Inhibition of neuroinflammatory nitric oxide signalling supresses protein glycation and recovers neuronal dysfunction in prion disease

Background : Several neurodegenerative diseases associated with protein misfolding (Alzheimer’s, Parkinson’s disease) exhibit oxidative and nitrergic stress following initiation of neuroinflammatory pathways. Associated nitric oxide (NO)-mediated post-translational modifications impact upon protein functions that can exacerbate pathology. Non-enzymatic and irreversible glycation signalling has been implicated as an underlying pathway that promotes protein misfolding, but the direct interactions between both pathways are poorly understood. Methods : Here we investigated the potential therapeutic effects of supressing neurotoxic NO signalling during early progression of prion disease. Tg37 mice aged 3-4 weeks were inoculated by intracerebral injection with either 1% brain homogenate of Rocky Mountain Laboratory (RML) scrapie prion protein or control normal brain homogenate (NBH). Hippocampal gene and protein expression levels of oxidative and nitrergic stress markers were analysed and electrophysiological characterisations of pyramidal neurons were performed in 6-10 weeks old RML and NBH mice. Mice were injected with a NO synthase (NOS) inhibitor and the time course of disease markers was compared to controls. Electrophysiology, immunoblotting and immunocytochemistry studies were performed to identify the effects of NOS inhibition on neurophysiology, glycation, prion protein misfolding and cell death. Statistical analyses employed two-tailed unpaired Student’s t -test, one-way or two-way ANOVA as required and data were considered significant with P<0.05. Results: Increased neuroinflammatory signalling was observed in mice between 6 and 10 weeks post inoculation (w.p.i.) with scrapie prion protein which was characterised by enhanced nitrergic stress and associated with a decline in hippocampal neuronal function by 9 w.p.i.. Daily in vivo administration of the NOS inhibitor L-NAME between 6 and 9 w.p.i. at 20 mg/kg abolished the functional degeneration of hippocampal neurons in prion mice. We further found that this intervention in diseased mice ameliorated 3-nitrotyrosination

3 Conclusions: Our data suggest that alleviating nitrergic stress during early phases of neurodegeneration reduces neurotoxic post-translational NO signalling and glycation-assisted prion misfolding in the hippocampus, a mechanism which might be applicable to other protein misfolding neurodegenerative conditions. Background Diseases associated with old age, including Alzheimer's disease (AD) and other forms of dementia, are increasing in prevalence and, although symptomatic treatments exist, no cure to prevent the progression has yet been described. Neurodegenerative diseases associated with the accumulation of misfolding proteins including AD, Parkinson's disease (PD) and various forms of prion disease share many common features such as upregulation of neuroinflammatory signalling [1]. In particular, neuroinflammatory signalling can compromise synaptic plasticity [2] in early degeneration representing an initiation phase for dysfunction. It is well established that cellular stress signalling in neurodegenerative diseases is associated with aberrantly elevated levels of nitric oxide (NO) and activation of oxidative/nitrergic stress pathways [3][4][5]. NO has both physiological and pathological roles in the nervous system where it is synthesized by NO synthases (neuronal NOS [nNOS], inducible

NOS [iNOS] and endothelial NOS [eNOS]
). Excessive NO production reported in prion disease, AD and PD is leading to aberrant post-translational protein modifications and unbalanced neurotoxic nitrergic signalling [6][7][8][9]. Protein cysteine residues can be directly and reversibly S-nitrosylated by NO [9] which can lead to protein dysfunction to facilitate disease progression [10,11]. NO also reacts with free radicals such as superoxide anion O 2resulting in the generation of peroxynitrite which accumulatively induces protein nitrotyrosination, the irreversible chemical addition of a nitro group to the tyrosine residues generating 3-nitrotyrosine (3-NT), which can result in a loss of physiological protein function [12]. Therefore, preventing disease-relevant, in particular, iNOS-mediated NO generation is a strong and promising therapeutic target across many neurological diseases [13][14][15].
In neurodegenerative conditions, oxidative and nitrergic stress is believed to be an initiator of many pathological pathways and can have both direct and indirect roles in the mechanisms underlying 4 neurodegeneration [16][17][18]. For example, 3-nitrotyrosination of β-amyloid (Aβ) oligomers is thought to be one of these direct pathways by increasing oligomer stability and toxicity [12,19]. Interestingly, Aβ42 oligomerisation also plays a key role by promoting nitrergic stress i.e. inducing 3nitrotyrosination of neuronal proteins [20]. One of these targeted proteins is the enzyme triosephosphate isomerase (TPI) which regulates the glycolytic flow by interconverting dihydroxyacetone phosphate (DHAP) and glyceraldehyde 3-phosphate (GAP). The nitrergic post-translational modification of TPI induced by the Aβ-peptide reduces its enzymatic activity [21] leading to a shift of balance of the metabolites towards DHAP formation. Subsequently, DHAP transforms spontaneously into the cytotoxic metabolic by-product methylglyoxal [22,23]. A nitro-oxidative environment favours high levels of 3-nitrotyrosinated TPI in AD patients [21] and as a result, inhibition of this enzyme leads to non-enzymatic methylglyoxal-mediated generation of protein glycation [21] and formation of advanced glycation end products (AGE).
The production of AGE is irreversible and glycated protein products have a propensity to aggregate [24] while exhibiting resistance to protease degradation. Notably, protein glycation is a marker of aging and it promotes inflammation via the overproduction of intracellular reactive oxygen species (ROS) in addition to impairing proteasomal activities. The resulting AGE-mediated induction of oxidative stress thereby leads to augmented oxidative damage in the brain presenting a positive feedback loop of neurotoxic pathways in different protein misfolding diseases [25][26][27]. In AD, the aggregation and deposition of AGEs have been observed in tau protein tangles and Aβ plaques [28,29] with the polymerization of Aβ being significantly accelerated by AGE-mediated protein crosslinking [30]. Glycation at specific residues causes a profound impact on the ability to form amyloid fibrils [31,32] confirming a contributing role of this modification in aggregation. AGEs are principally recognised by the receptor for AGE (RAGE) that belongs to the immunoglobulin gene superfamily.
RAGE is found on the surface of numerous cells types including microglia, monocytes, astrocytes, neurons and endothelial cells [33]. RAGE activation results in the stimulation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase [34], an enzyme that produces superoxide radicals. It also induces the production of cytokines via NF-kB expression, followed by upregulation of inflammatory 5 pathways [33]. Importantly, AGE-formation and RAGE expression are increased in brains from Creutzfeldt Jakob Disease (CJD) patients and are associated with enhanced oxidative and nitrergic stress in patient brains [35,36] suggesting the possibility that AGE-mediated modifications might, in part, play a role in the accumulation of the aberrant prion protein.
In this study, we investigated neurodegeneration induced by prion misfolding to assess the role of nitrergic stress caused by excess production of NO during the progression of the disease. We established that the decline of hippocampal CA1 pyramidal neuronal function is associated with increased nitrergic stress and reduced levels of antioxidants, and that prolonged in vivo treatment with a NOS inhibitor reduced prion protein aggregation and diminished aspects of neurodegenerative signalling such as functional neuronal decline. We further found that blocking NO signalling reduced nitrotyrosination of TPI and RAGE upregulation.

