signature of I/R injury (Chouchani et al. 2014; Narne et al. 2017a). 2-OG dehydrogenase is an interactive partner of SIRT3, which rescues it from product inhibition and plausibly allows demethylation reactions to proceed uninterrupted (Yang et al. 2016) (Fig. 4). Contrastingly, this could also be energetically disconcerting during spontaneous reperfusion, as augmented succinate dehydrogenase (SDH) activity promotes reverse electron transport, succinate oxidation and exacerbated ROS production. SIRT3 also directs minocycline-induced reversal of hypoxia-induced prolyl hydroxylase 2 (PHD2) inhibition in human brain microvascular endothelial cells (HBMVECs). The resultant decline in HIF-1α, MMP-2, MMP-9, VEGF levels and increased tight junction proteins expression possibly averts BBB disruption (Yang et al. 2015) (Table 1). Similarly, resveratrol potentiates AMPK-PGC-1α-ERRα-SIRT3 signaling pathway and attenuates oxidative injury in human vascular endothelial cells (Zhou et al. 2014). Nevertheless, at variation with all these findings, SIRT3 has recently been reported to exacerbate ROS production and hence brain injury after IR (Novgorodov et al. 2016). This is attributed to SIRT3-mediated deacetylation and activation of ceramide synthase isoforms, resulting in ceramide accumulation that would effectively block Complex III of ETC and hence mitochondrial respiration.

SIRT4

A notable feature of SIRT4 is its ability to preserve mitochondrial function and promote mitochondrial biogenesis in a two-pronged manner. By promoting ANT deacetylation, it promotes mitochondrial uncoupling, enhanced respiration and ATP production (Ho et al. 2013) (Fig. 4). SIRT4 knockdown mimics downregulated ANT2 expression that fosters energy deficits. Another distinct facet of SIRT4-regulated energy homeostasis is retrograde signaling, involving an enhanced AMPK-dependent PGC-1α activation that promotes fatty acid oxidation (Ho et al. 2013) (Fig. 4). This is ostensibly linked with the cooperative action of PGC-1α and PPAR-α in regulating the expression of nuclear genes encoding fatty acid oxidation enzymes. Further, ADP-ribosylating activity of SIRT4 inhibits glutamate dehydrogenase (GDH) activity that precludes the entry of glutamate and glutamine into metabolism (Fig. 4). This is characteristic of metabolic syndrome that worsens the risk for stroke (Guarente 2006). The precise role of SIRT4 in IS pathophysiology in relation with above discussed putative aspects of mitochondrial metabolism is not well-explored and needs to be deciphered.

SIRT5

Mitochondrial SIRT5, a weak deacetylase is promiscuous in that, it can remove succinyl, malonyl and glutaryl moieties from substrate enzymes and effectively regulate metabolic pathways as glycolysis, krebs cycle, ketogenesis and aminoacid degradation (Verdin et al. 2010; Rardin et al. 2013; Nishida et al. 2015). SIRT5 utilizes NAD⁺ as a cofactor in the demalonylation and desuccinylation of target proteins, generating NAM and 2'-O-malonyl-ADP-ribose and 2 -O-succinyl-ADP-ribose respectively together with demalonylated and desuccinylated target proteins. SIRT5 negatively regulates pyruvate dehydrogenase (PDH) and SDH, as robust succinylation of these enzymes is observed in SIRT5 knockout mice that associate with increased mitochondrial respiration (Rardin et al. 2013) (Fig. 4). In a metabolic scenario of repressed SDH activity and increased succinate accumulation during ischemic conditions, it would be tempting to speculate that SIRT5 can indirectly inhibit 2-OGDHs. Recent work posits SIRT5 as a therapeutic target in CI/R setting, as its expression increases in MCA following stroke in a tMCAO model, HBMVECs and peripheral blood monocytes of acute IS patients (Diaz-Cañestro et al. 2018). It supposedly precipitates CI/R injury by increasing BBB permeability through occludin (tight junction protein) degradation. Interestingly, SIRT5 is indispensable for mediating the effects of PKCε activation in brain, and reversing ETC perturbations and cellular damage in an ischemia–reperfusion setting (Zou et al. 2018). The primacy of SIRT5 is further demonstrated by inhibition of cortical degeneration following MCAO and PKCε-mediated decrease in cell death following OGD in cortical cultures in/from SIRT5^−/− mice (Morris-Blanco et al. 2016) (Fig. 4). Another study posited purine metabolism as a common metabolic pathway regulated by SIRT5, PKCε and ischemic preconditioning. This implies that SIRT5 is pivotal for regulation of pathways concerning brain metabolism, cogently linking it with ischemic tolerance (Koronowski 2018). Taken together, the above discussed findings and a panoply of proposed modalities involving NAD⁺ network and SIRTs present promising avenues for development of efficient therapies, to limit acute brain injury as in IS.

SIRT2—The Cytosolic Sirtuin

SIRT2 has been implicated in inflammatory response, as SIRT2-mediated deacetylation of NF-κB/p65 at K310 abrogates the expression of inflammatory genes including aquaporin 4 (AQP4), MMP-9, and pro-inflammatory cytokines (Rothgiesser et al. 2010; Yuan et al. 2016). The primary targets are oligodendrocytes in myelin-rich regions of ischemic hemisphere, wherein SIRT2 prevents ROS-induced inflammasome aggregation and microglial activation. Contrastingly, SIRT2 supposedly mediates myelin-dependent neuronal dysfunction in early phase after IS (Krey et al. 2015). This is substantiated by SIRT2 deficiency resulting in fewer neurological deficits due to disrupted necroptotic signaling. Alternatively, it is also hypothesized to affect axonal function with a reduced influence on pure global stroke volume or post-ischemic inflammatory processes.

It is indeed noteworthy that SIRTs act in a coordinate manner in mediating neuronal growth and death. With respect to neurogenesis, in vitro administration of NAD⁺/NMN promotes NSC growth and differentiation through SIRT1, SIRT2 and SIRT6 activation (Zhao et al. 2015). Inhibition of corresponding SIRTs disrupts NAMPT-NAD axis, that in turn suspends pro-growth and pro-differentiation of NSCs. SIRTs coordination also manifests in two distinct cellular paradigms of neuronal apoptosis viz., cerebellar granule neurons (CGNs) placed in low potassium (LK) medium and hippocampal derived HT-22 neuroblastoma cell lines subjected to homocysteic acid treatment (Pfister et al. 2008). While SIRT1 and 5 protect CGNs from LK induced apoptosis, SIRT2,3 and 6 emerge as pro-apoptotic. SIRT5-mediated neuronal rescue is determined by its subcellular localization, as it is anti-apoptotic in CGNs when localized in cytosol and apoptotic when located in mitochondria in HT-22 cells. Paradoxically, pharmacological modulators like NAM and sirtinol do not impede SIRT1’s neuroprotective effect in these paradigms. Also, the proven benefits of resveratrol with respect to SIRT1 activation are not manifested. Taken together, as neuronal apoptosis and regeneration are canonical to ischemic brain pathology, it would be interesting to decipher the effects of sub-cellular compartmentation of SIRTs on neuronal damage, rescue and restoration.

SIRT1 and Axonal Regeneration—A Plausible Pharmacological Opportunity

SIRT1 is the effector of increased nuclear NAD⁺ levels that prevents axonal self-destruction (Araki et al. 2004; Wang et al. 2005). In this milieu, treatment with sirtinol or SIRT1 knockdown precludes NAD⁺-dependent axonal protection after axonal transection. This, however is not observed following treatment with PARP-1 inhibitor (3-AB). SIRT1 also mediates delay in Wld in DRGs treated with resveratrol, in NAD⁺ dependent manner (Calliari et al. 2014). While SIRT1’s role in promoting neuron survival is sufficiently advocated through promotion of neurite outgrowth in cultured neural cells and axonogenesis of primary neurons, it is also implicated in suppressing axon regeneration (Guo et al. 2011a, b; Li et al. 2013; Tang 2019). In conjunction with miR-136, SIRT1 regulates intrinsic mammalian axon regeneration ability by forming a mutual negative feedback regulatory loop (Liu et al. 2013). Also, embryonic SIRT1^−/− DRGs are seemingly protected against axonopathy. This alludes to PTEN-mTOR axis, as SIRT1-mediated deacetylation and activation of PTEN negatively regulates mTOR (Tang 2019) (Fig. 2b). Further, PTEN and mTOR are prominent regulators of axon regeneration and neurogenesis (Park et al. 2008). PTEN deletion and mTOR upregulation promotes CNS neuron regeneration. Suppression of mTOR reportedly impairs synthesis of new proteins in axotomized retinal ganglion cells that impedes regeneration. Though the precise connection between these entities is unclear, a cogent link between PTEN and the role of mTOR in nutrient homeostasis in regulating axon regeneration is conceivable. This involves protein translation and autophagy, as focal inhibition of protein translation or axonal mRNA transport hinders axon regeneration (Fig. 2b). Alternatively, TSC2 inhibition also activates axon regeneration, as TSC2 negatively impacts mTOR activity. This observation contrasts SIRT1’s ability to downregulate mTOR protein levels and its phosphorylation with no effect on mRNA levels, thereby promoting CNS neuron growth and survival (Guo et al. 2011a, b). Also, SIRT1 inhibition stymies axonal development in embryonic hippocampal neurons. SIRT1 upregulation reportedly promotes not only axonogenesis, but also axon elongation in an Akt-decetylation dependent manner (Li et al. 2013). As Akt is PTEN target and SIRT1 unequivocally activates both of them, it needs to be resolved as to which of these targets (if any) is deacetylated during Wld in IS.

PARP-1 in IS: A Potential Druggable Target in CI/R Realm

PARP-1 is a significant molecular entity along the continuum of cell death, apoptosis and cell necrosis. It serves as a molecular switch for cell death paradigms by functioning as NAD⁺-utilizing enzyme, an activity integral to poly(ADP-ribosylation) (PARylation). PARylation involves transfer of ADP-ribose residues from NAD⁺ onto both automodification domain of PARP (auto-PARylation) and glutamate and K residues of other acceptor proteins including histones, TFs, HMGB proteins and components of DNA repair machinery viz., topoisomerases I and II (trans-PARylation). PARP-1 hyperactivation is ubiquitously damaging to all the components of neurovascular unit i.e., neurons and stromal cells (Moroni and Chiarugi 2009). Overactivation of PARP-1 resulting from peroxynitrite (ONOO-)-induced DNA damage (single-stranded DNA breaks) under pathological conditions typified by CI/R causes extensive PARylation. This initiates an energy consuming vicious cycle resulting in rapid depletion of cellular energetic pools (NAD⁺ and ATP) (Alano et al. 2010; Luo and Kraus 2012). The protracted deficit of ATP and NAD⁺, indicative of cellular bioenergetic collapse slackens glycolysis and mitochondrial respiration. In neurons, PARP-1 mediates a distinct form of cell death viz., ‘parthanatos’ characterized by translocation of PAR polymers to mitochondria, loss of mitochondrial integrity through mitochondrial depolarization and MPT (Narne et al. 2017b; Liu S et al. 2021) (Fig. 2b). This propels the release of AIF that translocates to nucleus (neuronal death defining step), binds to DNA, induces chromatinolysis and subsequent neuronal death (Chaitanya et al. 2010; Wang Y et al. 2011; Chaitanya et al. 2011; Baxter et al. 2014; Chaitanya and Babu 2009). Alternatively, PARP-1 directly influences mitochondrial oxidative metabolism through its effect on HK in the glycolytic cascade (Bai et al. 2015; Kadam et al. 2020). This involves interaction of PAR polymers bound to HK with voltage-regulated ion channels and regulation of ATP/AMP exchange between mitochondria and cytosol. PARylation of HK promotes its dissociation from the mitochondrial membrane, thereby intercepting the glycolytic cascade and carbon flux. This, summarily affects mitochondrial oxidative metabolism and hence ATP generation (Andrabi et al. 2014; Fouquerel et al. 2014) (Fig. 2b). In keeping with this, NAD⁺ repletion has been shown to subvert PARP-1 mediated NAD⁺ depletion and glycolytic inhibition (Ying et al. 2003). Incubation of hippocampal slices with PARP-1 inhibitor PJ34 or NAM (5 mM) prior to hypoxia abrogates NADH hyperoxidation following reoxygenation and enhanced ATP content thereby contributing to improved neuronal recovery (Shetty et al. 2014). A new arrow in the quiver for curtailing glutamate excitotoxicity-induced parthanatos is Iduna, an endogenous N-methyl-D-aspartate receptor (NMDAR)-induced survival protein that can bind PAR polymers (Andrabi et al. 2011) (Table 1). PARP-1 inhibition also distinctively protects BBB during inflammation-induced endothelial dysfunction (Rom et al. 2015). By targeting inflammatory response with PJ34 in the direction of reducing CI/R-induced damage, it effectively diminishes COX2, inducible nitric oxide synthase (iNOS), and pro-inflammatory cytokine levels (Jiao and Li 2021) (Table 1). In another prospective treatment modality, 14,15-EET counteracts neuronal parthanatos through upregulated expression of antioxidant enzymes viz., GPx, heme oxygenase-1 and SOD in cortical neurons following I/R (Zhao H et al. 2021) (Table 1). Concurrently, ROS generation, PARP-1 cleavage, and AIF nuclear translocation also markedly reduce in cortical neurons. With another neuroprotective agent baicalein, reduced cerebral infarct volume and neurological scores were demonstrated, owing to abrogation of cytokine release, PARP-1 activation and nuclear translocation of AIF and macrophage migration inhibitory factor in CI/R rats (Li et al. 2020) (Table 1). Consequent of this, there is inhibition of apoptosis, dampened oxidative stress, preserved mitochondrial function and Δψm in OGD cells. PARP-1 inhibition also reportedly attenuates microglial activation and post-stroke inflammation, with peroxiredoxin acting as a critical mediator (Chen et al. 2020) (Table 1). Thus, an armamentarium of PARP inhibitors is being developed for targeting CI/R/IS, given the distinguished role of PARP-1 in various cellular processes, cell death sub-routines and adaptive regimes and its moon-lighting functions (Koehler et al. 2021).

