Hypoxia-inducible factor (HIF) prolyl hydroxylase inhibitors induce autophagy and have a protective effect in an in-vitro ischaemia model

This study compared effects of five hypoxia-inducible factor (HIF) prolyl hydroxylases (PHD) inhibitors on PC12 cells and primary rat neurons following oxygen-glucose deprivation (OGD). At 100 µM, the PHD inhibitors did not cause cytotoxicity and apoptosis. MTT activity was only significantly reduced by FG4592 or Bayer 85–3934 in PC12 cells. The PHD inhibitors at 100 µM significantly increased the LC3-II/LC3-I expression ratio and downregulated p62 in PC12 cells, so did FG4592 (30 µM) and DMOG (100 µM) in neurons. HIF-1α was stabilised in PC12 cells by all the PHD inhibitors at 100 µM except for DMOG, which stabilised HIF-1α at 1 and 2 mM. In primary neurons, HIF-1α was stabilised by FG4592 (30 µM) and DMOG (100 µM). Pretreatment with the PHD inhibitors 24 hours followed by 24 hour reoxygenation prior to 6 hours OGD (0.3% O2) significantly reduced LDH release and increased MTT activity compared to vehicle (1% DMSO) pretreatment. In conclusion, the PHD inhibitors stabilise HIF-1α in normoxia, induce autophagy, and protect cells from a subsequent OGD insult. The new class of PHD inhibitors (FG4592, FG2216, GSK1278863, Bay85-3934) have the higher potency than DMOG. The interplay between autophagy, HIF stabilisation and neuroprotection in ischaemic stroke merits further investigation.

The hypoxia inducible transcription factors (HIF) are regulators of cellular responses to hypoxia in mammals. HIF mediated responses involve increases in the expression of multiple genes in a context dependent manner, including those associated with erythropoiesis and angiogenesis 1 . Under normoxic conditions, efficient catalysis by the ferrous iron and 2-oxoglutarate (2OG) dependent HIF prolyl hydroxylases (PHDs 1-3 in humans) promotes binding of HIF-α to the VHL (Von Hippel Lindau) tumour suppressor protein which is a targeting component of an E3 ligase complex, resulting in proteasomal degradation of HIF-α 2 . The action of a second 2-oxoglutarate (2OG) dependent HIF hydroxylase, the asparginyl hydroxylase, factor inhibiting HIF (FIH), regulates HIF activity by reducing its binding with histone acetyl transferases 2 . During hypoxia, PHD activity is reduced resulting in the stabilization and accumulation of HIF-1α in the cytoplasm, which then translocates to the nucleus forming the HIF transcription factors that upregulate expression of multiple genes 3 . The array of genes targeted by the HIF system makes it an appealing pharmacological target, in particular via PHD inhibition mediated HIF upregulation for treatment of diseases including anaemia, ischaemic stroke, and wound healing 4 . Currently, four PHD inhibitors i.e. FG4592 (Roxadustat) from FibroGen, GSK1278863 (Daprodustat) from GlaxoSmithKline, Bay85-3934 (Molidustat) from Bayer and AKB-6548 (Vadadustat) from Akebia are in clinical use or trials for anaemia treatment in patients with chronic kidney disease (CKD) 5,6 .
A number of HIF PHD inhibitors (e.g. GSK360A, FG4497, FG2216, DMOG, DFO) have been studied in stroke models either in vivo or in vitro, where these compounds showed neuroprotective effects following an ischaemic insult [7][8][9][10][11][12] . Nevertheless, the role of HIF in stroke pathophysiology remains debatable 13 . Neuronal specific HIF-1α knock out mice resulted in a worse neurological outcome and larger infarct volume following 30 min middle cerebral artery occlusion (MCAO) 14 ; however, the neuronal specific HIF-1α knock out mice had better neurological outcomes after 75 min global ischaemia 15 . Single HIF1α or HIF-2α knock out mice had a similar infarct volume following MCAO compared to the respective wild type mice, possibly due to a mutual www.nature.com/scientificreports www.nature.com/scientificreports/ compensation 16 . HIF-1α/2α double knock out mice were observed to be significantly more impaired 72 hour after the MCAO suggesting HIF can be involved in functional recovery after cerebral ischaemia 16 . The effects of HIF on the ischaemic brain vary depending on the severity and/or duration of the stroke 13 . Whilst indirect induction of HIF, via genetic ablation of PHD1 or PHD2, reduced infarct volume and improved sensorimotor function following transient ischaemia 17,18 . The pharmacological inhibition of PHD2, and/or PHD1 could stimulate adaptations that provide protection against damage from ischaemic stroke 4,13 . A pharmacological approach for ischaemic cerebral tolerance is clinically appealing due to its non-invasive application. Currently, no available drug specifically inhibits a single PHD isoform, as such, researchers have mostly focused on generating inhibitors targeting PHD2, as it is proposed to be the most important PHD isoform in regulating HIF levels in normoxia 5 .