Mice and Prion inoculation
Hemizygous Tg37 mice that overexpress the cellular mouse prion protein (PrP C ) have been used as described previously [37]. Mice were fed ad libitum with a standard mouse chow. Tg37 mice aged 3 to 4 weeks were inoculated by intracerebral injection into the right parietal lobe with 1% brain homogenate of Rocky Mountain Laboratory (RML) misfolded prion protein (PrP Sc ) [37]. Control mice received 1% normal brain homogenate (NBH) and hippocampi and cortices' of both groups were used for analyses.

Animal drug treatment
Tg37 mice inoculated with NBH or RML brain homogenate are denoted as NBH or RML mice. RML mice were treated (i.p.) with vehicle (5% glucose) or N(ω)-nitro-L-arginine methyl ester (L-NAME, 20 mg/kg) daily from 6 weeks post inoculation (w.p.i.) for 3 weeks. This dosage provides a ~70% inhibition of lipopolysaccharide-induced NO x generation (via iNOS induction) in mice with an ED 50 for L-NAME of 5 mg/kg [38]. The presence of the L-NAME metabolite and active inhibitor of NOS in the brain, L-NOARG [39], was confirmed by high-performance liquid chromatography (HPLC) at a concentration of 11.8±1.1 nM/g tissue (n=8 mice).
Liquid chromatography-mass spectrometry (LC/MS) analysis of the metabolite L-NAME was injected (20 mg/kg; i.p.) daily for 5 days to reach steady-state levels before the mice were sacrificed and the brains dissected and snap-frozen in liquid nitrogen. The tissue samples were weighed still in the frozen state and disrupted using a micropestle in an Eppendorf tube. A two-stage extraction was performed to ensure denaturation, membrane disruption and extraction of the metabolite L-NOARG. A mix of chloroform/methanol (0.5 ml, 2:1) was added to the tubes and vortexed for 20 min. After centrifugation, the supernatant was removed to another Eppendorf tube and the pellets were re-suspended in 0.5 ml 50% methanol/water. They were again vortexed and recentrifuged. The solvent mixes were pooled for each sample and then dried in a centrifugal evaporator. Samples were reconstituted in 100% methanol by vortexing and re-centrifugation. The HPLC system used was a modular LC Shimadzu LC system, equipped with pump, cooled autosampler and column oven at 45⁰C. A Phenomenex Gemini 18 3µ column 100x2mm was used at 45⁰C, with mobile phase A (0.01% formic acid (pH 4.6) in water) and mobile phase B acetonitrile. The compound was eluted on a gradient from 0.0% B to 60% B in 5 min with a 2 min wash at 100% B. The MS system used was a Sciex 4000 QTrap run in ES+ve MRM mode. Four MRM channels (220.143>87.000, 220.143>174.200, 220.143>59.100 and 220.143>70.200) were monitored to ensure specificity.
Quantification was performed against a dilution series of the standard and the extraction efficiency was included in the calculations.