PARP-1 and Axonal Damage—A Formidable Feature for Pharmacological Intervention

The influence of SIRT1 and PARP-1 on axon degeneration and regeneration primarily relies on cellular energy dynamics (Bai et al. 2011). As regards PARP-1, pharmacological inhibition or genetic deletion of PARP-1 has been shown to promote axon regeneration (Brochier et al. 2015; Byrne et al. 2016). PAR formation inhibits axon regeneration as observed in γ-amino-butyric acid (GABA) neurons in C.elegans and in mammalian cortical neurons. Enhancing PAR turnover through increased PAR glycohydrolase (that removes PAR chains) activity, potentiated by delta-like 1 homolog activity enhances axon regeneration (Byrne et al. 2016). Injured environment in neurons is non-conducive for axon growth and is induced by myelin debris and reactive astrocytes. This poses a molecular and physical impediment for axon growth. At this juncture, PARP-1 critically mediates multiple growth-inhibitory signals (Brochier et al. 2015). PARP-1 inhibition conceivably augments growth cone motility by inhibiting growth-inhibitory cascades. This modality involves an altered growth state achieved by (i) energy (NAD⁺ and ATP) depletion by excessive PARP-1 activation (as discussed earlier) (Fig. 2b) and (ii) the ability of PAR polymers to form a part of the local signal that necessarily governs axon-growing response to non-permissive extrinsic substrates like myelin-associated glycoprotein (MAG), Nogo-A and chondroitin sulphate proteoglycans (CSPG) in primary cultured neurons (Brochier et al. 2015) (Fig. 2c). Intracellular effects of axon-growth inhibiting MAG and Nogo-A is mediated by small-guanosine-tri-phosphate-binding protein i.e., Rho that directly activates ROCK, while CPSG initiates a different process. Nevertheless, both these events collectively contribute to PARP-1-mediated growth cone collapse and impaired axon regeneration of damaged CNS neurons (Brochier et al. 2015) (Fig. 2c).

PARP-1 has been shown to localize in the axon following exposure to an injured environment, typified by disruption of axonal tracts and accumulation of growth inhibitory proteins at the site of damage. The ensuing PAR formation along the axon hinders axon regeneration in a dose-dependent manner. Integral to this is NAD⁺ and ATP depletion that affects axonal survival and growth cone motility, as both these substrates are essential for dynamic modulation of actin and tubulin polymerization (Fig. 2b, c). ATP decline precedes NAD⁺ depletion, which is in part ascribed to PARP-1 interaction with HK, that blocks glycolytic cascade thereby impeding mitochondrial ATP generation (Andrabi et al. 2014; Fouquerel et al. 2014). Consequently, motility of growth cones enriched with mitochondria is affected (Wang et al. 2008) (Fig. 2b). Concurrently, NAD⁺ depletion following PARP-1 hyperactivation abrogates SIRT1 and SIRT3 activation both of which are crucial for axonogenesis and neurite outgrowth (Liu et al. 2013) (Fig. 2b). The Krebs’ cycle substrate pyruvate could alleviate this by circumventing the glycolytic cascade and promoting mitochondrial ATP production, thereby permitting neurite growth on MAG-expressing CHO cells (Brochier et al. 2015). It needs to be seen if a decline in sirtuin activity due to PARP1-dependent NAD⁺ depletion directly impacts neurite growth in an inhibitory environment. Based on all these findings, it is a tenable proposal that PARP-1 inhibition offers the advantage of both neuroprotection and reduced sensitivity of injured neurons to altered growth states, defined by accumulation of inhibitory molecules hostile to axon growth. Taken together, this kind of precise engagement of all the components of NAD⁺ metabolism viz., NAD⁺ biosynthetic and metabolizing enzymes, nucleotidases and presumptive transporters based on their tissue distribution and expression levels generates a comprehensive or holistic concept, that connects NAD⁺ metabolism with neurodegenerative etiology like IS in question and offers therapeutic opportunities (Langley and Sauve 2013; Ruggieri et al. 2015; Yoshino et al. 2018).

NMNAT1, PARP-1 and SIRT1—The Nexus and an Emergent Therapeutic Premise

The undisputed connection between NMNAT1, PARP-1 and sirtuins in regulating gene expression with significant epigenetic footprint has been illuminated by studies in the recent past. In this regard, the body of evidence relating to an ischemic brain has only begun to emerge. Therein, the next few sections seek to elucidate the molecular underpinnings of this potentially interesting and druggable nexus in normal and an altered physiological milieu as CI/R. A cogent explanation of above-mentioned epigenetic antecedents and testable theses in CI/R milieu will be presented and discussed.

Nuclear localization of NMNAT1 is essential to maintain substrate specificity and regulate non-redundant NAD⁺-dependent nuclear functions (Berger et al. 2005). Localization of NMNAT1, PARP-1 and SIRT1 to the same sub-cellular compartment enables a rapid and coordinate action, in response to various stimuli and intracellular milieu (Berger et al. 2005; Ryu et al. 2018). In line with this, a functional relationship exists between enzymatic activities of NMNAT1 and PARP-1 at target gene promoters that regulates gene expression. This is evident from binding of NMNAT1 to auto-PARylated PARP-1 and allosterically stimulating its activity, thereby amplifying PARylation (Berger et al. 2007) (Fig. 5b). Further, a signal-regulated role of NMNAT1 in controlling PARP-1 function is determined by the phosphorylation status of NMNAT1 (Zhang et al. 2012). PKC-mediated phosphorylation of NMNAT1 serine residues or substitution of serine with arginine residues precludes binding of NMNAT1 to PAR. This interactive maneuver proceeds independently of NAD⁺ production and highlights the functioning of NMNAT1 as a ‘chaperone’. Alternatively, NMNAT1 as a ‘feeder’ is recruited to target gene promoters by PARP-1 so as to generate a distinct nuclear NAD⁺ pool. This is functionally compartmentalized from cytosolic and mitochondrial pools to sustain PARP-1 activity and attendant gene expression in MCF-7 cells (Fig. 5b). Nevertheless, interaction of NMNAT1 with PARP-1 is direct without PAR-dependence and NMNAT1 activity is essential for PARP-1-dependent PARylation at target gene promoters. In this milieu, depletion of NMNAT1 or PARP-1 has been shown to affect the expression of about two hundred genes displaying a ten percent overlap of two gene sets (Zhang et al. 2012). Each set comprised up and down-regulated genes and an overlapping set constituted twenty-three genes. Of these, it was observed that five genes viz., ATXN10, SOCS2, PEG10, TMSNB, and NELL2 were differentially regulated by NMNAT1 and PARP-1.

Fig. 5.

a PARPs use NAD⁺ as a substrate for polymerization of ADP-ribose residues on to acceptor proteins, generating poly(ADP-ribosyl)ated (PARylated) target proteins and nicotinamide (NAM) in the process. b–d Interactive dynamics between NMNATs, SIRTs and PARPs. Multiplicity of NMNAT1 function. b NMNAT1 functions as a ‘chaperone’ by binding poly-ADP ribosylated (PARylated) PARP-1 and allosterically stimulating it, so as to transcriptionally activate estrogen receptor-alpha (ERα)-responsive genes. Also, NMNAT1 activity is regulated by protein kinase C (PKC)-mediated serine (Ser) phosphorylation that precludes its interaction with PARP-1. NMNAT1 in the capacity of ‘feeder’ generates NAD.⁺ for optimal functioning of PARP-1. c NMNAT-1 also interacts with and allosterically activates SIRT1, which interacts with transcription factors FOXOs, NF-κB, p53, SOX9 and coregulators (coreg) p300, NCoR, SMRT, to regulate target gene expression with an interspersed deacetylation of H4K16 acetylation marks. d Reciprocal activation of PARP-1 and SIRT1 under stress conditions and determination of cell death subroutines (Refer “NMNAT1, PARP-1 and SIRT1—The Nexus and an Emergent Therapeutic Premise” and “PARP-1 and Sirtuins in NAD⁺ Utilization—Druggable Axis in CI/R Milieu” sections for more details) (PCAF p300/CREB-binding protein (CBP) associated factor (PCAF), ERE estrogen response element)

Focusing on the direct protein targets of PARylation, i.e., the PARylome, PARP-1 binds and PARylates, inter alia, estrogen receptor (ER-α) when potentiated by estradiol (ligand of ER-α) (Zhang F et al. 2013). PARP-1 promotes its binding to the estrogen response element (ERE) in the target gene promoters, thereby facilitating ER-α-mediated gene transcription in cultured vascular smooth muscle cells (Fig. 5b). The mechanistic feature of PARP-1-dependent gene regulation involves obscuring protein–protein interaction sites through charge repulsion, by strongly negatively charged PAR polymers (Ryu et al. 2018). It is indeed noteworthy, that addition of NMNAT1 to reaction with unlabeled ATP stimulates PARP-1 activity markedly, enabling synthesis of longer PAR chains (Zhang F et al. 2013). Absence of NAD⁺ abolishes PARP-1-mediated ER-α-dependent transcription in an in vitro transcription assay with chromatin templates, which nevertheless could be relieved by NAD⁺ repletion. Non-availability of precursor NMN also inhibits PARP-1 activity, owing to no net synthesis of NAD⁺ despite the presence of NMNAT1 and ATP in the reaction. It could thus be inferred that distinct ‘regulatory inputs’ modulating the activity and localization of PARP-1 can be articulated with ‘regulatory outputs’, culminating in various molecular and organismal end points, necessarily governed by PARP-1 functioning.

As regards the other nuclear client protein of NAD⁺ i.e., SIRT1, its deacetylase activity at endogenous promoters is controlled by NMNAT1 and NAMPT. This occurs through cellular NAD⁺ production and NAMPT-dependent cytosolic NAM removal, as NAM avidly inhibits SIRT1 and PARP-1 (Verdin et al. 2010; Zhang and Kraus 2010). SIRT1 is recruited to the promoters of common target genes of NAMPT, NMNAT1 and SIRT1. This requires interaction with sequence-specific TFs like FOXO family members, NF-κB, p53, SOX9 and transcription coregulators like p300, NCoR, SMRT. NMNAT1 interacts with SIRT1 at target gene promoters (Fig. 5c). This proximity seemingly favors a substrate channeling mechanism that creates a microdomain of high NAD⁺ levels, facilitating efficient NAD⁺ utilization by SIRT1 (Wang P et al. 2011; Zhang and Kraus 2010; Grubisha et al. 2005) (Fig. 5c). Akin to PARP-1, colocalization of NMNAT1 with SIRT1 possibly allows allosteric regulation of the latter. In addition, dynamic recruitment of NMNAT1 to target gene promoters is regulated by cellular signaling inputs like protein kinases, stress, cell differentiation and nutrient availability. This, in turn affects NAD⁺ levels by regulating NMN (Berger et al. 2007; Zhang et al. 2009; Wang P et al. 2011). With respect to NAMPT, the idea of its compartmentalization in organelles is substantiated by differential effect of FK866, that significantly reduces cytosolic NAD⁺ while leaving mitochondrial NAD homeostasis unaffected (Pittelli et al. 2010). This decline in NAD⁺ is increasingly attributed to PARP-1 activity rather than SIRT1, and hence could be rescued with PARP-1 inhibition or NAD precursor kynurenine. Taken together, NMNAT1 dynamically regulates PARP-1 and SIRT1-mediated gene expression at endogenous promoters in the capacity of both ‘chaperone’ and ‘feeder’ in response to an integrated input from signaling pathways.

