Neuroprotection by dimethyloxalylglycine following permanent and transient focal cerebral ischemia in rats

Abstract

Dimethyloxalylglycine (DMOG) is an inhibitor of prolyl-4-hydroxylase domain (PHD) enzymes that regulate the stability of hypoxia-inducible factor (HIF). We investigated the effect of DMOG on the outcome after permanent and transient middle cerebral artery occlusion (p/tMCAO) in the rat. Before and after pMCAO, rats were treated with 40 mg/kg, 200 mg/kg DMOG, or vehicle, and with 40 mg/kg or vehicle after tMCAO. Serial magnetic resonance imaging (MRI) was performed to assess infarct evolution and regional cerebral blood flow (rCBF). Both doses significantly reduced infarct volumes, but only 40 mg/kg improved the behavior after 24 hours of pMCAO. Animals receiving 40 mg/kg were more likely to maintain rCBF values above 30% from the contralateral hemisphere within 24 hours of pMCAO. DMOG after tMCAO significantly reduced the infarct volumes and improved behavior at 24 hours and 8 days and also improved the rCBF after 24 hours. A consistent and significant upregulation of both mRNA and protein levels of vascular endothelial growth factor (VEGF) and endothelial nitric oxide synthase (eNOS) was associated with the observed neuroprotection, although this was not consistently related to HIF-1α levels at 24 hours and 8 days. Thus, DMOG afforded neuroprotection both at 24 hours after pMCAO and at 24 hours and 8 days after tMCAO. This effect was associated with an increase of VEGF and eNOS and was mediated by improved rCBF after DMOG treatment.

Introduction

N-oxalylglycine is an inhibitor of most 2-oxoglutarate-dependent oxygenases, as well as of numerous other enzymes (Hewitson et al, 2004; Nagel et al, 2009). Hence, dimethyloxalylglycine (DMOG), an ester of N-oxalylglycine that penetrates cells readily, is expected to inhibit all prolyl-4-hydroxylase domain 1–3 (PHD 1–3) and factor-inhibiting hypoxia-inducible factor (HIF) (Jaakkola et al, 2001; Takeda et al, 2007). These enzymes are known as oxygen sensors and determine the stability and activity of the hypoxia-inducible transcription factors (Bruick and McKnight, 2001).

DMOG acts to stabilize HIF-1α expression under normal oxygen tension in cultured cells, at concentrations between 0.1 and 1 mmol/L (Jaakkola et al, 2001), whereas HIF-1α mediates changes in the expression of a large number of genes, with, for example, resultant increases in erythropoiesis, angiogenesis, and metabolism (Semenza, 2001, 2007). Inhibition of PHDs with other compounds has previously been reported to stabilize HIF-1α in the brain (Siddiq et al, 2005). DMOG itself, in models of ischemic heart and muscle disease and inflammatory bowel disease, is reported to attenuate proinflammatory mediators and create a pro-angiogenic microenvironment (Cummins et al, 2008; Milkiewicz et al, 2004; Ockaili et al, 2005). Nonetheless, the roles of DMOG in brain ischemia in in vivo experiments of brain ischemia are relatively unexplored (Chen et al, 2008). Moreover, there are no studies about the pharmacokinetics or pharmacodynamics of DMOG, and the diversity of 2-OG-dependent enzymes has only begun to unravel (Loenarz and Schofield, 2008).

This study was designed to characterize the effects of different DMOG treatment regimes on the outcome of permanent and transient middle cerebral artery occlusion (p/tMCAO) in the rat. Serial magnetic resonance imaging (MRI) was used to monitor both cerebral blood flow (CBF) and infarct volume. So far, it is still unclear whether elevated HIF levels are harmful or protective in the setting of cerebral ischemia and what role the time course for HIF activation and downstream gene effects play. Hence, we chose to compare two doses of DMOG and two treatment regimens in different models to see whether there are dissimilar responses. We show that DMOG exerts neuroprotective properties, but the increase in HIF-1α and putative neuroprotective target genes alone does not entirely explain the observed effects in outcome as the greatest protection was seen with only modest increases in HIF-1α protein levels.

Materials and methods

Animals and Experimental Design

This study complied with local ethical requirements and was performed in accordance with the Animals (Experimental Procedures) Act, 1986, under project licence from the United Kingdom Home Office.