Western blotting
Mouse hippocampi were dissected into ice-cold phosphate-buffered saline (PBS) containing protease inhibitors and phosphatase inhibitors (Complete, Roche Diagnostics) and samples were flash-frozen in liquid nitrogen until further analysis. All the following procedures were then performed at 4⁰C. Frozen tissue was homogenised in ice-cold RIPA buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 3 mM EDTA, and 1% Nonidet P-40) and 1 x protease inhibitor cocktail with a tissue homogenizer (IKA T10 basic ULTRA-TURRAX®

Immunocytochemistry
Mice were anaesthetised with 3% isofluorane (2l/min O 2 ) and transcardially perfused with 20 ml of ice-cold PBS, followed by 20 ml of ice-cold 4% PFA. Following fixation, brains were immediately removed, and further fixed overnight in 4% PFA at 4⁰C. Brains were processed in paraffin wax and sliced at 5 µM using a microtome. Following antigen retrieval, sections were washed in TBS + 0.1% triton X-100 and blocked for 2 hours at RT in TBS, 0.1% triton X-100, 10% goat serum and 5% BSA.

NADPH-diaphorase assay
Mice were anaesthetised and transcardially perfused with PBS then 4% PFA. The brain was removed and fixed for 24 h in PFA before being embedded in paraffin. Five μm sections were cut and deparaffinised before being incubated in a mixture of 2.5 mg nitroblue tetrazolium, 10 mg β-NADPH, and 0.2% Triton X-100 in 10 mL of 0.05 M Tris buffer for 30 min at 37 °C. The reaction was terminated with 0.05 M Tris buffer washes.

qRT-PCR
Total RNA was extracted from isolated hippocampal tissue using RNeasy mini kit (Qiagen) and treated 8 with deoxyribonuclease 1. For each tissue, 1 µg of total RNA was reverse transcribed using an iScript™ cDNA synthesis kit according to the manufacturer's instructions (BioRad). Real-time PCR reactions were performed using 25-100 ng of template cDNA per reaction and Taqman gene expression 'assay on demand' assays. The target primer/probes used were FAM-labelled Ncf1 (Mm01208835_m1), and Stip1 (Mm00489584_m1). Reactions were multiplexed and normalised to a VIC-labelled primer/probe for β-actin (Actb; Mm00607939_s1). Relative mRNA expression in each region was determined by normalising to the average amount of hippocampus or cortex from NBH mice using the ΔΔCT method.

Electrophysiology
Brain slices were prepared from animals killed by cervical dislocation. Whole-cell patch recordings were made from visually identified mouse CA1 neurons in acute brain slices (300 μm thick) of the hippocampus as described previously [40,41]. Patch pipettes were pulled from glass capillaries with osmolarity between to 280-290 mosmol/l. Synaptic stimulation at the Schaffer collateral was achieved using an isolated stimulator (Digitimer, Welwyn Garden City, UK; 1-10 V, 0.1-0.2 ms) via a bipolar platinum electrode.
Potassium and sodium currents were recorded in CA1 pyramidal neurons by whole-cell voltage clamp, action potentials (AP) in current clamp. Holding potentials were -60 mV and voltage command steps ranged from -110 mV to +50 mV (200 ms) for potassium current measurements and pre-pulse potentials (400 ms) ranged from -110 mV to +50 mV for sodium current recordings. APs were evoked by 50 pA-step current injections ranging from -100 pA to +400 pA.
Cumulative postsynaptic current analysis: The apparent size of the ready releasable pool of vesicles (RRP) was probed by the method of cumulative evoked excitatory postsynaptic current (eEPSC) amplitudes. Stimulus trains (30 Hz for 1 s) were calculated as the difference between peak and baseline before stimulus onset of a given eEPSC. Receptor desensitization was not blocked as it did not affect eEPSC amplitudes, because a comparison of the decay of the first and the last eEPSC within a train did not reveal any significant difference in decay kinetics. The number of release-ready vesicles pool size was obtained by back extrapolating a line fit to the linear phase of the cumulative eEPSC plot to time zero.

Ultrahigh performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS)
Mass spectrometry studies were performed by Metabolon. Samples were prepared using the automated MicroLab STAR® system (Hamilton Company). To remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions: two for analysis by two separate reverse phase (RP)/UPLC-MS/MS methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC-MS/MS with negative ion mode ESI, one for analysis by HILIC/UPLC-MS/MS with negative ion mode ESI and one sample was reserved for backup. Samples were placed briefly on a TurboVap® (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis. The platform used by Metabolon integrates the chemical analysis, including identification and relative quantification, data reduction, and quality assurance components of the process. Workflow including extraction of raw data, peak-identification, and quality control processed on the assay platform, has been described previously [8].

Statistical analysis
Results are expressed as the means ± S.E.M. Statistical comparisons were made by one-way analysis of variance (ANOVA) with Newman-Keuls posteriori analysis, two-way ANOVA with Bonferroni's multiple comparisons test, Kolmogorov-Smirnov test for non-parametric data distributions or unpaired Student's t-test, as appropriate. A P-value smaller than 0.05 was considered as significant. All statistical tests were performed using GraphPad Prism v7 software.