PARP-1 and Sirtuins in NAD⁺ Utilization—Druggable Axis in CI/R Milieu

As NAD⁺ is a common substrate for PARP-1 and SIRTs, a significant crosstalk is conceivable between these molecular entities. Under physiological conditions, PARP-1 exhibits stronger affinity for NAD⁺ at existing cellular NAD⁺ levels (200–500 µM), owing to its comparatively lower K_M values (20–60 µM vs 150–200 µM). This strongly argues in favor of PARP-1 activation over SIRT1 (Bai et al. 2015). By means of their interaction with nuclear receptors, that function in coupling metabolic signals with steering of transcriptional programs, PARPs can serve as secondary retrograde response signal. This would facilitate metabolic adaptation during exigent conditions as CI/R injury. PARP-1 functions as a transcriptional co-activator that requires its acetylation at specific K residues by p300/cAMP-response-element-binding protein (CREB) in response to inflammatory stimuli. This event precedes its interaction with p50 and synergistic coactivation of p300. This, in turn promotes transcriptional activation of NF-κB by p300 and mediator complex and induces expression of target genes viz., like iNOS, TNF-α, IFN-γ (Hassa et al. 2005). In this connection, one of the salutary effects of PARP-1 inhibition is attenuation of inflammation, by precluding NF-κB nuclear translocation and phosphorylation. This inhibits iNOS and ICAM-1 expression, as demonstrated in atherosclerotic model of low shear stress (Qin et al. 2013). Mechanistically, mild stress triggers PARP-1 activation independent of DNA damage. This occurs through K acetylation, preferentially in N-terminal 1–214 amino acids by histone acetyltransferases (HATs) p300/CREB-binding protein (CBP) associated factor (PCAF) or by p300/CBP in the amino acid stretch of 477–525 (Hassa et al. 2005) (Fig. 5d). Activated PARP-1 then PARylates target proteins with concurrent NAD⁺ utilization.

To prevent a decline in cellular NAD⁺ levels, SIRT1 regulates PARP-1 negatively both at the transcriptional and post-translational levels. This was observed in macrophages and cardiomyocytes in the context of NF-κB-dependent immune and stress responses (Rajamohan et al. 2009). SIRT1 per se and also prebound to NAD⁺ reduces PARP-1 activity, by binding it through its conserved core domain and deacetylating K in the canonical amino acid stretches (preferentially in 1–214 aminoacids) of PARP-1. Indeed, Km values of PCAF-acetylated PARP-1 (995 nM) and Km of SIRT1 for PARP-1 deacetylation (1.2 μM) are comparable, which is suggestive of both the enzymes being equally active under a low NAD⁺ concentration. Nevertheless, in stressed cells these values might differ for endogenous reactions (Rajamohan et al. 2009). Alternatively, SIRT1 can regulate PARP-1 activity by deacetylating PCAF/p300/CBP (Pittelli et al. 2010) (Fig. 5d). In the context of escalating cellular stress, PARP-1 activation and PARylation is extensive, resulting in depletion of NAD⁺ levels and augmentation of NAM levels. Consequently, SIRT1 is inactivated together with a decline in its expression (Vida et al. 2017). This situation parallels that of SIRT1^−/− cells, wherein a marked increase in PAR synthesis occurs resulting in AIF-mediated cell death (Kolthur-Seetharam et al. 2006). Increased acetylation of PARP-1 promotes hyperacetylation of substrates (common with SIRT1) like p53 (which is increasingly retained in nucleus) and Ku70 resulting in apoptotic cell death (Motta et al. 2004; Rajamohan et al. 2009) (Fig. 5d). Further, an interactive association between PARP-1 and NMNAT1 also explains the absence of apportioning localized finite NAD⁺ pool generated by NMNAT1. This indeed signifies the competition between SIRT1 and PARP-1 for NAD⁺ utilization under varied stress conditions and suitably explains cell death mechanism in the absence or inhibition of SIRT1.

An increase in NAD⁺ concentration inhibits PARP-1-SIRT1 interaction in vitro. This observation underscores the rubric of subcellular compartmentation, as NAD⁺ increase only activates SIRT1 with no effect on the activity of mitochondrial sirtuins. This influences mitochondrial biogenesis through deacetylation and activation of transcriptional cofactor FOXO3A and coactivator PGC-1α. Alternatively, supplementation with NR augments both nuclear and mitochondrial NAD⁺ levels with consequential activation of both SIRT1 and SIRT3. This augments oxidative metabolism thereby impacting higher order metabolic regulation (Cantó et al. 2012; Wang S et al. 2018). Amelioration of NAD⁺ availability and hence its adequacy promotes oxidative metabolism by stimulating mitotropic tone of gene expression and signal transduction. This can be achieved by PARP-1 inhibition as demonstrated in PARP-1^−/− mice (Bai et al. 2011; Vida et al. 2017). They seem to phenocopy several aspects of SIRT1 activation. Therein, it would be tenable to say that SIRT1 modulates PARP-1-deficient phenotype. In a putative context, PARP-2 acts as a negative regulator of SIRT1 as it binds to SIRT1 gene promoter and inhibits its transcription as demonstrated in cardiac myotubes (Bai et al. 2011). This results in suspension of mitochondrial gene expression and oxygen consumption due to reduced deacetylation and activation of PGC-1α. Similar to nuclear PARP-1, its mitochondrial counterpart is also activated during oxidant stress. A slew of cell-damaging events follows, which include release of PARP-1 from mitofilin into mitochondrial matrix, impairment of ETC by PARylation of respiratory chain complexes with concomitant reduction in mitochondrial NAD⁺. Mitochondrial uncoupling ensues resulting in bouts of ROS release (Brunyanszki et al. 2016).

As regards the remainder sirtuins, hippocampal neurons assailed by NMDA excitotoxicity exhibit a sustained increase in SIRT3 mRNA and protein expression following PARP-1 hyperactivation. There is also concurrent increase in mitochondrial ROS generation and NAD⁺ modulation (Kim et al. 2011). In such a scenario, SIRT1 and SIRT3 impinge on FOXO3A-MnSOD axis, so as to preserve mitochondrial integrity and promote mitochondrial biogenesis. Alongside, an imbalance in mitochondria and nuclear-encoded mitochondrial proteins (OXPHOS subunits) occurs. This constitutes a retrograde signal and elicits a mitohormetic and nuclear adaptive response. Therein, mitochondrial unfolded protein response (UPRmt) ensues, preserving mitochondrial function and hence neuronal survival through an improved response to mitophagy. Further, ADP-ribosylation activity of SIRT6 is harnessed in inducing repair of double stranded DNA breaks, by physically interacting with and mono ADP-ribosylating PARP-1 on K527. Activated PARP-1 then repairs the breaks through NHEJ and homologous recombination (Mao et al. 2011). Taken together, regulation of metabolic flux can be achieved by sirtuins and PARPs by means of their dependence on cellular NAD⁺, and its reciprocal utilization to mediate varied stress responses. This forges an epigenetic nexus with intermediary metabolism that provides a context for development of new therapies that could possibly attenuate CI/R-associated sequelae.

PARP-1 and SIRT1 in Restructuring Chromatin—An Emergent Therapeutic Proposal

PARP-1 and SIRT1 in Chromatin Context

PARP-1-mediated ADP-ribosylation of histones is integral to histone code. In fact, the ability of PAR polymers to interact with other chromatin modifiers enables PARylation to be considered an epigenetic mark in histone code (Ciccarone et al. 2017; Ummarino et al. 2021). PARylation can influence the extent of methylation and active demethylation. Non-covalent attachment of auto-PARylated-PARP-1 at the catalytic domain of DNA methyltransferase (DNMT1) in a ternary complex, together with transcriptional repressor-CTCF (chromatin architectural protein entrusted with maintenance of genome methylation) abrogates its activity (Caiafa et al. 2009) (Fig. 6a). This could extensively affect histone-DNA interactions and exert varied regulatory influences on corresponding gene expression patterns. PARP-1 also functionally interacts with DNA-demethylating TET1, so as to initiate an epigenetic program to direct transcriptional activity at defined loci (Doege et al. 2012) (Fig. 6a). While non-covalent binding between PAR polymers and TET1 catalytic domain precludes its activity, covalent interaction activates TET1 independently of DNA damage (Ciccarone et al. 2017). 5-hydroxymethylcytosine (5hmC) generated by TET1-mediated DNA hydroxylation is abundant in neurons, and is implicated in neural activity-induced demethylation in mature neurons of adult brain (Guo et al. 2011a, b). TET1-mediated DNA hydroxylation also regulates oxidative stress-induced neuronal cell death (Xin et al. 2015). TET1 is essential for hippocampal neurogenesis as TET1 deficiency tapers the adult NPC pool in SGZ and hampers their proliferation. Concurrently, there is hypermethylation and downregulation of a gene cohort related to progenitor proliferation, that would significantly affect spatial learning and memory formation (Zhang RR et al. 2013) (Fig. 6a). Based on these considerations it could be surmised that PARP-1 activity would have a footprint in mediating the effect of various epigenetic modifications, on post-stroke neuronal plasticity and behavioral outcomes.

Fig. 6.

a–d Epigenetic features of the role of PARP-1 and SIRT1 in regulating neurogenesis, synaptic plasticity and varied long-term memory processes. a PARP-1 influences gene expression patterns through its interaction with varied epigenetic enzymes. In this context, auto-poly(ADP-ribosyl)ated (PARylated) PARP-1 together with transcriptional repressor-CTCF, interacts with DNA methyltransferase1 (DNMT1) forming a ternary complex and abrogating DNMT1 activity. This exerts a regulatory influence on transcription at defined loci. PARP-1 through covalent and non-covalent interaction with DNA-demethylating ten-eleven translocase-1 (TET1), partakes in spatial learning and memory processes. This involves regulation of TET-mediated active DNA demethylation generating 5-hydroxymethylcytosine (5hmC) that promotes transcriptional activation of gene cohort related to proliferation of neural progenitors. The affinity of PARP-1 for trimethylated H3K4 (H3K4me3) marks displaces lysine demethylase 5B (KDM5B) from chromatin with concurrent increase in H4 acetylation (H4KAc), resulting in transcriptional activation of specified genes in cortical neurons (Refer “PARP-1 and SIRT1 in Chromatin Context” section for additional details.). b A cross-talk between histone acetylation and (poly ADP-ribosyl)ation (PARylation) elicits significant cellular consequences. Histone 4 Lysine 16 (H4K16) is the common acceptor site for PARP-1 and SIRT1. H4K16 acetylation (H4K16Ac) disallows H4 PARylation (H4PAR). H4K16Ac marks are reduced following PARP-1 activation by double-stranded breaks. H4K16Ac is negatively influenced by NAMPT or NMNAT1 owing to an increased availability of NAD⁺ to support SIRT1 activity. Histone deacetylase 1(HDAC1)-directed deacetylation of H4K16 occurs in primary neurons following oxygen- glucose deprivation (OGD), that alters the expression of genes related to mitochondrial metabolism, neurogenesis and stress response (Refer “PARP-1 and SIRT1 in Chromatin Context” section for additional details). c Cross-talk between PARylation and histone methylation can influence DNA-damage response. Following DNA damage, PARP-1-mediated PARylation of H3K9me3-writing enhancer of zest-2 (EZH2) inhibits its activity, while that of H3 causes EZH2 to dissociate from chromatin, with an associated reduction in distribution of trimethylated H3K27 (H3K27me3) marks on chromatin, that promotes a DNA damage response. As EZH2 activity is essential for neurogenesis and distinct memory processes, it needs to be deciphered if PARP-1-EZH2 interaction impacts these processes. (Refer to “PARP-1 and EZH2—A Case in Point” section for details.) d NMNAT1 and SIRT1 in modulation of nucleolar integrity. In a nucleolar milieu, SIRT1 together with trimethylated H3K9 (H3K9me3)-writing suppressor of variegation 3–9 Homolog 1 (Suv39H1) enzyme and nucleomethylin (NML) forms an energy-sensitive eNoSC (energy-dependent nucleolar silencing complex). eNoSC establishes heterochromatin by H3 deacetylation and H3K9 dimethylation thereby suspending ribosomal DNA (rDNA) transcription. Under conditions of energy (ATP) depletion, NMNAT1 is recruited to eNoSC to support SIRT1 activity through localized NAD⁺ generation followed by heterochromatin formation conferring genome protection until energy balance is reinstated. (Refer to “PARP-1, SIRT1, rRNAs and Nucleolar Integrity in Synaptic Plasticity—An Emerging Leitmotif” section for more details.)