This study aimed to characterize effects of three clinical HIF-PHD inhibitors (FG4592, GSK1278863, Bay85-3934) alongside FG2216 (IOX3) and DMOG (dimethyloxalylglycine), a non-specific 2OG analogue, using an in-vitro oxygen-glucose deprivation (OGD) model. These chemicals activate the HIF signalling pathway in normoxia, with somewhat different effects 19 , which we considered could be reflected in specific effects on ischaemic neurons.

PC 12 cells. Effects of PHD inhibitors on PC12 cell viabilities and apoptosis in normoxia. PHD inhibitors are
widely used to mimic hypoxia in vitro, where they are initially dissolved and administered with DMSO. However, DMSO is a redox active small molecule and therefore could affect cell behaviour. In order to determine effects of DMSO on PC12 cells, a range of DMSO concentrations (0.5-10%) were administered to PC12 cells for 24 hours in normoxia. Both 5% and 10% DMSO produced significant reduction in MTT activity and increase in LDH release. Moreover, 2% DMSO produced a slight but significant reduction in MTT activity. The remaining concentrations of DMSO had no significant effects on MTT activity or LDH release compared to PC12 cultures into which no DMSO was added (Fig. S1). A concentration of 1% DMSO was therefore used in subsequent work to dissolve the PHD inhibitors.

Effects of PHD inhibitors on autophagy.
We thereafter studied the effect of 24 hours treatment with the PHD inhibitors (100 µM) on LC3b (microtubule associated protein light-chain 3b), p62 (sequestosome-1, SQSTM1) and Beclin1 (BCN1) expression. Rapamycin, an established, autophagy inducer, was used as a positive control. There was a significant increase in LC3b-II/LC3b-I ratio and reduction in p62 in cells exposed to 1 and 10 µM of Rapamycin (Fig. 3). A significant increase in the LC3b-II/LC3b-I ratio was also seen in cells exposed to DMOG, FG2216 and FG4592, but at slightly lower levels compared to GSK1278863 and Bay85-3934. In comparison to 1% DMSO (vehicle) treated cells, there was significant downregulated in p62 in cells exposed to DMOG, FG2216, FG4592, GSK1278863 and Bay85-3934. A significant increase in Beclin1 was seen in cells exposed to FG4592, GSK1278863 and Bay85-3934 (100 μM) but not in cells exposed to Rapamycin (1, 10 μM) and DMOG (100 μM).
Effects of PHD inhibitors on HIF-1α protein expression. Next, we investigated HIF-1α stabilisation at 24 hours PHD inhibitor treatment in normoxia. No significant stabilisation of HIF-1α was seen at concentrations of 1, 10 and 50 µM (data not shown). Significantly increased HIF-1α levels were observe in cells exposed to 100 µM of all the PHD inhibitors (FG2216, FG4592, GSK1278863 and Bayer 85-3934) (Fig. 4B), except for DMOG. Further studies demonstrated that HIF-1α was significantly stabilised in cells exposed to 1 mM and 2 mM of DMOG for 24 hours (Fig. 4C).
Effects of preconditioning with PHD inhibitors and re-oxygenation prior to OGD in PC12 cells. We then studied whether pre-treatment with the PHD inhibitors followed by re-oxygenation was protective for PC12 cells against www.nature.com/scientificreports www.nature.com/scientificreports/ subsequent OGD insult. The effects of various concentrations (1, 10, 50 and 100 µM) were studied. PC12 cells were subjected to 24 hours PHD inhibitors treatment followed by a period of 24 hours reoxygenation prior to an OGD insult (6 hours, 0.3% O 2 ). Six hours OGD (0.3% O 2 ) led to about ~40% of cell death, of which ~90% were early apoptotic, ~10% were late apoptotic/necrotic, and caused significant increase of LDH release (24.3% ± 2.3% vs 6.4% ± 2.2%) and reduction in MTT activity (73.6% ± 12.3% vs 100%), compared to the normoxia control.

Discussion
Stroke is the fourth main cause of mortality in the UK and the leading cause of adult disability, with cerebral ischaemia representing the most prevalent type of stroke (accounting for ~87% of cases) 20 . Over the past 30 years, various neuroprotective strategies for ischaemic stroke have failed to succeed in clinical trials due to the complex nature of stroke. Most of these strategies were aimed at targeting single molecules in the ischaemic cascade 21 . Recent focus has shifted towards targeting the brain's own endogenous protective mechanism 22,23 . Our study investigated one such promising approach known as ischaemic tolerance 7,10,11 .
Our study focused on analysing the most recent clinically advanced small molecule PHD inhibitors at biologically safe concentrations. These PHD inhibitors are being pursued for treatment of anaemia 24 . We discovered that the PHD inhibitors stabilised HIF-1α in normoxia, and upregulated a number of known HIF target genes expression in PC12 cells. The PHD inhibitors induced autophagy, but not apoptosis in normoxia. Importantly, these PHD inhibitors protected the PC12 cells from OGD insults when given as preconditioning agents. Our studies were conducted using a PC12 cell model of ischaemia tolerance 25 , as PC12 cells have been extensively used in neurobiological studies as models of Alzheimer's disease, Parkinson's disease, Huntington's disease and www.nature.com/scientificreports www.nature.com/scientificreports/ ischaemic stroke [25][26][27] . Since the 4 "clinical" PHD inhibitors had the similar effects on the PC12 cells, we then used FG4592 as a representative novel PHD inhibitor alongside with DMOG, a non-specific 2-OG analogue, in primary rat neurons, to validate the results obtained from the PC 12 cells. Both FG4592 and DMOG stabilized HIF-1α in the neurons at lower concentrations compared to PC12 cells; suggesting neurons are more sensitive to PHD inhibitors than PC12 cells. Similarly, both FG4592 and DMOG protected the neurons from OGD insults when given preconditioning.