Study approval
All animal work conformed to the United Kingdom Home Office regulations. All procedures (both nonregulated and regulated) were conducted under a Home Office project licence awarded to Joern Steinert under the Animals (Scientific Procedures) Act 1986. Results downregulated at 6 w.p.i. (P=0.038, two-way ANOVA, n=4 mice each, Fig. 1a). RNA expression for the peroxisome proliferator-activated receptor-gamma coactivator, PGC-1α, which controls the expression of genes related to the generation of ROS and prevents oxidative stress by reducing the production of ROS [42], is also downregulated at this early stage of disease (P=0.0015, two-way ANOVA, n=4 mice each, Fig. 1c), as is the prion-interacting protein stress-inducible phosphoprotein 1, STIP1 [43] (P=0.0013, two-way ANOVA, n=4 mice each, Fig. 1d). At 10 w.p.i. we found a strong increase in mRNA levels for Ikba (P=0.012) as well as NCF1 (P<0.0001), one of the components of NADPH oxidase (p47-phox, Fig. 1a, b, n=4 mice each). Furthermore, PGC-1a (P<0.0001, two-way ANOVA) and STIP1 (P=0.0047, two-way ANOVA) mRNA levels remain reduced at this later disease stage (Fig. 1c,   d). Importantly, Nos2 mRNA (iNOS) is increased at 10 w.p.i. (P=0.0266, two-way ANOVA, Fig. 1e) with Nos1 (nNOS) and Nos3 (eNOS) mRNA levels being strongly reduced at 6 and 10 w.p.i. (Additional file 1: Figure S1). The enhanced iNOS activity is likely a result of activated astroglia signaling which we further illustrated using the NADPH diaphorase assay at 6-12 w.p.i. (Fig. 1f). We have previously shown that NBH mice lack any significant NADPH diaphorase signals at 10 w.p.i. [8]. This data illustrates a strong increase of NOS (NADPH activity) from 8 w.p.i. onwards in RML mice with signals reminiscent of activated astroglia in the CA1 region spreading to the whole hippocampus at 10 and 12 w.p.i.. Further characteristics for upregulated astroglia were detected by immunostaining for GFAP which showed a significantly stronger expression at 10 w.p.i. in RML hippocampi (P=0.045, Student's t-test, Fig. 1g). Altogether our data indicate that an early onset neuroinflammatory contribution to prion disease is detectible from 8 weeks onwards. Thus our approach to start targeting NO signaling will focus on time points following 6 w.p.i. onwards.

Rescue of functional neuronal decline in neurodegeneration by reducing nitrergic stress.
We asked whether interference with neuroinflammatory NO signalling prior to detectable prion misfolding might alleviate some of the neuronal symptoms reported in the disease. In order to address this question, subsequent studies only focussed on RML mice and the effects of treatment since we did not find any indications for aberrant nitrergic signalling in NBH controls (Fig. 1). We quantified the function of CA1 pyramidal neurons and their Schaffer collateral-mediated synaptic inputs as a read-out of neuronal health in RML prion-infected mice treated daily from 6 w.p.i. with the NOS inhibitor, L-NAME in comparison to age-matched vehicle RML controls. Both cohorts were assessed by electrophysiology at 7 and 9 w.p.i.. These time points were chosen to study the effects of NOS manipulation based on data presented in Figure 1 at which we expect pathological neuroinflammation to become dominant. Fig. 2a shows representative Schaffer collateral-evoked excitatory postsynaptic currents (eEPSC) in pyramidal CA1 neurons in RML mice at 7 and 9 w.p.i.
following stimulation by a 30 Hz train. The recordings illustrate a vast deterioration in neuronal function by 9 w.p.i.. Mean train eEPSC amplitudes from vehicle (RML) or L-NAME treated mice are summarised in Fig. 2b, c (RML, black; +L-NAME, red, n=9 mice each).
Following 3 weeks of daily treatment with L-NAME, mean initial current amplitudes were increased at 9 w.p.i. compared to RML vehicle control mice (P=0.029, two-way ANOVA, Fig. 2d). Cumulative postsynaptic current analysis (Fig. 2e, f), a measure to estimate available vesicle pool sizes based on the cumulative eEPSC amplitude plots, revealed a strong recovery of the vesicular pool size at 9 w.p.i.
following L-NAME treatment (P=0.043, two-way ANOVA, Fig. 2g) suggesting an improvement of synapse function. An important regulatory mechanism for synaptic release is the control of release probabilities. As a measure to determine the initial synaptic release probabilities we assessed pairedpulse ratios (PPR, eEPSC 2 /eEPSC 1 amplitudes) at 33 ms inter-spike intervals of Schaffer collateralstimulated eEPSCs in diseased and treated mice. PPR values of synaptic responses in hippocampi from diseased mice revealed a significant increase at 9 w.p.i. indicating a reduction in release probability over time (P=0.0145, two-way ANOVA, Fig. 2h). L-NAME treatment, however, suppressed this increase in PPRs, which suggests that NOS inhibition prevents different aspects of synapse dysfunction (Fig. 2h). A further assessment criterion of neuronal health is the ability of synapses to spontaneously release vesicles. These spontaneous or miniature EPSCs (mEPSC) provide important signals for physiological neuronal homeostatic signaling [44]. Consistent with a general neuronal dysfunction in prion disease, we found that miniature EPSC (mEPSC) amplitudes and spontaneous firing frequencies were reduced at 9 w.p.i. (Fig. 3a-c). Analyzing the relative cumulative distributions of mEPSC amplitudes and inter-spike-intervals (ISI) revealed a strong leftward shift in amplitudes and 13 rightward shift in ISI in prion disease at 9 w.p.i. compared to 6 w.p.i. (Fig. 3b, c, amplitudes: P<0.0001, D=0.571; ISI: P<0.0001, D=0.738, Kolmogorov-Smirnov test: 9 vs 6 w.p.i.) confirming again compromised neurotransmission. A protective effect of L-NAME treatment was observed at 9 w.p.i. (Fig. 3b, c, amplitudes: P<0.0001, D=0.571; ISI: P<0.0001, D=0.683, Kolmogorov-Smirnov test: 9 w.p.i. vs 9 w.p.i. + L-NAME treatment). The effects of prion disease are further illustrated for mean mEPSC amplitudes and frequencies (Fig. 3d, e, black, P<0.05, one-way ANOVA) and importantly, L-NAME treatment leads to recovery of both parameters over the time course of the disease (Fig. 3d, e, red, P<0.05, one-way ANOVA).
We also assessed neuronal function by measuring the ability of CA1 pyramidal neurons to generate current-evoked action potentials (AP), and characterized the underlying whole-cell potassium and sodium currents. Both, the ion channel functions and AP propagation play crucial roles in regulating neuronal performance and allow the neurons to sustain information transmission. AP parameters such as half width and amplitude are main characteristics, both predominately determined by potassium and sodium channel activities, respectively. L-NAME treatment led to a reduction of the increased AP half-width at 9 w.p.i. indicating an augmentation of potassium currents (P=0.002, one-way ANOVA).
(P=0.018, one-way ANOVA) following L-NAME treatment (Fig. 4a, b). As both measures are determined by the activities of voltage-gated potassium and sodium channels, we directly assessed both, potassium and sodium currents by voltage-clamp. Current-Voltage (IV) relationships show a strong decline of outward potassium currents (Fig. 4c, d) and inward sodium currents (Fig. 4e) at 9 w.p.i. confirming the observed changes in AP parameters, with L-NAME treatment recovering both currents substantially at 9 w.p.i., substantiating the AP data at 9 w.p.i. and confirming a neuroprotective role of L-NAME treatment. The data on neuronal physiology were supported by measurements of levels of neuronal protein involved in synaptic function and neuronal fidelity. Protein levels of several synaptic markers, including MUNC 18, SNAP-25 and complexin 1/2 were reduced with others showing strong tendencies of lower expression (synaptophysin, synaptobrevin, synapsin) at 9 w.p.i. (Additional file 2: Figure S2). Protein expression of the voltage-gated potassium channel Kv3.1, 14 the main conductance contributing to the IV relationships in Fig. 4c, d and determinant of AP half width, shows a decline at 9 w.p.i.. Altogether, the data illustrate the broad neuronal dysfunction at later disease stages and the potent effects of L-NAME treatment to recover several aspects of neuronal dysfunction by suppressing excessive NO signaling in vivo.