As regards histone methylation, sequential and differential PARylation of histones, in part, guides histone methylation by altering the histone substrate specificity (Caruso et al. 2018). In relation with this, the designated ADP-ribose acceptor sites in regulatory Ks of core histones are K13 of H2A, K30 of H2B, K27 and K37 of H3, and K16 of H4 in the amino-terminal basic tails. Covalent ADP-ribosylation of H3K27 and H3K37 prevents their subsequent methylation by SET7/9, a H3 histone methyltransferase while promoting methylation of H1.4 (Messner et al. 2010). Conversely, methylated H3 and H1 are amenable for PARylation, and in the absence of PARylation compete for SET7/9-mediated methylation in the consensus motif. PARP-1 increasingly associates with transcriptionally permissive H3K4me3 marks at cis-acting sites like promoters and enhancers. Therein, inhibition of chromatin-binding ability of histone lysine demethylase (KDM)5B by PARylation enriches the putative genes with transcriptionally active H3K4me3 marks (Fig. 6a). This could also corroborate an increase in H4 acetylation in cortical neurons following PARP-1 activation. An insult like DNA damage causes PARP-1 to get globally redistributed away from the promoters to the sites of DNA damage that could partially correspond with reduced global reduction in transcription. Therein, it could be inferred that PARP-1 and PAR glycohydrolase localize at the gene promoters, modulate histone acetylation in a concerted manner, and regulate gene expression in a context dependent fashion (Verdone et al. 2015).

On the other hand, SIRT1 mediates facultative heterochromatin formation through deacetylation of H1K26, H3K9, H4K16 and hypomethylated H3K79. This produces an overall effect of chromatin silencing through H1 recruitment in proximity of promoters and spreading of hypomethylated H4K79 (Vaquero et al. 2004). It needs to be clarified if H1 modification by PARP-1 and SIRT1 has any interspersed cross-talk in effecting transcriptional outcomes that concern LTP and hence cognitive dysfunction in a pathological milieu like CI/R. H4K16 is the common acceptor site for PARP-1, H4K16 HATs-KAT5 (TIP-60), KAT8 (MYST1,MOF), KAT2A (GCN5) and histone deacetylases like HDAC-1 and SIRT1 and hence suggests an interplay of these entities in modulating chromatin topology, protein interactions and global activation of gene expression. Under such stipulations, a cross-talk between acetylation and PARylation could have significant implications, as H4K16 acetylation severely impairs H4 PARylation (Messner et al. 2010). This could be of instructional consequence to DNA repair, as TIP60-mediated H4K16 acetylation (H4K16Ac) initiates NHEJ (Hsiao and Mizzen 2013; Panier and Boulton 2014). A marked decrease in H4K16Ac to the tune of forty to eighty percent in PC12 cells and primary neurons following an acute insult like OGD occurs and is maximally attributed to HDAC1 activity rather than SIRT1 (Dmitriev and Papkovsky 2015) (Fig. 6b). Also, H4K16Ac can sufficiently be restored by trichostatin treatment and also following reperfusion. Such a transient expulsion of chromatin-relaxing H4K16 mark could condense the chromatin so as to suspend local transcription. This could plausibly affect mitochondrial metabolism (precisely linked to the transcriptional regulation of UCP2), neurogenesis and also stress response; with reference to notch signaling pathway components. It could also be surmised to affect DNA repair, as H4K16Ac marks immediately decline after formation of double-strand breaks that activate PARP-1 and are restored after DNA damage accrual (Messner et al. 2010). Also, H4K16 acetylation increases following NAMPT inhibition or NMNAT-1 knockout (Zhang et al. 2009). This is attributed to decreased SIRT-1 activity at target gene promoters lacking the complex of NAMPT and NMNAT-1 and hence decreased NAD⁺ levels. (Fig. 6b). These features present a study premise for exploring the interaction and effects of H4K16 acetylation dynamics, pertinent to NAMPT expression, with an impending effect on DNA damage and repair modalities in a pathological milieu as CI/R. Taken together, a comprehensive understanding of chromatin dynamics articulated by NAD⁺-consuming epigenetic enzymes like SIRT1 and PARP-1, as mentioned above would enable identification of emerging prospects, hitherto unknown in the epigenetic landscape of an ischemic brain, that are amenable for therapeutic intervention.

PARP-1 and EZH2—A Case in Point

Enhancer of zeste 2 (EZH2), a transcriptionally repressive H3K27me3 writing enzyme, is a crucial component of polycomb repressive complex (PRC2) that compacts chromatin and promotes heterochromatin formation. PRC2 belongs to evolutionarily conserved polycomb group (PcG) proteins that can assemble into distinct protein complexes. These constitute well-surveyed regulatory systems with an imperious role in development by striking precise balance between stem cells and progenitor cells (Schuettengruber et al. 2017). They sculpt the chromatin structure by mediating histone PTMs, with a consequential effect of dynamically regulating gene expression patterns during cortical development (Albert et al. 2017). These substantially contribute to quantitative and qualitative variation in cellular composition, neuron number, shape, size and structure of the neocortex between species, that measurably define the cognitive abilities of humans (Bölicke and Albert 2022). PRC1 and PRC2 are pre-eminent PcG protein complexes that act in concert, in order to establish transcriptionally repressive chromatin through deposition of H3K27me3 at target gene promoters. Studies have shown that EZH2 is essential for maintenance of balance between self-renewal and differentiation through H3K27me3 chromatin marks in cortical progenitor cells. Loss of function of EZH2 results in deregulation of transcriptional circuitries, with an altered timing of cortical development, by limiting the neurogenic period for progenitor cells and diminishing neuronal output (Pereira et al. 2010). Current research is delving into the minutiae concerning its role in various debilitating neurologies involving dysregulated molecular networks, with resultant effects on synaptic plasticity and cellular differentiation trajectories (Buontempo et al. 2022; Park et al. 2022).

As regards PARP-1, it directly interacts with EZH2 in regulating gene silencing and hence global gene transcription. PARP-1 inhibition or KO precludes the occupancy of TF E2F4 at EZH2 gene promoter and upregulates EZH2 expression thereby increasing global H3K27me3 marks (Martin et al. 2015). An extension of EZH2 function into mediating DNA damage response involves PARylation and PARP-1 activity. In this context, a two-pronged mechanism operates following DNA damage that involves (i) EZH2 PARylation that diminishes its activity and (ii) H3 PARylation that dampens the affinity of EZH2 for chromatin causing its subsequent dissociation (Fig. 6c) (Caruso et al. 2018). Therein, PARP-1 indulges in a cross-talk not only with acetylation but also with methylation, so as to steer the DNA damage response by recruiting PRC2 components like EZH2 and SUZ12 to sites of DNA (Ciccarone et al. 2017). As for SIRT1, it engages with EZH2 in PRC4 complex in undifferentiated embryonic stem cells and EZH2-over-expressed cancerous cells and its implications are less well understood (Jing and Lin 2015). EZH2 also has a footprint in regulating fear memory consolidation as H3K27me3 marks attenuate phosphatase and tensin homolog (PTEN) expression resulting in potentiation of AKT/mTOR signaling (Fig. 6c). This kind of translational control impacts learning-dependent synaptic plasticity in neurons (Jarome et al. 2018). In a similar vein, EZH2 promotes expansion of NPCs by inhibiting PTEN, the negative regulator of neurogenesis and activation of Akt-mTOR signaling (Zhang et al. 2014). This could be of consequence for neurogenesis, as PTEN deletion in granule neurons results in increased soma, apical dendrites and spine density while its deletion in adult NSCs in the subgranular zone (SGZ) augments proliferation. The self-renewal ability of NSCs is augmented following conditional deletion of PTEN in adult NSCs in ependymal zone that stimulates constitutional neurogenesis in olfactory bulb (Gregorian et al. 2009). EZH2 deletion abrogates NSC proliferation and differentiation and its KO impairs spatial learning and memory, contextual fear memory and spatial pattern separation. This indeed accentuates the role of EZH2 in behavioral outcomes (Zhang et al. 2014). In relation with an ischemic brain, PcG proteins are preponderant in an ischemic tolerant brain rather than pre-conditioned or ischemic-damaged brain. They suppress the transcriptional activity of certain potassium channel genes that feature in the hibernation patterns of gene expression characteristic of an ischemic tolerant brain, with an upshot of suppressed outward potassium currents (Stapels et al. 2010). As EZH2 is the catalytic core subunit of PRC2, with all the plausible links between PARP-1, EZH2 and above discussed sequelae, it would be interesting to explore if PARP-1-EZH2 rapport could serve in identification of effectors of ischemic tolerance. This could hold the turn-key for alleviating the neurological sequelae associated with CI/R and hence could be considered for therapeutic opportunity.

PARP-1, SIRT1, rRNAs and Nucleolar Integrity in Synaptic Plasticity—An Emerging Leitmotif

Ribosomal RNA (rRNA) synthesis (28S, 18S, and 5.8S ribosomal components) is a pre-requisite for long-term synaptic plasticity (LTSP). Synaptic plasticity is a pre-eminent determinant of cognitive function. LTSP is integral to LTP and underlies long-term memory (LTM) formation. rRNA synthesis constitutes the rate-limiting step of ribosomal biogenesis that in turn governs nucleolar integrity. While ribosomal biogenesis is central to nucleolus existence, recent studies have proposed a non-ribosomal function concerning synaptic plasticity and memory. This relates to PARP-1-dependent activation of IEG and activity-dependent late phase genes, essential for consolidation of late-phase synaptic plasticity or potentiation (L-LTP), as observed in forskolin-stimulated mouse hippocampal slices (Allen et al. 2014). PARP-1 partakes in ribosomal biogenesis by controlling precursor rRNA processing. This involves targeting of nucleolar proteins to the proximity of precursor rRNA, post-transcriptional modification, and pre-ribosome assembly targeting (Boamah et al. 2012). PARP-1 and PARs regulate both RNA polymerase I-driven transcription and nascent rRNA transcript processing, thereby regulating generation of new and qualitatively different plasticity-dependent ribosomes. Subsequently, the emergent RNA granules are transferred to active synapses wherein memory is consolidated through associated local protein synthesis. Indeed, L-LTP is hampered following selective inhibition of RNA polymerase I due to nucleolar fragmentation, and eviction of PARP-1 from functional nucleolar compartments (Boamah et al. 2012). Further, with reference to LTF, rRNA genes act as PARP-1-dependent effector genes, as demonstrated in isolated ganglia from serotonin (5-HT) exposed mice (Sharma 2010). 5-HT activates PARP-1 that induces a long-lasting and dynamic increase in expression of transcripts of rRNA genes, supporting ribosomal biogenesis for LTF of sensori-motor synapases (Hernández et al. 2009). Loss of nucleolar PARP-1 is of deleterious consequence, as evidenced by synaptic dysfunction and memory impairment in post-mortem brains of Alzheimer’s disease (AD) patients (Zeng et al. 2016). This is marked by a decline in hippocampal pyramidal neurons positive for PARP-1 and nucleolar marker fibrillarin. Also, a decrease in ribosomes in the inferior parietal lobe is typical of AD brains, owing to epigenetic silencing of rDNA by DNA methylation that inhibits rRNA gene expression and ribosomal biogenesis (Zeng et al. 2016). In this connection, it would be interesting to explore the influence of nucleolar integrity intertwined with PARP-1 activity and NAD⁺ network on LTSP in CI/R paradigm.