At baseline, none of these compounds at 1-100 μM led to cytotoxicity (as assayed by LDH assays), cell death (as assayed by trypan blue exclusion assay) or apoptosis (as assayed by Annexin V/7-AAD staining) in PC 12 cells. MTT activities were unchanged by the compounds except with FG4592 (100 µM) and Bay 85-8934 (50 & 100 µM), which significantly decreased MTT activities. In primary neuron culture, no significant LDH release and MTT activity alteration was seen by FG4592 (100 µM) and DMOG (100 µM). On the other hand, DMOG (500 µM and above) resulted in significant LDH release and MTT activity reduction following 24 hours incubation in both PC 12 cells and primary neurons. MTT assay assesses mitochondrial catabolism and cell metabolic activity, which can be reduced in cells whilst retaining cell membrane integrity represented by LDH assay 28 .

Figure 5.
Effects of HIF PHD inhibitors on hypoxia gene expression. Expression of hypoxia genes, HIF1α, Bnip3, Phd2, Vegf, Pfkfb1, Pfkfb3, Ldha, after treatment with 100 µM PHD inhibitors in normoxia for 24 hours were normalised to β-actin levels. Significant upregulation of Bnip3 and Vegf was seen with FG4592 and Bay85-3934. Phd2 was significant upregulated by FG4592. Pfkfb3 was significantly upregulated by DMOG, FG2216, FG492, GSK1278863, and Bay85-3934. Ldha was significantly upregulated by FG2216, GSK1278863 and Bay85-3934. HIF1α was not changed by any of the HIF PHD inhibitors. Each data point represented the mean and standard deviation of the relative fold change with respect to 1% DMSO-treated sample normalised to reference gene β-actin level. The dotted line represented basal gene expression. Statistical significance was indicated as *P < 0.05 (Two-way ANOVA, Tukey's post-hoc analysis).
www.nature.com/scientificreports www.nature.com/scientificreports/ Next, the role of the PHD inhibitors on three major autophagic marker, LC3, p62 and Beclin1 was studied using rapamycin as a positive control inducer of autophagy 29 . Beclin1 plays an important role in induction of autophagy. Upon induction of autophagy, cytosolic LC3-I is conjugated to phosphatidylethanolamine to form LC3-II, which is incorporated into the autophagosomal membrane. The p62 protein (sequestosome 1) binds to polyubiquitinated proteins through an ubiquitin-associated domain and combines with LC3-II through its LC3-interaction region domain for attachment to the autophagosome, which is finally degraded in autolysosome. In the process of autophagy, p62 is continuously degraded. Beclin1, LC3-II and p62 have been used to monitor autophagic activity in various studies 30 . Our results showed that both the PHD inhibitors (100 µM) and rapamycin (1, 10 µM) significantly increased LC3b-II/LC3b-I expression ratio. GSK1278863 (100 µM) and Bayer 85-3934 (100 µM) promoted LC3-II expression to a similar extent as rapamycin (10 µM). FG4592, FG2216 and DMOG induced LC3-II expression to a lesser extent, but still significant extent. A subsequent downregulation of Figure 6. The effect of preconditioning with HIF-PHD inhibitors followed by 24 hours re-oxygenation and 6 hours OGD on PC12 cells. PC12 cells were exposed to the PHD inhibitors (1, 10, 50 and 100 µM) for 24 hours in normoxia (21% O 2 ) followed by 24 hours re-oxygenation (21% O 2 ) and subsequent 6 hours OGD insult (0.3% www.nature.com/scientificreports www.nature.com/scientificreports/ p62 was seen in cells treated with rapamycin and the PHD inhibitors in comparison to the vehicle (1% DMSO) -treated cells indicating degradation by autolysosomes was initiated. Our results are consistent with a study by Duran et al. 31 reporting inhibition of PHDs by DMOG for 5 hours in U2OS cells was sufficient to induce autophagy. Similarly, DMOG induced IC3-II accumulation and P62 degradation in HeLa cells 32 . In a rat model of CKD, DMOG treatment for 1 week significantly increased autophagy 33 . Li et al. 34 , reported that FG4592 induced autophagy in SH-SY5Y cells via LC3II/LC3I upregulation and p62 downregulation. The "clinical" PHD inhibitors significantly upregulated Beclin1 in the PC12 cells, which is line with the study by Lu et al. 35 showed that HIF-1/ BNIP3/Beclin1 signalling pathway modulates autophagy and contributes to hypoxia preconditioning-induced protection against OGD/Reperfusion. Although rapamycin promoted autophagy (shown by LC3-II upregulation and p62 downregulation), it did not upregulate the expression of Beclin1. This is in line with studies by Li et al. 36 www.nature.com/scientificreports www.nature.com/scientificreports/ and Grishchuk et al. 37 , which found that rapamycin promoted autophagy via a Beclin1 independent mechanism. They showed that rapamycin inhibited mTOR resulting in activation of Atg1 kinase complex to regulate autophagy.