Protein 3-nitrotyrosination is driven by oxidative stress and enhances protein glycation in prion disease.
Protein 3-nitrotyrosination is characteristic of protein-misfolding neurodegeneration [22,45]. 3-NT occurs under conditions of oxidative stress and aberrant nitrergic activity. We confirmed a decreased ratio of reduced to oxidized glutathione GSH/GSSG in prion diseased mice at 9 w.p.i. (P=0.0039, control NBH vs RML mice, Student's t-test, Fig. 5a) providing evidence for enhanced oxidative stress as one hallmark of the disease and confirming previously reported increases in oxidative stress levels in prion-diseased mice [8]. Together with the enhanced levels of nitrergic stress associated with augmented expression levels of iNOS mRNA (Fig. 1), we found total amounts of 3-nitrotyrosinated proteins to be >5-fold elevated in prion infected mice compared to NBH controls at 9 w.p.i.
Both AD and CJD are associated with neurotoxic glycation signalling as well as a strong presence of AGEs [22,23,36]. AGEs are produced by the non-enzymatic glycation of amino compounds with reducing sugars through a series of sequential and irreversible reactions. The predominant neuronal precursor to induce protein glycation is DHAP which spontaneously forms methylglyoxal, a strong glycating agent [46]. We have measured DHAP by mass spectrometry and found levels to be ~2-fold increased (P=0.0027, Student's t-test, Fig. 5c) in the hippocampus of prion-infected mice. As TPI is the upstream enzyme in the cascade to produce DHAP, a reduction of its enzymatic activity would be responsible for the observed changes in the levels of its metabolites. It has been previously reported that mutations of TPI carrying the tyrosine substitutions Y165F and Y209F, both of which are located at the catalytic center, render this enzyme inefficient when expressed in human neuroblastoma cells thereby shifting the balance between GAP and DHAP towards the latter and reducing cell viability [23]. Based on these findings, we tested whether we could detect a nitrergic post-translational modification of TPI in prion-diseased mice that would ultimately be responsible for an accumulation of DHAP. Indeed, when performing immunoblotting for 3-nitrotyrosination after having immunoprecipitated TPI, we detected an increase of 3-NT TPI protein levels in disease (P=0.033, one-way ANOVA, RML vs NBH, Fig. 5d). Importantly, treatment with L-NAME prevented the increase in 3-NT TPI levels and reversed it back to control levels (P=0.484, one-way ANOVA, NBH vs RML+L-NAME, Fig.   5d). Total amounts of TPI protein were unchanged (P=0.163, one-way ANOVA, Fig. 5f). The production of AGE following TPI nitrotyrosination is directly linked to the activation of the receptor for AGE (RAGE) and the expression of RAGE reflects a positive confirmation for AGE signaling [47]. To test whether we could confirm AGE signaling in prion mice we examined RAGE expression and found that prion diseased mice exhibited markedly enhanced levels of RAGE (P=0.0074, one-way ANOVA, Fig. 5e). We tested the effects of NOS inhibition on RAGE expression and found that treatment with L-NAME ameliorated these increases (P=0.0243, one-way ANOVA, Fig. 5e, note the lower band represents actin) providing strong evidence for the NO-mediated 3-NT/TPI/DHAP/RAGE signaling cascade in prion disease which signifies a disease-associated neurotoxic pathway.