As for SIRT1, in conjunction with suppressor of variegation 3–9 Homolog 1 (Suv39H1), a H3K9me3 writing epigenetic enzyme, SIRT1 forms a complex viz., eNoSC (energy-dependent nucleolar silencing complex) with nucleomethylin (NML). This complex regulates the transcriptional responsiveness of rDNA loci through heterochromatin formation that involves H3 deacetylation and H3K9 dimethylation by sensing cellular energy status (Murayama et al. 2008) (Fig. 6d). This transcriptional module restricts or suspends rDNA transcription. This aspect links SIRT1 and Suv39H1-mediated energy-dependent transcriptional repression, restoration of energy balance and genome protection. A cohesive link between NMNAT1 and rRNA transcription is provided by paucity of energy that drives NMNAT1 recruitment to eNoSC. This enables generation of a finite pool of NAD⁺ for SIRT1 activity with subsequent establishment of heterochromatin (Song et al. 2013) (Fig. 6d). This paradigm is pertinent to an ischemic brain that is indeed emblematic of glucose deprivation. Further, implications of decreased rRNA synthesis in the context of neuronal plasticity and cognition/memory processes that could elucidate their plausible footprint in an ischemic brain are yet to be deciphered.

PARP-1 and SIRT1 in Component Processes of Long-Term Memory—Emerging Druggable Opportunities

An aspect that is increasingly relevant to IS recovery and rehabilitation is synaptic plasticity, that is necessary to surmount the ‘post-stroke depression’ (Calabresi et al. 2003; Pekna et al. 2012; Das and Rajanikant 2018). During CI/R, memory processing is impacted by the disruptions in local network activity in hippocampal and extra-hippocampal components like Papez circuit (Escobar et al. 2019). A time-limited window of neuroplasticity opens up in post-stroke epoch that is very much similar to that during development (Murphy and Corbett 2009). This critical period is endowed with the ability for true recovery and behavioral compensation through formation of new structural and functional circuits, that serve to remap the putative cortical regions. CI induces protective and reparative forms of neuronal plasticity that could essentially contribute to ischemic tolerance and post-stroke recovery (Calabresi et al. 2003). The pathological synaptic plasticity induced by CI/R contributes to delayed neuronal death in hippocampus and striatum and spreading of cortical infarct to penumbra. Persistent increase in synaptic strength defines LTP and activity-dependent LTP of excitatory neurotransmission is central to associative learning and memory formation (Raymond 2007). PARP-1 and SIRTs partake in these processes and an increased comprehension of the role of PARP-1 and SIRTs in connection with LTM formation would enable the identification of targets amenable for therapeutic intervention. This could plausibly avert the cognitive sequelae following IS and aid in preserving cerebral function during CI/R.

PARP-1 and LTM

Post-translational modifications like phosphorylation, acetylation, methylation, ubiquitination, SUMOylation play an important role in LTM formation (Sharma 2010; Sun and Kennedy 2013). Following suit, PARP-1-dependent PARylation is essential for LTM acquired during learning (Cohen-Armon et al. 2004; Goldberg et al. 2009). In addition to high affinity binding to DNA breaks, PARP-1 activation also occurs in response to physiological stimuli and associative training procedures that elicit LTM (Visochek et al. 2005; Hernández et al. 2009; Wang SH et al. 2012). Memory stabilization or consolidation requires changes in synaptic plasticity, which, in part, is achieved by PARP-1 activity and PARylation of proteins over a specific time-course. Activity-dependent LTSP requires additional gene expression and protein synthesis that increasingly relies on chromatin dynamics (Sutton and Schuman 2006; Richter and Klann 2009). In this order, the underlying mechanism for long-term sustenance of synaptic changes, in part, involves (i) PARP-1-mediated fast and transient decondensation of chromatin, that unplugs the expression of genes governing LTSP and hence LTM (Rouleau et al, 2004) and (ii) PARylation of nuclear TFs and recruitment of additional regulatory binding proteins at promoters of memory-related genes. PARP-1 is preponderant at specific DNA loci that correspond with subsets of actively transcribed genes. It promotes binding of RNA polymerase II and additional transcriptional machinery at corresponding promoters (Schreiber et al. 2006).

PARP-1 partakes only in long-term behavioral sensitization (LTS), an analog of LTM (Cohen-Armon et al. 2004; Goldberg et al. 2009). It does not necessarily correlate with all forms of synaptic plasticity like long term depression (LTD) evoked by inhibitory neuropeptides and short-term sensitization induced by single noxious stimulus. An essential contribution of PARP-1 to LTM and spatial memory formation becomes evident from two operant-conditioned learning tasks in Aplysia (Cohen-Armon et al. 2004). These involve (i) LTS of defensive withdrawal reflexes from recurrent and multi-positioned noxious stimuli in pleuro-pedal ganglia, and (ii) conditioning of food response in buccal and cerebral ganglia, paired with reinforcing negative stimulus. The former involves LTS-stimulated release of neurotransmitter 5-HT, that facilitates sensory to motor neuron synapses and essentially involves PARP-1 activity. In the latter task, there is consolidation of learning that supposedly entails PARP-1-dependent gene expression and development of LTM. PARP-1 inhibition with 3-aminobenzamide (3-AB) prior to training reportedly does not affect survival or behavior and only affects LTM, while inhibition after training does not affect LTM. Similar observations were made in mammals in relation with two distinct learning paradigms like object recognition and passive avoidance task. Also, PARP-1 suppression in CNS hinders LTM formation while leaving short term memory unaffected (Goldberg et al. 2009). PARP-1-dependent PARylation is also a vital cog in gene expression-dependent memory processes like contextual-fear memory consolidation and reconsolidation in hippocampus and long-term contextual-fear memory extinction in medial pre-frontal cortex (mPFC) (Inaba et al. 2015). Inhibition of dorsal hippocampal and mPFC PARP-1 with 3-AB or PJ34 prior to training disrupts the above memory processes, with no effect on locomotor activity or anxiety-related behavior. This is due to activation of PARP-1 during early phase of memory consolidation.

SIRTs and Synaptic Plasticity

Akin to PARP-1, SIRT1 is indispensable for learning and memory formation and regulates synaptic plasticity. SIRT1 KO mice are characterized by impaired cognitive features associated with immediate memory, classical conditioning and spatial learning. These ostensibly relate to decreased short and long-term hippocampus-dependent memory (Michán et al. 2010). While basal synaptic transmission, NMDAR function and dendritic spine structure are virtually unaffected in SIRT1 KO brain; dendritic branching, branch length and complexity of dendritic arbors significantly decrease. SIRT1 inhibition also impairs synaptic plasticity and hence LTP. In a pathological milieu such as hypoxia–ischemia, SIRT1-mediated attenuation of NF-κB signaling cascade ameliorates hypoxia–ischemia-induced long-term memory deficits following resveratrol administration. It is important to note that alleviation of neuroinflammation correlates with an increased dendritic spine density, synaptic plasticity and synaptic protein expression all of which are integral to learning and LTM formation (Peng et al. 2021). Additionally, SIRT1-mediated inhibition of miR-134 through a repressive complex consisting of transcription factor YY1, upregulates the expression of CREB and brain derived neurotrophic factor (BDNF) genes (Gao et al. 2010) (Fig. 7a). This form of direct engagement of SIRT1 in regulating normal brain function is distinct from the ones related to SIRT1-mediated cell survival. This could provide a cue for future research in an ischemic brain, as CREB activation by NMDAR stimulation generates a growth environment for residential neural precursors, by stimulating BDNF expression (Zhu et al. 2004) (Fig. 7a). This could impact post-stroke neurological outcomes, as CREB-mediated facilitation of NPC survival is integral to experience-dependent changes in neurogenesis in DG, hippocampus-associated learning and spatial learning (Tully et al. 2003). Based on these features, it is tenable to mention that SIRT1 activation constitutes a formidable therapeutic option for correction of IS-related cognitive deficits.

Fig. 7.

a, b Epigenetic underpinnings of PARP-1 and SIRTs involvement in regulating long-term memory acquisition during varied learning paradigms. a PARP-1 activation is essential for long-term synaptic plasticity (LTSP), long-term potentiation (LTP) that underlie hippocampal-associated learning and spatial memory formation. Putative to this are the molecular events involving selective release of histone H1 in PARP-1 dependent manner, at promoters of cAMP-response-element-binding protein (CREB)-dependent (immediate-early genes (IEG)-c-fos, zif268, arc) and NF-κB dependent genes associated with learning and memory. Coincident to this is redistribution of heterochromatin protein 1γ (HP1γ) and methyl CpG-binding protein (MeCP2). The same is recapitulated following neuronal stimulation with potassium chloride (KCl)-induced membrane depolarization, hormones and growth factors. SIRT1 is essential in memory processes and for neurogenesis and neural precursor survival. SIRT1-mediated inhibition of miR-134 through a repressive complex consisting of transcription factor YY1 upregulates the expression of CREB and brain derived neurotrophic factor (BDNF) genes. This facilitates LTSP essential for learning and spatial memory formation (Refer “PARP-1, Sirtuins and Chromatin Dynamics in LTM Formation” section for additional details). b Nerve growth factor (NGF) invokes hippocampal-dependent long-term memory (LTM) and synaptic plasticity through PARP-1-mediated protein PARylation and activation of PKA-CREB pathway. A panoply of post-translational modifications, including PARylation, phosphorylation and acetylation of various targets stimulate target gene expression. Along with PARP-1, extracellular signal -regulated kinase-2 (ERK2) is implicated in synaptic plasticity and LTM acquisition. ERK2 translocates, binds to and activates PARP-1 in nucleus to replace histone H1 by PARylation, thereby gaining access to substrates like transcription factors-Elk1 and CREB. Phospho-ERK2 phosphorylates and activates Elk1 with core histone acetylation and Elk1 target gene c-fos activation. A coincident ERK2-mediated phosphorylation and activation of histone acetyltransferase (HAT) activity of p300/CBP and recruitment of histone H4 results in local unwinding of chromatin. Following this, is an active transcription of immediate early genes (IEGs) like c-fos, zif268 and arc, essential for synaptic plasticity and LTP. SIRT1 also induces ERK1/2 phosphorylation and regulates hippocampal genes related to synaptic function, lipid metabolism and myelination that impacts the maintenance of dendritic structures (Refer “PARP-1, Sirtuins and Chromatin Dynamics in LTM Formation” section for additional details)

In relation with memory processes, long-term contextual fear memory is enhanced following SIRT6 deletion. Disrupted fear conditioning, owing to SIRT6 overexpression is observed in hippocampal CA1 of mice subjected to Pavlonian fear conditioning paradigms that increasingly rely on optimal hippocampal functioning (Yin et al. 2016). An impaired contextual fear conditioning without virtually affecting spatial memory, is associated with SIRT6 overexpression in forebrain excitatory neurons. Additionally, SIRT6 as a negative modulator of LT fear memory formation, markedly reduces expression of proteins related to IGF-2 signaling pathway (Akt/p-Akt, mTOR/p-mTOR). As regards mitochondrial sirtuins, SIRT3 deficiency is also associated with a poor remote memory and dampened synaptic activity that could be mapped to reduced neuronal number in anterior cingulate cortex (Kim et al. 2017). Taken together, there seems to be a functional interplay of PARP-1 and sirtuins in regulating LTP. This has not been cogently explored in an ischemic brain and could present a slew of therapeutic opportunities, when investigated in relation with NAD⁺ availability and functioning of other major components of NAD biosynthesis network.

PARP-1, Sirtuins and Chromatin Dynamics in LTM Formation

PARP-1 plays a dual role of chromatin ‘architectural protein’ through its specific binding to nucleosomes and that of a histone ‘chaperone’ without affecting core histones or assembly of nucleosomes through its intrinsic enzymatic activity (Kim et al. 2004; Kraus 2008; Muthurajan et al. 2014; Ummarino et al. 2021). Indeed, its abundance to the tune of 106 molecules/cell accords it the status of a ‘structural component’ and modulator of chromatin in vivo, as its incorporation into chromatin generates higher order chromatin structures. Arguably, the chaperoning function of PARP-1 is ascribed to higher affinity of PAR for H1 compared to DNA per se (Beneke 2012; Ummarino et al. 2021). In relation with LTM, the stringent requirement of PARP-1 for memory consolidation and LTM preservation is based on the reciprocal binding dynamics of PARP-1 and linker histone H1. In line with this, treatment of pleuro-pedal ganglia with serotonin (5-HT) induces H1 PARylation and substitution with PARP-1, thereby generating relaxed chromatin. This renders DNA transiently accessible for transcription of LTM-related genes and hence long-term facilitation (LTF) (Cohen-Armon et al. 2004; Kraus 2008). In this connection, PARylation of PARP-1 and H1 could be detected in cerebral cortex and hippocampus immediately after training session in mammals (Goldberg et al. 2009). This associative maneuver also provides an epigenetic premise for PARP-1 activity through a canonical model termed as ‘histone shuttling’. It generates a relaxed chromatin conformation that corresponds with active transcriptional states and epigenetic memory (Krishnakumar et al. 2008; Strickfaden et al. 2016; Ciccarone et al. 2017).