Autophagy is a self-catabolic process of damaged or dysfunctional cellular components, which are recycled for providing energy and nutrients 29 . A number of studies demonstrate that mild/moderate autophagy has a neuroprotective effect on ischaemic stroke whereas insufficient or excessive autophagy results in nerve damage and cell death 22,29 . A study by Park et al. 38 showed ischaemic preconditioning (IPC) increased generation of autophagosomes and lysosomal activity. Blocking autophagic activity by 3-methyladenine during IPC ameliorated the neuroprotective effects of IPC, indicating that autophagy participates in IPC-induced neuroprotection. Similarly, Sheng et al. 39 showed that autophagy activation during IPC plays an important role in developing tolerance against a subsequent fatal ischaemic insult. Yan et al. 40 showed that elevated autophagic activity contributed to increased tolerance against transient focal cerebral ischaemia in hyperbaric oxygen preconditioning while inhibition of autophagy using 3-methyladenine supressed the protective effective. Additionally, pre-and postconditioning with rapamycin were found neuroprotective in various in-vivo and in-vitro studies 22,41 .
At 100 μM, all the inhibitors except DMOG were able to accumulate HIF-1α significantly in the PC12 cells after 24 hours incubation at normoxia. We subsequently demonstrated, despite resulting in significant cytotoxicity DMOG at both 1 and 2 mM stabilized the HIF-1α in PC12 cells. This indicates that the 4 'clinical' compounds have higher potency compared to DMOG for inducing HIF-1α in PC12 cells. This conclusion is consistent with studies by Zhdanov et al. 42 in PC12 cells and by Chan et al. 19 , in MCF-7 cells, where 1 mM DMOG significantly stabilised HIF-1α. In the primary neurons, the HIF-1α was stabilised by FG4592 (30 μM) and DMOG (100 μM), whose concentrations were significantly lower compared to those used in PC 12 cells. FG4592 clinical trial concentration is 30 µM 43 . 50 µM of FG4592 was shown to upregulate HIF-1α expression in SH-SY5Y cells in normoxia 31 . In line with our results in primary rat neurons, Ogle et al. 9 , treated primary rat neurons with DMOG (up to 500 µM) for 24 hours at normoxia and showed increased levels of HIF-1α protein. Badawi and Shi 44 treated primary rat neurons with 0.5, 1 and 2 mM DMOG and HIF-1α protein levels were significantly increased by 57%, www.nature.com/scientificreports www.nature.com/scientificreports/ 83% and 93% respectively. DMOG at 2 mM is close to it maximal effect as a higher concentration of DMOG did not cause further increase in the HIF-1α protein level.
Six hours OGD (0.3% O 2 ), led to about 40% of cell death. Pre-treatment with the PHD inhibitors at 100 µM for 24 hours with 24 hours reoxygenation provided significant cytoprotection of both PC12 cells and primary neurons following 6 hours OGD. This is consistent with the study by Ogle et al. 9 that preconditioning with DMOG (100 μM) for 24 hours prior to OGD significantly reduced OGD-induced neuron cytotoxicity. Jones et al. 45 reported that pre-treatment with DMOG > 30 µM for 20 hours prior to OGD significantly reduced OGD-induced neuronal death in a dose-dependent manner. DMOG induced protection has also been demonstrated in-vivo, in which the systemic application of DMOG before onset of cerebral ischaemia and throughout led to increased acute cerebral tissue preservation 7 . Whilst, post-treatment with DMOG or FG4497 attenuated sensorimotor dysfunction in mice 3 days or 7 days after ischaemia-reperfusion injury 9,11,18 .
There are few studies on the 4 'clinical' PHD inhibitors in stroke research. GSK1278863 (Daprodustat), a pyrimidinetrione-glycinamide, is at least partially, selective for the PHDs over other 2-OG dependent enzymes, which minimises off-target effects whilst also signifying an advantage over DMOG 5,46 . Another PHD inhibitor from GlaxoSmithKline, GSK360A has been shown to decrease the infarct volume and improve behaviour after transient MCAO in rats 12 . FG2216 (IOX3) is structurally related to FG4592, and was replaced in clinical development by FG4592 following the appearance of adverse effects in clinical trial phase II 24 . There is an additional phenoxy-group on the phenyl isoquinolyl ring in FG4592, resulting in more efficient binding than FG2216 5 . FG2216 offered neuroprotective when pre-treating mice followed MCAO 10 .