Nitric oxide and glycation signalling impact on prion misfolding.
The pathological scrapie form of the prion protein is heavily glycated resulting in mainly N ε -(carboxymethyl)lysines at up to eight lysine residues, whereas PrP C does not contain glycated amino acids [48,49]. It has been shown that mono-or di-glycation affects the molecular weight of the prion protein resulting in multiple bands following immunoblotting. We thus tested for expression of prion protein in hippocampal tissues. Importantly, we noticed the occurrence of bands above the expected PrP c molecular weight (~36 kDa) in hippocampal samples of RML mice at 9 w.p.i. detected using the ICSM35 antibody raised to specifically recognise β-sheet-rich structures within prion proteins [50,51] ( Fig. 6a). These higher weight bands might therefore reflect glycated and β-sheet-enriched prion protein.
We next assessed hippocampal tissue by immunocytochemistry and stained for prion protein to investigate the localisation of the misfolded protein in NBH and RML mice with and without L-NAME treatment. In order to specifically visualise β-sheet-rich PrP Sc that is prerequisite for prion aggregation we immunostained samples using the ICSM35 antibody [50,51]. We found that prion signals were almost exclusively detected in the cytosol of cubiculum neurons in RML mice reminiscent of possible protein aggregates formed within the cytosol [49,52] (Fig 6b, c). Cubiculum areas from NBH mice did not show any labelling in neurons (Fig. 6b) confirming PrP Sc specific detection in RML mice (note also the strong loss of DAPI signals in RML samples indicative of substantial neuronal loss which was reversed after L-NAME treatment, Fig. 6b). Following counting of ICSM35-positive (PrP Sc ) cells in the hippocampus, we found larger proportions of cells in RML mice compared to RML mice receiving L-NAME treatment suggesting a reduction in the occurrence of prion protein misfolding (RML: 38.8 ± 4.1%, RML+L-NAME: 23.4 ± 2.7%, total cell number: 679, P=0.0346, Student's t-test, [n=5 RML mice, n=3 RML+L-NAME treated mice], Fig. 6b, c). These data therefore indicate that reducing nitrergic stress could slow down the progression of protein misfolding and/or aggregation. We further noticed that the absolute numbers of pyramidal neurons in the hippocampus of L-NAME treated mice was similar from control NBH hippocampi whereas vehicle treated RML mice showed a significant reduction in cell numbers (NeuN positive cells per mm 2 : NBH: 0.0021 ± 0.0001 (n=3 mice); RML: 0.0007 ± 0.0001 (n=5 mice), P=0.0324 vs NBH; RML+L-NAME: 0.0013 ± 0.0003 (n=3 mice), one-way ANOVA; Fig. 6b, d (representative images of NeuN and DAPI labelled tissues). This data suggest that L-NAME treatment results in a reduction of neuronal loss consistent with the L-NAME mediated rescue of neuronal function at this time point of disease (Fig. 2-4).
Previously we reported that the metabolome of the hippocampus as well as the cortex is altered in prion disease, with both regions showing significant increases in nitrergic and oxidative stress signalling at 10 w.p.i. [8]. Furthermore, diseased mice exhibit behavioural deficiencies in burrowing activity at 9 w.p.i. [37], a behaviour associated with hippocampal but also cortical dysfunction [53]. In parallel to the hippocampal increases in DHAP levels, we also confirmed enhanced levels of DHAP in the cortex of RML mice (scaled intensity: NBH: 0.81 ± 0.07, n=8; RML: 1.07 ± 0.07, n=8; P=0.023, Student's t-test, data not shown) suggesting a similar upregulated glycation signaling which could underlie the observed prion misfolding. To test if we can detect potential signs of prion misfolding in the cortex as well and beneficial effects of L-NAME treatment in cortical regions at 9 w.p.i. we assessed cortical tissues for misfolded prion protein by immunocytochemistry. We found that in the cortex from RML mice prion protein positive signals exhibit clusters within the cytosol showing typical punctate form of localisation pattern as seen in the immunocytochemistry data in the hippocampus reflecting misfolding of the protein which was absent in tissue from NBH mice (Additional file 3: Figure   S3a). Following counting of cells exhibiting such PrP Sc clusters, we found that RML samples exhibited high numbers which were greatly reduced following L-NAME treatment suggesting a reduction of neurons possessing b-sheet-rich prion protein (RML: 95.0 ± 2.2%, RML+L-NAME: 24.2 ± 5.1%, total cell number: 1892, P<0.0001, Student's t-test, [n=5 RML mice, n=3 RML+L-NAME treated mice], Additional file 3: Figure S3b). Importantly, we also assessed the total number of cortical neurons and found that L-NAME treatment reversed the reduction in cell loss caused by disease (cells per mm 2 : NBH: 0.0022 ± 0.0002 (n=3 mice); RML: 0.0014 ± 0.00006 (n=5 mice), P=0.0119 vs NBH; RML+L-NAME: 0.0018 ± 0.00004 (n=3 mice), one-way ANOVA; Additional file 3: Figure S3a, c).
Together, our data provide evidence that in vivo NOS inhibition for 3 weeks can reduce the amount of prion protein misfolding which positively impacts on neuronal health. We propose that nitrergic stress exerts its effects via post-translational modification, and associated inhibition, of TPI leading to an accumulation of glycated proteins. The resulting generation of advanced glycation end products may impact on prion protein misfolding as AGE-assisted conversion of PrP C into PrP Sc itself as it is reported for α-synuclein [26,27] or Ab [31].