Similar to LTSP, consolidation of object recognition memory requires PARP-1 activation in hippocampus. This essentially involves PARP-1-mediated H1 PARylation and associated local chromatin changes driving H1 eviction. H1 PARylation in the hippocampus occurs after acquisition period, with H1 being selectively released in a PARP-1 dependent manner at promoters of CREB and NF-κB- dependent genes, associated with learning and memory. Prominent among such rapidly and selectively upregulated CREB-dependent genes are immediate early genes (IEGs) viz., c-fos, c-jun, zif268, BDNF, Egr1 (Cohen-Armon et al. 2004; Fontán-Lozano et al. 2010). Release of H1 is however repressed following PARP-1 inhibition or deletion, phosphorylation inhibition or H1 becoming non-PARylatable or phosphorylatable. PARP-1-induced CREB-dependent IEG expression (c-fos, c-jun, BDNF, Egr1) is also canonical to neuronal stimulation by potassium chloride-induced membrane depolarization (Azad et al. 2018) (Fig. 7a). Therein, PARP-1 regulates hyperdynamic binding of H1, that links neuronal depolarization and chromatin plasticity through CREB-IEG expression. Depolarization also induces global rearrangement of heterochromatin, in addition to modulation of H1 binding dynamics by PARP-1. This involves redistribution of H3K9 trimethylation marks, heterochromatin protein 1γ and methyl CpG-binding protein (MeCP2) in stimulated neurons (Azad et al. 2018). In a similar vein, sensory experience-dependent neuronal stimulation is also associated with Ca²⁺-dependent triggering of signal transduction pathways with eventual transcriptional activation of IEGs (c-fos, c-jun, Arc) (Fig. 7a).

IEGs are quintessential in synaptic plasticity and hence memory trace formation. IEG positive neurons encode and store information acquired during learning which is used during memory recall (Minatohara et al. 2016; Carmichael and Henley 2018). In relation with IS, IEGs induction reportedly enhances brain plasticity and behavioral recovery following stroke (Sharp et al. 2000). IEGs like c-fos and c-jun are expressed in the contralateral hemisphere after focal ischemia. They are induced diffusively in the non-ischemic regions of ipsilateral hemisphere in MCAO rats, owing to spreading depression (Takemoto et al. 1995; Weinstein et al. 2004). c-fos expression is required for induction of doubly dissociable memory processes viz., consolidation and reconsolidation along with extinction of contextual fear memory (Inaba et al. 2015). PARP-1 transcriptionally activates c-fos expression by impinging on cAMP-PKA-CREB axis. It is noteworthy that inhibiting c-fos expression following infusion of 3-AB into hippocampus and mPFC impedes synaptic plasticity events. In addition to c-fos, there is a specific requirement for BDNF expression for memory consolidation and zif-268 for brain plasticity, learning and memory, both of which are positively regulated by PARP-1 activity (Bozon et al. 2003; Lee et al. 2004). At this juncture it is interesting to mention that CAMKII-NAMPT^−/− mice exhibit a decreased expression of IEGs and a defective LTD component of LTM (Stein et al. 2015). It is also worth mentioning that CREB activity and NMNAT2 expression are downregulated prior to neurodegeneration like tauopathy. This can affect PARP-1-dependent activation of CREB-related IEGs that would necessarily impact the component processes of memory (Ljungberg et al. 2012). Therein, it would be interesting to decipher if any functional interplay of PARP-1 and NMNATs is involved in conferring the constellation of phenotypes in CAMKII-NAMPT^−/− mice, given the unifying feature of NAD⁺-related energy deprivation and reciprocal involvement of PARP-1 and NAMPT in LTD. Given the prominence of all these above discussed molecular entities per se and their interactions in an ischemic brain, such studies could provide cues for implications of NAMPT downregulation on PARP-1. This could also inform about resultant steering of transcriptional programs thereof, eventually affecting the cognitive and motor deficits in CI/R paradigm.

Along with PARP-1, there is also a formidable coplayer viz., extracellular signal -regulated kinase-2 (ERK2), the phosphorylated form of which (pERK2) associates with synaptic plasticity and LTM acquisition (Cohen-Armon et al. 2007). In response to an exposure to neurotrophins and neurotrophic peptides, PARP-1 is highly activated that promotes neurite outgrowth and synaptic plasticity (Visochek et al. 2005; Visochek et al. 2016). In line with this, nerve growth factor (NGF) administration facilitates hippocampal-dependent LTM and synaptic plasticity through PARP-1-mediated protein PARylation and activation of PKA-CREB pathway in cortical neurons. This is abrogated following co-application of 3-AB. In an intertwined manner, ERK2 translocates, binds to and activates PARP-1 in the nucleus to displace H1 by PARylation following high frequency electric stimulation and also in cortical neurons treated with NGF (Visochek et al. 2016). This provides facilitated access of PARP-1-bound pERK2 to substrates like TFs-Elk1 and CREB (Fig. 7b). Concurrently, PTMs including PARylation, phosphorylation and acetylation of discrete entities potentiate target gene expression. In NGF-treated cortical neurons, pERK2 phosphorylates and activates Elk1, together with core histone acetylation and Elk1 target gene c-fos activation (Cohen-Armon et al. 2007). As regards CREB, a coincident ERK2-mediated phosphorylation and activation of HAT activity of p300/CBP and H4 recruitment results in local unwinding of chromatin. Following this, an active transcription of IEGs like c-fos, zif268 and arc ensues, all of which are crucial for synaptic plasticity and LTP (Visochek et al. 2016) (Fig. 7b). All these events are evidenced in hippocampus, wherein LTP of CA3-CA1 synaptic connections is accepted as a competent cellular model of LTP. Also, the criticality of PARP-1 requirement for LTP is evidenced from PARP-1-KO, inhibition or silencing that impairs LT synaptic potentiation in CA1-CA3 hippocampal connections, a linear correlate of impaired learning. However, single-stranded DNA breaks could disrupt PARP-1-ERK2 synergism as binding of PARP-1 to nicked DNA occludes ERK2 binding sites (Visochek et al. 2016). Overlapping with the target specificity of PARP-1, SIRT1 also induces ERK1/2 phosphorylation. This is reduced following SIRT1 KO, thereby altering the expression of hippocampal genes related to synaptic function, lipid metabolism and myelination and distorting dendritic structures (Abe-Higuchi et al. 2006) (Fig. 7b). It is nevertheless interesting to note that SIRT1-dependent hippocampal ERK2 phosphorylation associates with an anti-depressive behavior. From the above discussion it would be tenable to assume that the penalty imposed by stroke on synaptic plasticity and its repercussions in the form of impaired sensory and motor responses, could supposedly be rectified by PARP-1-dependent activation of IEGs and also SIRTs modulation with an overall effect of LTM restoration and forging of LTP.

In conjunction with the consideration that PARylation constitutes an early response to DNA damage, it needs to be recounted through the above discussion that secondary messenger cascades invoked by sensory experience-induced LTM can also initiate PARylation. Nevertheless, the activating mechanisms could be disparate involving different secondary messenger pathways, owing to varied neuronal types. An attempt could be made to delineate the underlying mechanism through electric stimulus-induced depolarization of rat cortical neurons that promotes Ca²⁺ mobilization to nucleus in a PKC-dependent manner. It is not yet clarified if serotonin induces nuclear rise in Ca²⁺. 5-HT induces expression of ubiquitin C-terminal hydrolase as an early response gene (Cohen-Armon et al. 2004). An increase in ubiquitin-mediated proteolysis of regulatory subunits of cAMP-dependent protein kinase produces a persistently active kinase required for LTF in sensory cells. A parallelism and coordination could be implied here, as PARP-1 activation also affects proteolysis by nuclear proteasomes. In addition, an intertwined role of nitric oxide in multiple memory processes post-learning, that could be possibly generated through its action on guanylyl cyclase is also implicated (Katzoff et al. 2002). As nitric oxide itself functions as an epigenetic modulator it would be interesting to decipher its role in potentiating cognition or behavioral processes in an ischemic brain (Narne et al. 2019). Taken together, it could be inferred that molecular mechanisms underlying associative learning and memory formation could subsume PARP-1 and SIRT-mediated cell’s response to stress or injury. These molecular intricacies also provide opportunities for therapeutic targeting in the direction of cognition preservation in a pathological setting like CI/R.

Flavin Adenine Dinucleotide and Lysine Specific Demethylases—Emerging Therapeutic Opportunities in Counteracting CI/R injury

FAD Dynamics

The dihydroflavins flavin adenine mono/dinucleotide (FMN/FAD) are essential cofactors for the functionality of riboflavin (precursor for FMN and FAD)-based coenzymes i.e., flavozymes which primarily catalyze reactions pertaining to oxidative metabolism of carbohydrates, aminoacids, fatty acids and mitochondrial electron transport chain (Walsh and Wencewicz 2013). In addition, a significant proportion of flavoproteins (dehydrogenases, reductases, oxidases) utilize FAD as a cofactor for effecting a plethora of chemical transformations in secondary cellular metabolic pathways and diverse regulatory processes encompassing ROS generation, protein folding, antioxidant defense and chromatin remodeling (Barile et al. 2013; Berndt et al. 2020; Tolomeo et al. 2021). Perhaps, they typically contrast nicotinamide coenzymes (NAD(P)H/NAD(P)) that predominate the catalysis realm in primary metabolic pathways (Walsh and Wencewicz 2013; Giancaspero et al. 2013; Barile et al. 2016). An enzyme-tailoring paradigm such as a redox reaction catalyzed by flavins, is majorly constituted by oxidation reactions of amine and alcohol groups and desaturation to olefins. It characteristically entails two reductive half reactions. The latter is attributed to the property of FAD and FMN to remain attached to protein partner after each catalytic turnover unlike nicotinamide coenzymes (Walsh and Wencewicz 2013; Giancaspero et al. 2013; Barile et al. 2016). This exigency is principally evolutionarily driven, given the lability of FAD and FMN to undergo rapid and un-controlled auto-oxidation in the absence of a controlled microenvironment like an enzyme active site.

The enzyme synthesizing FAD i.e., FAD synthase (FADS) occurs both in cytosol and mitochondria underlying the fact that a dynamic pool of flavin cofactors occur in the nucleus (Giancaspero et al. 2013). Albeit its mitochondrial preponderance, recent studies have suggested an extra-mitochondrial, rather nuclear source of FAD (Giancaspero et al. 2013; Barile et al. 2016). Flavin cofactor homeostasis is primarily governed by the coordinate action of varied FADS isoforms that drive subcellular compartmentalization of FAD biosynthesis and FAD-degrading FAD pyrophosphatase. In addition, molecular components enabling flavin-trafficking across the subcellular membranes also populate the ‘flavin network’ that, inter alia, is responsible for regulation of cellular energy balance. Studies have indicated that deficiency of flavin enzymes and impaired FAD homeostasis are complicit in cardiovascular diseases, anemia, cancer, abnormal fetal development, and neuromuscular pathologies and neurological disorders (Barile et al. 2013; Berndt et al. 2020; Tolomeo et al. 2021). In relation with brain hypoxia, recording FAD fluorescence as a measure of mitochondrial redox state is emerging as a valuable addition to multimodal bed-side monitoring modalities like oxygen availability and subdural electrocorticography, that would enable patient risk stratification (Berndt et al. 2020). Hence, sufficient understanding of factors affecting FAD homeostasis is warranted, as it formidably aligns with the metabolic upshots arising from steering transcriptional networks through epigenetic mechanisms. This in turn would necessarily illuminate and inform about the features linked with identification of pharmacologically amenable states.