Our study reveals that 6 hours OGD mainly results in apoptosis, rather than necrosis in the PC 12 cells, consistent with the previous studies 47 . Pre-treatment with PHD inhibitors protected PC12 cells from apoptosis following OGD. These findings correlate with a finding of a study by Li et al. 34 , where FG4592 reduced MPP + (1-methyl-4-phenylpyridinium) induced apoptosis and improved mitochondrial function in SH-SY5Y cells. Apoptosis is energy-dependent, reversible programmed cell death that results in rapid clearance of cells by phagocytosis. HIF-1α stabilisation induces apoptosis by stabilising of tumour suppressor gene p53, which induces Bax and Bak proteins regulating the release of cytochrome C 48 . HIF-1α stabilisation also results in upregulation of proapoptotic family members such as BNIP3, Noxa, Nix and downregulation of Bcl-2 49 . By contrast, HIF-1α can also protect cells from apoptosis by elevating Bcl-2 and Mcl-1 levels, Bcl-xL induction, and decreasing pro-apoptotic Bid, Bax, and Bak levels 50 . In our study, the PHD inhibitors (100 µM) treatment for 24 hours stabilised HIF-1α, and protected the cells from a subsequent OGD injury.
Hif1a gene expression was not significantly altered by the PHD inhibitors in the PC 12 cells despite the HIF-1α protein levels being significantly increased by the PHD inhibitors. Our observations are consistent with various studies that indicate that HIF-1α levels are mainly regulated at the post-transcriptional level 51,52 . During ischaemia, cells switch from oxidative phosphorylation to anaerobic glycolysis. Significant upregulation of Pfkfb3 was seen all the five PHD inhibitors, which is consistent with the study by Minchenko et al. 53 that revealed cobalt chloride, DFO and DMOG stimulated Pfkfb3 mRNA expression in Hep3b cells. PFKFB3 is ubiquitously expressed in several proliferating cells and tissues 54 . No significant changes in Pfkfb1 in the PC12 cells by the PHD inhibitors were seen in this study. FG2216, GSK1278863 and Bay 85-3934 significantly upregulated the Ldha expression in PC12 cells, while FG4592 and DMOG did not. HIF signalling possesses feedback loops, e.g. PHD2 and PHD3 are upregulated in a negative feedback loop with HIF-1 stabilisation 55 . The Phd2 expression was significantly upregulated by FG4592 in PC12 cells. FG4592 and Bayer 85-3934 significantly upregulated Bnip3 and Vegf expression. www.nature.com/scientificreports www.nature.com/scientificreports/ It has been recently discovered that the upregulation of BNIP3 and BNIP3L provides a protective effect through promoting mitochondrial autophagy via competing with Bcl-2 releasing Beclin-1 30 . VEGF plays an important role in stimulating angiogenesis and neurogenesis resulting in increased blood flow and metabolism. VEGF also promotes cell survival by promoting survival-promoting signalling pathways, such as serine-threonine protein kinase Akt signal transduction system and induction of anti-apoptotic pathways 56 . VEGF is reported to reduce caspase-3 activation and cell death in an in-vitro model of cerebral ischaemia 57 . Nevertheless, VEGF was shown to induce vascular permeability, resulting in oedema formation after stroke 56,57 .
In conclusion, the HIF PHD inhibitors possess the capability to stabilise HIF-1α and to upregulate expression of a number of hypoxia genes that promote ischaemic tolerance. The PHD inhibitors induce autophagy in both PC 12 cells and primary neurons, but do not cause apoptosis or cell death at 100 μM. DMOG is less potent in stabilising HIF-1α, autophagy induction and ischaemic tolerance compared to the 'clinical' PHD inhibitors. The pharmacological induced hypoxia adaptation requires coordination of intricate pathways and mechanisms, such as HIF, mTOR, autophagy. Further studies for precise understanding of the interplay of these mechanisms could lead to the development of new pharmacological strategies to minimize the damaging effects of strokes. Glucose-free Dulbecco's modified eagle medium, Neurobasal medium, Glucose-free Neurobasal medium, penicillin and streptomycin (10000 units/mL & 10000 µg/mL), TrypLE (synthetic trypsin), Glutamax supplement, sodium pyruvate (100 mM), Hank's Balanced Salt solution (HBSS), L-glutamine (200 mM), B27 supplement (50×, serum free), Pierce BCA protein assay kit and Pierce ECL western blotting substrates were from ThermoFisher Scientific (Loughborough, UK). Laemmli buffer(4x), 4-15% Mini-PROTEAN TGX Precast polyacrylamide gel, skimmed milk, Precision Plus Protein Dual Color Standard were from Bio-Rad (Hertfordshire, UK). Amersham ™ Protran ® Premium nitrocellulose blotting membranes were from VWR (Leicestershire, UK), RIPA (radio-immuno precipitation assay) buffer(10x) were from New England Biolabs Ltd (Hertfordshire, UK), Mouse anti-HIF-1α monoclonal antibody was from Novus Biologics (Abington, UK), rabbit anti-p62 (SQSTM1) polyclonal antibody, rabbit anti-LC3B polyclonal antibody, and rabbit polyclonal anti-β-actin antibody were from Abcam (Cambridge, UK), rabbit polyclonal anti-Beclin1 polyclonal antibody from Cell signalling technology (Danvers, MA, USA), goat polyclonal anti-mouse IgG HRP affinity, anti-rabbit IgG HRP affinity were from Dako, Agilent (Santa Clara, CA, USA). Monoclonal mouse anti-Tuj1 was obtained from Biolegend (San Diego, CA, USA). Vectashield mounting medium with DAPI was obtained from Vector Laboratories (Burlingame, CA, USA).