Discussion
This study was set out to investigate the role of nitrergic stress in the development of protein misfolding neuropathology using prion-mediated neurodegeneration as a well-established disease system. We show that prion disease is characterised by a range of dysfunctional pathways involving oxidative stress and neuroinflammatory nitrergic signalling as well as activation of the glycation-AGE 18 signalling cascade. Importantly, we also show that diminishing cytotoxic NO levels throughout early disease stages can reduce the progression of the neuronal pathology as revealed by sustained synaptic and neuronal function, the associated suppression of glycation signalling and the reduced generation of misfolded prion protein in the hippocampus and cortex of infected animals. Although NO signalling plays an important role in physiological neuronal function including regulation of ion channel activities [40, 54,55] or several aspects of neurotransmission [56,57], NO dysregulation and nitrergic stress have long been associated with numerous neurodegenerative diseases [3,9,19,58,59]. Its main disease-relevant neurotoxic signaling involves post-translational protein modifications, such as S-nitrosylation and 3-nitrotyrosination. For instance, it has been shown in various systems that S-nitrosylation of synaptic proteins such as AMPA receptors [60], syntaxin [61], complexin [62], stargazin [63,64] or dynamin [65] can modify their functions (for review [4]). Several studies have confirmed that supressing excess NO production might be beneficial under various conditions of neuroinflammation which is a major hallmark of protein misfolding diseases. As such targeting NO signalling and NO-mediated post-translational modifications presents an attractive target for treatment in neuropathology [9]. For instance, a potent iNOS inhibitor, GW274150, has been used in Phase II clinical trials to treat migraine and other NOS inhibitors are known to provide neuroprotection in PD [13] or under conditions of cerebral ischemia and cerebral artery occlusions [66,67]. In various models of neurodegeneration pharmacological inhibition of iNOS produced beneficial effects [14,15] confirming that blocking nitrergic neurotoxicity can protect the central nervous system [68]. In addition, two NOS blockers, L-NNA and L-NAME, have been used in human trials to treat septic shock, although with mixed outcome, this requires further investigations (for review see [69]).
However, the exact downstream targets of NO and signaling routes in disease remain elusive despite the identification of numerous modified proteins. However, there is growing evidence that high nitrergic activity in conjunction with enhanced oxidative stress can directly result in activation of a cascade involving glycation signalling [20,[70][71][72], presenting a positive feedback loop to promote cell death. Furthermore, nitrergic and oxidative stress facilitates the formation of peroxynitrite, which in turn, leads to 3-nitrotyrosination of cellular proteins. For example, 3-NT formation of Aβ at tyrosine 10 promotes its aggregation [12,19] thereby increasing its synaptotoxicity. Glycation has been detected in AD and CJD in earlier studies [35,73,74], however, disease-related 3-NT signalling has only recently been proposed to involve glycation signalling in AD [21]. This specific pathway requires the dysregulation of glycolysis via the 3-NT-mediated suppression of triose phosphate isomerase activity which results in increased cellular levels of DHAP. DHAP spontaneously decomposes to methylglyoxal, which is a highly reactive a-oxo-aldehyde that can modify both, proteins and DNA to form AGEs.
Methylglyoxal can be metabolised by the glyoxalase pathway, which requires the presence of GSH (summary pathway, Fig. 7). As our data show reductions in neuronal GSH levels and associated enhanced oxidative stress levels in prion disease [8], it suggests that this detoxifying pathway might also be compromised. The downstream consequences of increased AGE levels are not fully understood but emerging evidence suggests that this irreversible modification renders many proteins inactive and susceptible to misfolding such as in the case of α-synuclein [26,27] or Ab [31] or further increases oxidative stress due to super oxide dismutase inhibition [75]. In our study, we could show that pharmacological suppression of NO signalling, predominantly targeting neuroinflammatory iNOS generated NO, leads to diminished 3-NT of TPI and subsequent reduction of AGE formation, as detected by reduced RAGE expression. These effects are associated with diminished levels of misfolded prion protein (PrP Sc ) and improved neuronal health as measured electrophysiologically and confirmed by higher numbers of pyramidal neurons in the hippocampus and cortex of treated animals. Since one mechanism that leads to the stabilisation and deposition of PrP Sc fibrils may be the result of specific AGE modifications of PrP Sc , it is possible that inhibition of NO-mediated AGE formation, as demonstrated here, might be an approach of slowing PrP Sc misfolding, thereby alleviating the progression of prion disease.