Lysine-Specific Demethylase 1 (LSD1) and Neuronal Milieu

The importance of FAD(H₂) in intermediary metabolism that in turn influences chromatin landscape remodeling, arises from its function as a redox coenzyme in OXPHOS and as a cofactor for lysine specific demethylases (LSDs) (Hino et al. 2012; Kim et al. 2021). LSD1 (KDM1A), one of the prominent members of FAD⁺-dependent amine oxidases and hence the moniker ‘client flavoprotein’, specifically targets the mono and dimethylated H3K4. LSD1, also known as AOF2 or BHC110, utilizes FAD as a coenzyme for removal of one or two methyl groups from a specific mono- or di-methylated lysine side chain of numerous nuclear proteins, including histone H3. The reaction involves two electron oxidation of lysine N-methyl amine group combined with two electron reduction of FAD, eventually generating a demethylated lysine and FADH₂ (Forneris et al. 2005). FADH₂ is reoxidized by molecular oxygen to generate H₂O₂ that can oxidize dG nucleobases in the vicinity forming 8-oxo-deoxy guanosine (8-oxo-dG) lesions which can hamper gene activation. LSD1 binds tetrahydrofolate (THF) and in the absence of the latter, formaldehyde is generated, whose nuclear toxicity is expected to be circumvented through its reduction and recycling to the major nuclear methyl group donor viz., S-adenosyl methionine (SAM). The non-availability of a free electron pair on amine nitrogen for catalysis prevents LSD1 from demethylating H3K4me3. These marks localize at the 5’ end of active genes involving transcription start sites and enhancers, and associate with functional form of RNA polymerase II. Thereby, LSD1 activity invokes transcriptional activation or repression in a context-dependent manner (Forneris et al. 2005; Ouyang et al. 2009; Zheng et al. 2015). LSD1 functions as a transcriptional co-activator through H3K9me1, me2 demethylation or as a suppressor by demethylating H3K4me1 and H3K4me2. This relies upon binding of LSH ternary complex (LSD1-CoREST-HDAC1/2 together with SUMO-2) to the core histones and nucleosomal substrates (Ouyang et al. 2009; Zheng et al. 2015). LSD1-mediated H3K4 demethylation is regulated by post-translational sumoylation of LSD1 and negatively controlled in a trans manner by H3T11 phosphorylation (Metzger et al. 2010).

In neuronal realm, LSD1 + 8a, an LSD1 isoform in conjunction with an interacting protein viz., supervillin mediates H3K9 demethylation and its inhibition hampers neuronal differentiation (Laurent et al. 2015) (Fig. 8a). LSD1 also physically associates with HEYL gene and represses it through demethylation of H3K4me2, during human fetal NSC neuronal differentiation (Hirano and Namihira 2016) (Fig. 8b). In a putative milieu, LSD1 elimination has been shown to foster ESC differentiation towards neural lineage. This is ascribed to LSD1 destabilization post-transcriptionally by a major LSD1 negative regulator viz., Jade-2, a E3 ubiquitin ligase (Han et al. 2014) (Fig. 8c). This feature serves as an antibraking system and adaptively steers epigenetic landscape, obviating the need for elaborate transcriptional changes. As regards neurogenesis, LSD1 that can control the development of cortical neurons in conjunction with CoREST and corepressor Rcor2 (Fuentes et al. 2012; Wang Y et al. 2016b). A guidance for LSD-mediated transcriptional activation and repression of NSC-related genes and hence NSC proliferation and progeny specification is provided by a LSD1 corepressor Rcor2 that is expressed in CNS. Inhibition of Rcor impedes NPC proliferation, reduces neuron population and decreases neocortex thickness and brain size (Wang Y et al. 2016b). LSD1 is also recruited to the promoters of cell proliferation regulator genes by Tlx (a nuclear receptor), an essential NSC regulator to repress their expression. LSD1 inhibition thereby precludes NSC proliferation in hippocampal DG of wild-type adult mouse brains (Sun et al. 2010). LSD1 regulates the proliferation of embryonic stem cells (ESCs) through acetylation of H4K16, a critical substrate of HDAC1 (Yin et al. 2014). LSD1 is posited to protect hippocampal and cortical regions in adult mice and LSD1 loss precipitates learning and memory deficits. Underlying this are LSD1-mediated transcriptional alterations in neurodegenerative pathways with concomitant activation of stem cell genes in degenerating hippocampus. These gene expression changes parallel those observed in AD and fronto-temporal dementia cases (Christopher et al. 2017). LSD1 function seems to be attenuated owing to its mis-localization in the pathological protein aggregates associated with these neuropathologies. In line with this, LSD1 overexpression in hippocampus of tauopathy mice significantly attenuates neurodegeneration and abrogate tau-induced expression changes (Engstrom et al. 2020). Taken together, given these formidable postulations of LSD1 significance in developing brain and neurodegeneration, it would be interesting to explore if LSD1 inhibition or over-activation would elicit transcriptional consequences in neurogenic niches and salvageable penumbral zone of an ischemic brain.

Fig. 8.

a–e Role of Flavin adenine dinucleotide (FAD) and lysine-specific demethylase 1 (LSD1) in regulation of gene expression concerning neurogenesis and molecular metabolism. FAD-dependent LSD1 acts as a transcriptional co-activator through mono and demethylated histone 3 lysine 9 (H3K9me1 and H3K9me2, respectively) demethylation or as a suppressor by demethylating H3K4me1, H3K9me2. a–c LSD1 in neural milieu: LSD1 modulates neuronal differentiation by a transcriptional upregulation of neuronal genes through decreased H3K9 demethylation in association with supervillin and reciprocal inhibition of non-neuronal genes, b inhibition of HEYL expression in conjunction with transcription factor RBP-J, with concurrent increase in H3K4 demethylation, abrogation of which ensures maintenance of un-differentiated state, c undergoing destabilization through jade-2-mediated polyubiquitination and degradation. Removal of LSD1 therein facilitates enhanced neuronal gene (Pax3, Ascl1, Zic1, Neurog1) expression that promotes neuronal differentiation and maturation. (Refer “Lysine-Specific Demethylase 1 (LSD1) and Neuronal Milieu” section for more details) d, e LSD1 in metabolic milieu: FAD availability determines LSD1 activity in varied metabolic scenarios. d During early hypoxia, LSD1 stabilizes hypoxia-inducible factor 1-alpha (HIF-1α) by demethylation of RACK1 [ubiquitination machinery (Elongin C complex) component] enabling its nuclear translocation and expression of HIF-1α target genes. Severe hypoxia limits FAD availability, reducing LSD1 activity and effects ubiquitin-mediated HIF-1α degradation. LSD1 can also stabilize HIF-1α by removal of H3K4 methyltransferase viz., SET domain-containing lysine methyltransferase 7/9 (Set7/9)-mediated methylation marks on HIF-1α that promotes tumor angiogenesis. e LSD1 promotes a metabolic shift/reprogramming by promoting HIF-1α-mediated increase in expression of GLUT1 and glycolytic enzymes in carcinoma cells. There is a concurrent reduction in oxidative metabolism owing to decreased expression of genes concerning mitochondrial metabolism, with an underlying LSD1-mediated decrease in transcription- activating H3K4me,me2 marks. LSD1 directs transcriptional repression of PGC-1α, PDK4, FATP1 and ATGL genes in adipocytes. LSD1 also inhibits gluconeogenic genes viz., FBP1 and G6Pase through H3K4me2 demethylation and affects glucose delivery during reperfusion (Refer to “LSD1 and Metabolic Realm—Plausible Therapeutic Strategy in CI/R Milieu” section for more details)

LSD1 and Metabolic Realm—Plausible Therapeutic Strategy in CI/R Milieu

Until recently, LSD1 activity was considered to subsist on mitochondrial FAD pool generated de novo from riboflavins and related activity of Krebs cycle, ETC and fatty acid β-oxidation, rendering LSD1 sensitive to fluctuating FAD/FADH₂ ratio. In relation with this, either FAD depletion or selective LSD1 inhibition blocks selective occupancy of target gene promoters by LSD1, which reaffirms FAD-dependency of LSD1 activity (Forneris et al. 2005). Added to this, flavoproteins like SDH, 2-OGDH, acyl CoA dehydrogenase, PDH subunit and AIF are preponderant in mitochondria. Consequently, FAD reserve available for nuclear flavoenzymes is rather finite, presumably causing LSD1 activity to be affected by modest FAD restriction. Nevertheless, subcellular compartmentation seems to be relevant, with the proposed existence of nuclear FAD pool that is essentially maintained by nuclear FADS and degrading enzymes (Giancaspero et al. 2013).

Another level of complexity is added to LSD1-mediated gene (dys) regulation by oxygen availability, a pre-requisite for FAD recycling through conversion of molecular O₂ to H₂O₂ (Walsh and Wencewicz 2013; Kim et al. 2021). This also serves to implicate hypoxia in epigenetic regulation of genes related to oxidative metabolism (Sims and Muyderman 2010; Xie N et al. 2020). Also, FAD synthesis from riboflavin is ATP-dependent and ATP depletion is a bona fide consequence of mitochondrial dysfunction during CI. In this regard, it could be surmised that reduced FAD availability impedes LSD1 activity during hypoxia (Yang et al. 2017). This theme is observed in a hypoxic setting, that outlines the ability of LSD1 to stabilize HIF-1α by preventing oxygen-independent degradation of HIF-1α (Lee et al. 2017). This occurs by LSD-mediated demethylation of RACK1 protein, a component of HIF ubiquitination machinery. The rate-limiting role of FAD in this context becomes apparent, as FAD levels decline under conditions of prolonged hypoxia resulting in reduced LSD1 activity. This in turn downregulates HIF-1α in later stages of hypoxia, with an ensuing metabolic shift. This comes further evident in a significant correlation between HIF-1α signature and expression of FAD-biosynthetic enzymes and LSD1 target genes in triple-negative breast cancers (Yang et al. 2017) (Fig. 8d). LSD1 can also stabilize HIF-1α that is methylated by Set7/9 on K32 and K391 residues in close proximity with oxygen-dependent degradation domain and prevent it from undergoing ubiquitin-mediated degradation (Lee et al. 2017) (Fig. 8d). Nevertheless, an increased LSD1 expression and distribution have been reported in neuron enriched hippocampal CA1 field and DG (region relatively resistant to neuronal death) and cerebral cortex following CI (Zhang YZ et al. 2010). This increase was rather rapid (within 1 h of ischemia) in CA1 and DG fields while it was relatively delayed (6 h post ischemia) in cortical neurons that distinctively correlated with H3K4me, me2, me3 patterns. Perhaps, this spatial and temporal variation in peaking of LSD1 activity signifies differential neuronal viability in the hippocampal and cortical regions following CI. In addition to this, knockdown or pharmacological inhibition of LSD1 by phenelzine and tranylcypromine (TCP) has been reported to confer neuroprotection, by antagonizing neuronal cell death. This was contrasted by a paradoxical absence of worsened ischemic injury following exogenous LSD1 overexpression. An increased LSD1 expression in an ischemic brain could also influence methylation status of histones that can be alluded to its SAM generating ability in the absence of THF as mentioned earlier (Forneris et al. 2005). In an I/R milieu there is a substantial reduction in the levels of THF that could signify an increased generation of SAM, which in turn could posit an aberrant methylation of selected genes (Narne et al. 2017a).