The Tetro cDNA synthesis kit and SensiFAST TM SYBR Hi-ROX kits were from Bioline Reagents Ltd (London, UK). The RNeasy plus Mini Kit was from Qiagen (Manchester, UK). The non-radioactive cytotoxicity assay kits were from Promega (Southampton, UK). The Guava cell dispersal reagent, Guava nexin kit, Guava instrument cleaning fluid, Guava Easycheck kit were from Merck Millipore (Burlingon, MA, USA). Plastic materials for cell cultures including pipettes, t25 cell culture vessel, 96-, 24-and 12-well plates, were from Greiner Bio-One (Gloucestershire, UK).
cell culture. PC12 cells. PC12 cells were cultured in 'complete' medium (high-glucose DMEM (containing 4 g/L glucose, L-glutamine and sodium bicarbonate, without pyruvate) supplemented with 5% FBS, 5% HS and 1% penicillin-streptomycin. PC 12 cells were incubated at 37 °C in a humidified atmosphere of 5% CO 2 in air (ESCO cell culture CO 2 incubator, Barnsley, UK). The plates and flasks used were coated with 5 mg/mL poly-D-lysine if required. www.nature.com/scientificreports www.nature.com/scientificreports/ Primary cortical rat neurons culture. Rat embryos E17-18 were removed from a pregnant rat, which was humanely killed under Schedule 1 according to the Animals Scientific Procedures Act (1986) (ASPA). The work is exempt from the need for Animal Welfare and Ethical Review Board (AWERB) approval under the ASPA and all subsequent amendments under both UK and European Law. All animals used in this study have been treated in accordance with ASPA guidelines. The embryonic brains were placed in HBSS on ice and dissected, and the cortices removed into an Eppendorf containing cold Neurobasal medium. The Neurobasal medium was then removed and replaced with pre-heated (37 °C) Neurobasal medium containing 0.05% trypsin and 100 µg/mL DNAse and the cortices dissociated through pipetting. When dissociated, the cortices were placed in a 15 mL falcon tube and placed in the incubator (37 °C) for 15 minutes. Thereafter, the cortices were removed and 6 mL pre-heated (37 °C) Neurobasal medium containing 10% FBS added in order to inactivate the trypsin. The cortices were further dissociated through pipetting before the cell suspension was centrifuged in a Harrier 18/80 R (MSE, East Sussex, UK) at 1500 rpm for 5 minutes. The supernatant was discarded, and the cells re-suspended in Neurobasal medium containing 2% B27 serum-free supplements, 2mM L-glutamine and 1% penicillin and streptomycin and filtered through a 70 micron filter. Neurons were then plated onto 5 mg/mL poly-D-lysine pre-coated plates at a density of approximately 0.15 × 10 6 cells per cm 2 . Thereafter, plates were placed in a standard incubator with a humidified atmosphere containing 5% CO 2 at 37 °C. Typically, experiments were performed at DIV 9-14.

Procedure of oxygen and glucose deprivation (OGD).
To mimic cerebral ischaemia in vitro, combined oxygen and glucose deprivation (OGD) was implemented. For OGD, the 'complete' medium was aspirated, and PC 12 cells rinsed twice with glucose-free media before incubation in glucose-free DMEM. PC 12 cells were then transferred into a purpose-built INVIVO 2 400 humidified hypoxia workstation (Ruskinn Technologies, Bridgend, UK) with atmospheric conditions established at 0.3% O 2 , 5% CO 2 , 94% N 2 and 37 °C for 6 hours. For OGD in primary cortical neurons, neurons were switched to glucose-free Neurobasal medium containing 2% B27 serum-free supplements, 2mM L-glutamine and 1% penicillin and streptomycin. All other experimental parameters remained constant with PC12 cells. an autophagy inducer (rapamycin) were initially dissolved in DMSO and subsequently diluted in the treatment appropriate culture medium to the indicated concentrations. To analyse the cytoprotective effects of the PHD inhibitors, final concentrations of 1, 10, 50 or 100 µM were applied in PC12 cells. The effect of rapamycin was studied at a final concentration of 1 and 10 µM. In primary rat cortical neurons, the PHD inhibitors DMOG and FG4592 were studied at varying concentrations as indicated. For the vehicle control group, a final concentration of 1% (v/v) aqueous DMSO was used throughout. To study the cytotoxicity of PHD inhibitors initial experiments involved exposure for 24 h to varying concentration during normoxia in 'complete' medium.
The cytoprotective effects of the PHD inhibitors were tested in an ischaemic tolerance model 25 . The cells were pre-treated with the PHD inhibitors for 24 hours, were then subjected to a period of reoxygenation (24 hours) by rapidly replacing the medium with normal 'complete' media and returning cells to normoxia, thereafter the cells were treated with an OGD (0.3% O 2 ) insult for 6 hours.