Conclusions
The established correlation between protein glycation -as reported for several misfolding proteinsand prion misfolding might link the NO signalling cascade to the neuroprotective effect seen following treatment with L-NAME. Together, we present new evidence for the link between NO-mediated post-20 translational modifications of an enzyme directly connecting NO and glycation signalling. We postulate that the pharmacologically abolished AGE formation, via NOS inhibition, positively impacts on prion protein misfolding thereby slowing or reducing prion aggregation. This could present an approach to diminish the detrimental neuroinflammatory effect seen in neurodegeneration [76] since our evidence of prolonged neuronal health as a result of treatment implicates this signalling route as an important contributor to neuronal degeneration (Fig. 7) and highlights nitrergic stress and glycation signalling as a putative target for disease-modifying treatments. To take this study forward

Availability of data and materials
All data generated or analysed during this study are included in this published article [and its supplementary information files].

22
The authors declare no conflict of interest.

Author contributions
JMB and JRS designed research; JMB, SWR, JGS, HS and CO performed studies, SJB and ABT contributed with analytical tools and had intellectual input. JMP, JRS and JGS wrote the paper with inputs from SJB and ABT. All authors read and approved the final manuscript.     the NOS antagonists L-NAME (+L-NAME, red) for 3 weeks (9 w.p.i.). a, representative eEPSC trains recorded at 30 Hz for 10 s in an RML mouse at 7 w.p.i. and RML mouse at 9 w.p.i.. b and c, mean train eEPSC amplitudes at 7 and 9 w.p.i. for RML mice and RML mice following treatment. d, single eEPSC amplitudes at both time points for RML and treated RML mice. e and f, cumulative eEPSC amplitude plots at 7 and 9 w.p.i. for RML mice and RML mice following treatment. g, estimation of vesicle pool sizes at both time points for RML and treated RML mice. h, paired pulse ratios (PPR) at both time points for RML and treated RML mice. Data are presented as mean ± SEM, n=9 RML and n=6 RML+L-NAME treated mice with n=4-6 neurons per mouse, two-way ANOVA, *P<0.05. 35 36 Figure 3 Spontaneous synaptic activity recovers following NOS antagonism at early time points.
Spontaneous release measured as miniature excitatory postsynaptic currents (mEPSC) recovered following NOS inhibition. a, raw traces showing spontaneous release events for RML mice at indicated time points. b and c, relative cumulative probability diagrams shown for mEPSC amplitudes and inter-spike-intervals. L-NAME treatment results in a significant shift of the data distributions for both parameters (P<0.0001, Kolmogorov-Smirnov test). d, mEPSC amplitudes and frequencies (e) show disease-dependent decline (black). L-NAME treatment reversed the reduction in mEPSC amplitudes and frequencies (red). Data are presented as mean ± SEM, n=9 RML and n=6 RML+L-NAME treated mice with n=4-6 neurons per mouse, one-way ANOVA, *P<0.05.
37 Figure 4 Postsynaptic neuronal health is protected by NOS antagonism. Pyramidal hippocampal 38 neurons were characterised by whole-cell patch clamp. a, representative current-evoked action potentials. b, action potential half width and amplitudes were analysed at 7 and 9 w.p.i. and are shown for vehicle treated (black) and L-NAME treated RML (red) mice. c, voltage-gated potassium currents declined at 9 w.p.i. in RML mice (raw traces in c and current/voltage relationship [I/V, black] in d). L-NAME treatment prevents the decline in voltage-gated potassium currents at 9 w.p. i. (d, red). E, voltage-gated sodium currents decline at 9 w.p.i. in RML mice and L-NAME treatment (red) leads to enhanced currents at 9 w.p.i. (representative recordings at 9 w.p.i.). Data are presented as mean ± SEM, n=9 RML and n=6 RML+L-NAME treated mice with n=4-6 neurons per mouse, one-way ANOVA, *P<0.05, **P<0.01.
39 Figure 5 NOS inhibition reduces protein nitrotyrosination and glycation signalling. a, the ratio of reduced/oxidised glutathione is diminished in hippocampal tissue of RML mice at 10 w.p.i.. b, levels of 3-NT proteins is enhanced in RML mice and reduced to control levels (NBH) Figure 7 Schematic representation of the pathways impacted on by excessive protein nitrotyrosination. Glycolysis generates glutaraldehyde-3-phosphate (G3P) that the enzyme triose phosphate isomerase (TPI) uses to form dihydroxyacetone (DHAP) which spontaneously and non-enzymatically decomposes to methylglyoxal. The glyoxalase cycle is the detoxification system by which methylglyoxal is transformed into lactate using glutathione (GSH) as a cofactor. Following activation of iNOS associated with enhanced levels of NO and peroxynitrite, 3-nitrotyrosination of TPI decreases its activity, which leads to an accumulation of DHAP and methylglyoxal. However, high concentrations of methylglyoxal will overload this clearance mechanism, leading to an accumulation of GSSG and the formation of advanced glycation end products (AGE) with subsequent activation of the receptors for AGE (RAGE). Under pathological conditions, glycated proteins are compromised and can aggregate which further contributes to inflammation and cell dysfunction.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.