Concerning energy metabolism, LSD1 induces transcriptional repression of a distinctive set of target genes viz., PGC-1α, PDK4, FATP1 and ATGL in adipocytes that regulate energy expenditure and oxidative metabolism involving mitochondrial respiration (Hino et al. 2012) (Fig. 8e). LSD1 also emerged as a significant player in effecting a metabolic shift/reprogramming through epigenome-metabolism cross-talk that in turn confers epigenetic plasticity in an alternative paradigm (Sakamoto et al. 2015). In hepatocellular carcinoma cells, LSD1 inhibition reduces glucose uptake, halts glycolysis and promotes mitochondrial respiration. This observation is supported by a concomitant HIF-1α inactivation with consequent reduction in expression of GLUT-1 and glycolytic enzymes (Sakamoto et al. 2015). Concurrently, there is an increased activation of mitochondrial metabolism genes with an increased H3K4 methylation in the corresponding gene promoters, with a significant and consistent overexpression of LSD1 and GLUT1 in carcinoma tissues (Fig. 8e). A similar occurrence could be envisaged in vulnerable neurons with an increased LSD1 expression and compromised oxidative mitochondrial metabolism. A glycolytic shift is also conceivable leading to an increased lactate accumulation in peri-infarct area and precipitation of energy failure. Also, LSD1 activity can suspend glucose substrate delivery by inhibiting gluconeogenic genes viz., FBP1 and G6Pase through H3K4me2 demethylation during reperfusion (Pan et al. 2013) (Fig. 8e). Further, LSD1 and SIRT1 represent reciprocal and centralized switches for energy homeostasis and forge an epigenetic connection with intermediary metabolism. LSD1 is diametrically opposed to SIRT1 in regulating PGC-1α gene, as the former negatively regulates PGC-1α and the latter promotes mitochondrial metabolism through PGC-1α activation (Sims and Muyderman 2010; Fu et al. 2014; Yang JL et al. 2018). This could be an alternative paradigm for PGC-1α activity, in view of the ability of PGC-1α to deploy anti ROS program involving SOD and UCP2 upregulation in hippocampal CA1 subfield following transient global ischemia. This could partially explain the beneficial effects of natural polyphenols like resveratrol, curcumin and quercetin in alleviating CI/R induced damage, as they can inhibit LSD1 and activate SIRT1 under in vitro conditions (Abdulla et al. 2013). LSD1 inhibition may therefore be viewed as a viable strategy for counteracting tissue acidosis in CI/R. However, it warrants a cautionary approach, as brain penetrant LSD1 inhibitors reportedly block memory consolidation and interfere with LTM formation (Neelamegam et al. 2012). LSD1 also shares non-histone protein substrates of SIRT1 viz., p53 and DNMT1 and has been shown to stabilize DNMT1 activity and maintain global DNA methylation (Tsai et al. 2008). LSD1 inhibits p53 by preferentially demethylating K370me2 that prevents the interaction of p53 with tandem Tudor domain of coactivator 53BP1. This disallows p53 binding to DNA thereby inhibiting the expression of p53 responsive genes. Whether the documented benefits of SIRT1 activation and DNMT1 inhibition on metabolic amelioration and cell viability can be mimicked by LSD1 inhibition during CI/R is yet to be ascertained. Further, the non-canonical functions of LSD1 in a demethylase-independent manner enumerates its interactions with an array of proteins in non-catalytic manner. Identification of this interactome further adds another wrinkle in expanding LSD1-based drug discovery towards targeting LSD1 interactome non-catalytically (Gu et al. 2020). Taken together, given the undisputed role of FAD-dependent LSD1 activity and effects in these diverse metabolic scenarios as discussed above, it would be fitting to precisely investigate the role of LSD1 in CI/R brain, afflicted by extensive perturbations in oxygen and glucose supply.

LSD1 Inhibition—Repurposing Therapeutic Agents for CI/R

With an increasing significance of LSD1 in epigenetic modulation of gene activity, it has been identified as one of molecular targets for therapeutic intervention. LSD1 exhibits similarity with flavin-dependent monoamine oxidases (MAO A/B) and polyamine oxidase enzymes, that spurred investigations into the applicability of the corresponding enzyme inhibitors in varied disease scenarios (Prusevich et al. 2014). Identification of MAO inhibitor viz., TCP as a potent LSD1 inhibitor catapulted further development of TCP-based irreversible LSD1 inhibitors. The mechanism-based inactivators viz., TCP and phenelzine and their analogs have been reported for their potent LSD1 inhibition activity and in conferring neuroprotection (Zheng et al. 2015; Prusevich et al. 2014; Tsutsumi et al. 2016). The chemical strategies pertaining to development of an arsenal of LSD1 inhibitors and associated clinical trials for targeting various disease etiologies (majorly cancers) are beyond the scope of this review and have been reviewed in detail in Zheng et al. 2015; Prusevich et al. 2014; Fang et al. 2019). Understanding the role of FAD-dependent LSD1 activity and its implications in IS milieu would enable identification of suitable LSD1 inhibitors for averting pathological implications of metabolic collapse in ischemic brain and for preserving neurological functions in post-stroke epoch.

Outlook

The mitochondrial and nuclear NAD⁺ pools represent critical junctures for metabolic regulation of epigenetic signatures by sirtuins and PARPs, the activities of which are contingent on NAD⁺ availability. In line with this, the prospects of utilizing the components of NAD⁺-synthesizing and metabolizing machinery, endogenous to metabolic pathways and cognitive features in IS realm is gaining traction in pharmacological parlance. The field is upbeat in terms of exploring and establishing salutary effects of NAD⁺ and associated precursors, through NAMPT/NMNATs-mediated NAD⁺ biosynthesis and restoration. These include preserving BBB integrity and cerebrovascular function, promoting neuronal survival, mitigating CI/R injury-induced motor and cognitive dysfunction, and improving neurovascular and behavioral outcomes in post-stroke epoch. Extensive exploration of these therapeutic modalities in pre-clinical models is providing direction for human clinical trials which could effectively circumvent usage of conventional tPA in treating IS.

The idea of ‘NAD world’ concept signifies functional hierarchy and frailty in determining the robustness of systemic NAMPT-mediated NAD synthesis, with an interplay of fine-tuned SIRT1 response in regulating tissue-specific metabolic tone. As neurons are primarily affected by CI/R, this transcriptional-enzymatic arm could provide workable cues in the direction of rescuing them from impending metabolic derangement with neurobehavioral implications in post-stroke epoch. Further, gaining insight into the epigenetic basis of energy dysfunction would enable identification of tractable drug targets and formulate suitable therapies to alleviate the neurological outcomes following IS. Tipping the balance towards energy sparing/preservation and correcting strayed cellular energetics in an ischemic brain by targeting various epigenetic axes comprising NAD⁺/NAD⁺-synthesizing machinery, SIRTs, PARPs and putative molecular entities (as discussed earlier) could bolster future epigenetic therapies concerning IS. These would necessarily hinge on modifying the epigenetic landscape and subsequent steering of transcriptional programs in the direction of mitigating CI/R-associated sequelae.

As NMNATs are associated with neuronal preservation, parsing out the regulatory dynamics between these molecules in an ischemic brain, in the direction of attenuating cognitive deficits and improving behavioral outcomes is warranted. Studying this particular NAD⁺ biosynthetic pathway in varied cell types in an CI/R brain provides critical insight into their vulnerability to pathophysiological stimuli. It could further enable the development of therapeutic interventions for preservation of cellular and BBB integrity. This could possibly extend further to other components of neurovascular unit, given their substantial contribution to preservation of brain functions by functioning in unison. These involve amelioration of endothelial dysfunction, attenuation of astroglial and microglial activation, preservation of cerebrovascular reserve etc. NMNATs, PARP-1 and SIRT1 while acceding to compartmentalization stipulations, regulate neuronal functions including synaptic plasticity, memory formation, cognitive functioning etc., under normal and stress-induced conditions. This also exemplifies their moon-lighting behavior, a norm of evolutionary conservation that probably emanated from the cell’s response to injury or stress. A deeper understanding of such moon-lighting functions through the molecular lens of underlying epigenetic underpinnings (as discussed in text) could provide targets for pharmacological interventions, while preserving integrity of distinct brain cell types. Sufficient exploration of such testable epigenetic theses (as discussed in text) could plausibly enable development of epigenetic therapies in this direction. These could substantially and holistically address the debilitating features of CI/R-induced injury like memory impairment, cognitive dysfunction, often including the predicament of neurodegeneration like Wld in an ischemic brain. Also, an emerging leitmotif concerning the role of NAMPT/SIRT1/PARP-1in ribosomal biogenesis and maintenance of nucleolar integrity is worth investigating for possible therapeutic targets in CI/R paradigm. Further, recent revelations of nuclear localization of FADS allows speculation of additional subcellular localizations of cofactor producing machinery. This would also add an additional level of complexity to the regulation of epigenetic architecture, that would depend on the dynamism of cofactor pools dispersed throughout the cell. Biochemical elucidation of such subcellular compartmentation of NAD⁺ and FAD would enable a deeper comprehension of their role in pathologically relevant (as concerning CI/R) epigenetic events. In relation with NAMPT and neurogenesis, while PC73 compounds offer reason for optimism as proof-of-concept drugs stimulating post-stroke neurogenesis, a moot question would be as to what proportion of such new cells survive in the niches and develop normally with optimal expression of neurotransmitters, integrate into neuronal networks and eventually contribute to functional recovery after stroke. Investigations on epigenetic modalities governing the salutary effects of NAD⁺-restoration and FAD-based LSD1 targeting in CI/R brain, could also address a slew of questions concerning their biology, rapid and dynamic action, and varied points of their integration into multiple pathways as discussed. These could provide necessary cues for expansion of existing array of NAD⁺ and LSD1- based therapeutic repertoire and repurposing them for application in CI/R paradigm. Expanding the arsenal of NAD⁺ donors, NAMPT, NMNATs and SIRT1 activators, PARP-inhibitors, LSD1 inhibitors by exploring the inter-twined relationships between these entities could enrich the prospects of neuroprotection, motor function and cognition preservation in post-stroke epoch.

Abbreviations

2-OGDH

2-oxoglutarate dehydrogenase

5hmC

5-Hydroxymethylcytosine

5-HT

5-Hydroxytryptamine (serotonin)

5mC

5-Methyl cytosine

ADP

Adenosine-di-phosphate

AIF

Apoptosis-inducing factor

AMPK

Adenosine monophosphate-activated protein kinase

ANT

Adenine nucleotide translocase

ATP

Adenosine-tri-phosphate

BBB

Blood brain barrier

CBF

Cerebral blood flow

CI/R

Cerebral ischemia/reperfusion

CKO

Conditional knock-out

CNS

Central nervous system

DG

Dentate gyrus

DRG

Dorsal root ganglion

ER-α

Estrogen receptor-alpha

ETC

Electron transport chain

EZH2

Enhancer of zest homolog-2

FAD

Flavin adenine nucleotide

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

GDH

Glutamate dehydrogenase

HDACs

Histone deacetylases

HIF-1α

Hypoxia inducible factor -1alpha

HKII

Hexokinase II

HR

Hypoxia-reoxygenation

IS

Ischemic stroke

JHDM

Jumonji domain-containing histone demethylase

LSD

Lysine-specific histone demethylase

LTD

Long-term depression

LTF

Long-term facilitation

LTM

Long-term memory

LTP

Long-term potentiation

LTSP

Long-term synaptic plasticity

MCAO

Middle cerebral artery occlusion

miR

Micro-RNA

MMP

Matrix metalloproteinase

MPTP

Mitochondrial permeability transition pore

mTOR

Mammalian target of rapamycin

NAD⁺

β-Nicotinamide adenine dinucleotide

NADH

(NAD + hydrogen)

NAM

Nicotinamide

NAMPT

Nicotinamide phosphoribosyltransferase

NF-κB

Nuclear factor kappa-light-chain-enhancer of activated B cells

NGF

Nerve growth factor

NHEJ

Non-homologous-end-joining

NMDAR

N-methyl-D-aspartate receptor

NMN

Nicotinamide mononucleotide

NMNAT

Nicotinamide mononucleotide adenylyl transferase

NPC

Neural progenitor cell

NSC

Neural stem cell

NR

Nicotinamide riboside

OGD/R

Oxygen glucose deprivation/reoxygenation

OXPHOS

Oxidative phosphorylation

PARP

Poly(ADP-ribose) polymerase

PARylation

Poly(ADP-ribosylation)

PDH

Pyruvate dehydrogenase

PGC-1α

Peroxisome proliferator-activated receptor- gamma coactivator-1alpha

PHD

Prolyl hydroxylase

PKC

Protein kinase C

PRC2

Polycomb repressive complex 2

PTEN

Phosphatase and tensin homolog

ROS

Reactive oxygen species

SARM1

Sterile alpha and TIR motif-containing 1

SDH

Succinate dehydrogenase

SGZ

Subgranular zone

SIRT

Sirtuin

SOD

Superoxide dismutase

SVZ

Subventricular zone

TET

Ten-eleven translocation

TF

Transcription factor

tPA

Tissue plasminogen activator

TSC2

Tuberous sclerosis complex 2

UCP2

Uncoupling protein 2

VDAC

Voltage dependent anion channel

VEGF

Vascular endothelial growth factor

Wld

Wallerian degeneration

Author Contributions

PN: Conceptualization, funding acquisition, resources, writing—original draft, review and editing; PPB: Resources, funding acquisition, supervision.

Funding

This work is supported by Department of Science and Technology (DST) (D.O. No. SR/CSRI/196/2016), India; Department of Biotechnology (BT/PR18168/MED/29/1064/2016), India. University Grants Commission (UGC) (UH/UGC/UPE-2/Interface studies/Research Projects/B1.4; UH/UPE-2/28/2015). Universities with Potential for Excellence—Phase II. Parimala Narne acknowledges the financial support from DST WOS-A Grant (SR/WOS-A/LS-126/2018).

Data Availability

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical Approval

This article is generated from literature survey and hence does not require ethical approval.

Consent to Participate

This article is generated from literature survey and hence does not require consent to participate or publish.

Consent for Publication

The article is a literature review and does not involve any research participants.