Assessment of cell viability. MTT assays. Cell viability was evaluated using the standard colorimetric assay for mitochondrial reductase catalysed reduction of yellow MTT to give a purple formazan product. PC12 cells were seeded on poly-D-lysine pre-coated 96-well plates at a density of 1.2 × 10 4 cells/well and cultured in 'complete' medium. Four hours prior to the completion of the experimental treatment, 10 μL of 5 mg/mL MTT solution diluted in PBS was added to the culture medium (final concentration, 0.5 mg/mL), and all samples incubated at 37 °C under treatment conditions. On completion of treatment, the supernatant was aspirated and the formazan crystals formed by surviving cells were solubilised in 50 μL DMSO, then incubated at 37 °C for 10 minutes. The optical density (OD) value of each well was determined by reading the absorbance at 540 nm using a microplate reader (Tecan Infinite M200 PRO, Switzerland). The DMSO background was subtracted from absorbance readings and the viability of cells for each treatment group was calculated based on Eq. 1; The viability of control cells (complete media in normoxic conditions) were assigned as 100% viable, while treatment samples were normalised against the control group OD value. Results are expressed as the percentage of cells possessing the ability to reduce MTT.
Lactate dehydrogenase (LDH) release assay. Cell-membrane integrity was analysed by measuring LDH activity of the culture medium using a non-radioactive cytotoxicity assay kit. PC12 cells were seeded on poly-D-lysine pre-coated 96-well plates at a density of 1.2 × 10 4 cells/well and cultured in 'complete' medium. Maximum LDH release control was generated by adding 10 µL 10x lysis solution to wells containing control cells, 45 minutes prior to adding the CytoTox reaction mixture. Following the completion of the experimental treatment, 50 µL of each www.nature.com/scientificreports www.nature.com/scientificreports/ sample medium was transferred to an unused 96-well flat bottom plate, into which, a further 50 µL of the reaction mixture containing tetrazolium salts was added. The subsequent mixture was incubated at room temperature for 30 minutes, protected from light. After 30 minutes, 50 µL of stop solution was added to each well and mixed. The amounts of formazan dye formed were assessed by measuring the absorbance with a microplate reader at 490 nm. The absorbance of the culture media background was subtracted from experimental values and the percent cytotoxicity calculated using Eq. 2: Experimental LDH Release Maximum LDH Release (%) 100 (2) The data are expressed as the mean percent of LDH release from the maximum control.
Trypan blue exclusion assay. Trypan blue exclusion was used to determine viable cells present in cell suspensions. This assay is based on the principle that live cells possess intact cell membranes and exclude trypan blue dye. PC12 cells were seeded at density of 1.2×10 6 cells in T25 and incubated in the complete DMEM media. Following treatments, the cell suspension was transferred in 15 ml sterile centrifuge tube and centrifuged for 5 minutes (100 relative centrifugal force (rcf)) using a Harrier 18/80 R centrifuge (MSE, London, UK). The medium was aspirated and the cells were re-suspended in complete DMEM media. Cells clusters were mechanically broken by pipetting. 100 μL of the suspension was stained with 100 μL 0.4% trypan blue for 5 minutes. 10 μL of the trypan-blue treated cell suspension was applied to each side of the haemocytometer. Live cells possessing a clear cytoplasm and non-viable cells possessing a blue cytoplasm were counted in each culture. Five samples were analysed for each treatment condition.

= + × live cells Viable cells Total number of cells Viable dead
Cell viability was expressed as percentage of viable cells in the total number of cells based on the Eq. 3. Immunofluorescence. Cell were fixed with 4% paraformaldehyde (PFA) for 15 minutes, and then were permeabilised using 0.1% Triton X-100 in PBS for 15 minutes and blocked by incubating with 5% bovine serum albumin (BSA) in PBS-T (PBS + 0.1% Triton X-100) for 1 hour in room temperature. This was followed by overnight incubation at 4 °C with primary antibody (mouse anti-Tuj1, 1:1000 in 1% BSA in PBS-T). Following three PBS washes, cells were incubated in secondary antibody (goat anti-mouse IgG-FITC, 1:200 in 1% BSA in PBS-T) for 2 hours at room temperature. Coverslips were then washed with PBS and mounted onto slides with Vectashield mounting medium with nuclear stain DAPI (4′, 6-diamidino-2-phenylindole). Images were taken by Nikon Eclipse 80i fluorescence microscope (Japan), Hamamatsu (C4742-95) digital camera and were double merged (consisting of FITC Tuj1; + DAPI stained nuclei) with NIS-Element BR 3.22.14 software (Nikon, Tokyo, Japan).

Flow cytometry analysis. A Guava
protein extraction and analysis. Cells were plated at a density of 1 × 10 6 in T25 flasks. After experimental treatments, the cells were washed with ice-cold PBS. Protein extraction buffer was made using 5 ml 1xRIPA buffer, 1 mM PMSF (50 μL) and a protease inhibitor cocktail (50 μL). Ice cold lysis buffer (0.5 ml per 75 cm 2 ) was added and mixed on orbital shaker for 15 minutes. Adherent cells were scraped using plastic cold scraper and transferred to an ice-cooled microcentrifuge tube. The sample was sonicated (3 cycles of 10 seconds pulse with 5 seconds interval) (MSE Soniprep 150 Plus, Wolflabs, UK), and centrifuged for 10 minutes. The resulting supernatant was transferred to a new microcentrifuge tube and the protein concentrations were determined using a BCA protein assay kit following the manufacturer's instructions. Twenty to 40 µg protein was denatured for 5 minutes in Laemmelli buffer at 95 °C. Samples were loaded on to pre-cast 4-20% polyacrylamide miniPROTEAN TGX gels (10 wells). 10 µl of Precision plus ProteinTM standards (10-250 kD) was loaded in each gel. The gels were run in 1x Tris/glycine/SDS running buffer in a BioRad Mini-PROTEAN ® tetra cell at 100 volts connected to a BioRad PowerPac Basic until the blue dye reached bottom (black line) of the gel. Following which the gels were removed from their cassettes into 1x transfer buffer. A trans-Blot SD Semi-Dry Electrophoretic transfer cell (Bio-Rad, Hertfordshire, UK) was used for the transfer. Gels were placed onto a nitrocellulose membrane and assembled with extra-thick filter paper. Transfer was performed www.nature.com/scientificreports www.nature.com/scientificreports/ at 12 volts for 1 hour. Membranes were blocked with 5% milk powder in 1x PBS-T for 1 hour then incubated overnight at 4 °C with primary antibodies against anti-HIF-1α (1:500), anti-Lc3b (1:500), anti-Beclin1 (1:500), anti-p62 (1:500) which were prepared in 1% milk powder in 1x PBS-T buffer. After the overnight incubation, membranes were washed in 1x PBS-T three times for 5 minutes each and were incubated for 1 hour in a horseradish peroxidise (HRP) conjugated secondary antibody, made in 1% milk powder in 1x PBS-T. After being washed three times with 1xPBS-T, the membranes were developed by Pierce ECL Western immunoblotting substrates, and imaged using FluorChem Western Blot imaging system (ProteinSimple, CA, USA) or ChemiDoc MP Imaging system (Biorad, UK). Thereafter, the membranes were stripped with mild-stripper buffer and re-probed with anti β-actin antibody (1:1000) and a subsequent HRP conjugated secondary antibody. The protein levels were quantified by densitometric analysis using Image J (NIH, USA). Values were normalized to β-actin and the corresponding controls.
Quantitative real-time polymerase chain reaction (qRt-pcR). RNA was extracted from cells using the RNeasy plus Mini Kit. The cell pellet was harvested by transferring the cell suspension to 15 mL centrifuge tube and centrifuging (5 minutes, 100 rcf). After which the supernatant was removed and 350 μL RLT plus containing 40 μM DTT was added to lyse the cell pellet. The homogenised lysate was then transferred to the supplied gDNA Eliminator spin column. The flow through was mixed with equal volume of 70% ethanol and transferred to an RNeasy spin column. The RNA pellets were washed with RW1 and RPE buffer following the manufacturer's instruction. The resultant RNA pellets were eluted in 50 μL nuclease-free water and the concentration measured on the NanoDrop 1000 Spectrophotometer (ThermoFisher Scientific, Loughborough, UK). The quality of the RNA samples was assessed using the 260/280 and 260/230 absorbance measurements. After extraction, each sample was diluted to a total RNA concentration 2 μg and transferred to a 1.5 mL RNase-free microcentrifuge tube on ice, ready for cDNA synthesis. Using Oligo(dT) 18 primers cDNA was synthesised using the Tetro cDNA synthesis kit in accordance with the manufacturer's protocol. Amplification of 100 ng cDNA template per reaction was performed using a Techne PROPLATE48, using the SensiFAST SYBR Hi-ROX kit. Using a Techne Prime Pro 48 Real-time qPCR machine (ThermoFisher Scientific, Loughborough, UK), an initial activation step was performed for 2 minutes at 95 °C before 40 cycles of a 3-step cycling program consisting of 95 °C for 5 seconds, 60 °C for 10 seconds and 72 °C for 15 seconds. The primers of a number of hypoxia genes [hypoxia-inducible factor 1α (Hif1α), BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (Bnip3), HIF target gene prolyl hydroxylase 2 (Phd2), vascular endothelial growth factor (Vegf), 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 1 (Pfkfb1), 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (Pfkfb3), lactate dehydrogenase A (Ldha)] used for analysis are in Table 1. Actin was used as a house keeping gene to normalise the relative levels of mRNA. β-Actin has been used as an internal loading control in PC12 cells for hypoxia treatment in previous studies 58,59 . Quantification of mRNA expression was performed using the comparative delta Ct method. Data analysis. Experiments employing on the 96 well plate reader were performed with 3-8 well replicates.
Independent experiments were performed in triplicate. The data is expressed as a mean value ± standard deviation (S.D.). One-way ANOVA with Tukey multiple comparison post-hoc was used to analyse comparisons among multiple groups. GraphPad PRISM 7 for Windows version 7.04 (GraphPad Software, Inc., CA, USA) was used for the analysis. Values of P < 0.05 were considered statistically significant. www.nature.com/scientificreports www.nature.com/scientificreports/