Adult male Wistar rats (250 g±10%) were randomly assigned to the treatment with a total volume of 0.5 mL DMOG in saline. In the pMCAO experiment, animals were injected i.p. with (1) vehicle (control group), (2) 40 mg/kg BW, or (3) 200 mg/kg BW DMOG at 0, 8, and 16 hours. At 4 hours after the third dose pMCAO was performed, and 4 hours thereafter a final dose was given. At 1, 3, and 24 hours after pMCAO, MRI was performed to assess regional (r)CBF and infarct evolution. In the two tMCAO experiments with 24 hours and 8 days reperfusion, respectively, animals were treated with either (1) vehicle (control group) or (2) 40 mg/kg BW once immediately after reperfusion through i.v. injection. In this experiment, animals were scanned either at 3 and 24 hours or at 8 days only after reperfusion. Each experimental MCAO group consisted of 10 animals and after each experiment, animals were decapitated under deep anesthesia and the brains were collected. Sham surgery was undertaken in four animals per MCAO experiment and these animals did not receive DMOG. The surgeon was blinded to the treatments.

Middle Cerebral Artery Occlusion Model

Animals were anesthetized with 2% to 3% isoflurane in O₂/N₂O (1:3), and maintained with 1.5% isoflurane. Occlusion of the right MCA was induced in 70 rats (plus 12 sham animals) either permanently for 24 hours or transiently for 60 minutes with a reperfusion period of up to 8 days. Briefly, a silicone-coated filament was introduced in the right external carotid artery and pushed through the internal carotid artery to occlude the origin of the MCA. The induction of focal cerebral ischemia was verified with Laser Doppler (LD) rCBF monitoring (Oxford Optronics, Oxford, UK). After thinning of the skull bone using a microdrill, a flexible LD probe was attached to the skull 1 mm posterior to the Bregma and 5 mm lateral to the midline. A drop in rCBF below 30% from baseline after the insertion of the filament was considered to be sufficient for induction of focal cerebral ischemia. After withdrawal of the filament, an increase of up to 70% from baseline was considered as successful reperfusion. The tail artery and vein were cannulated for measurement of arterial blood gases and for delivering the treatment in the tMCAO experiments. During surgery, the body core temperature of 37°C was controlled by a rectal thermometer connected to a feedback-controlled heating pad in all animals.

The rats were allowed to recover and were kept on a heating blanket. Buprenorphin (0.1 mg/kg BW, subcutaneously) was injected once after surgery in all animals; further treatments were applied only if animals showed signs of pain. Animals had free access to food and water and were re-anesthetized as above for MRI measurements.

Behavior

Rats were assessed 24 hours after pMCAO, and 24 hours and 8 days after tMCAO with a functional neuroscore for neurological outcome on a scale from 3 (most severe) to 18 (no deficit) (Garcia et al, 1995). Behavioral assessment consisted of scoring the spontaneous movement, the sensoric function, and the motor function.

Magnetic Resonance Imaging Protocol

Animals were anesthetized as described above and positioned in a quadrature birdcage coil (i.d. 5 cm). The MRI data were acquired using a 7-T horizontal bore magnet with a Varian Inova spectrometer (Varian, Palo Alto, CA, USA). During MRI scanning, the electrocardiogram and body temperature were monitored, and hypothermia was prevented with a hot water blanket system, although spontaneous hyperthermia could not be corrected.

Scout images were acquired to define a single 1 mm coronal slice located at the center of the striatum (5 mm posterior to the rhinal fissure), which was selected for the full imaging protocol. T₂-weighted (T2WI) images were acquired using a fast spin-echo sequence (TR=3.0 seconds; TE=37.5 milliseconds). Arterial spin labeling perfusion-weighted imaging (ASL-PWI) was performed using a flow-sensitive alternating inversion recovery labeling strategy, with an inversion time of 0.7 milliseconds (Kim, 1995). On the rCBF maps, calculated as the mean across the 32 ASL-PWI averages, the ipsilateral and contralateral hemispheres were traced and signal intensities were measured. Ipsilateral rCBF was expressed as percentage from the contralateral hemisphere after correction for baseline values.

At 1 and 3 hours after pMCAO, single-slice diffusion-weighted images were acquired using a navigated pulsed-gradient spin-echo sequence (TR 1.5 seconds; TE 0.0365 seconds; b=125, 500, and 1,000 s/mm², Δ=17.5 milliseconds; δ=12.5 milliseconds) as described earlier (Broom et al, 2005). At all other time points, a multi-slice diffusion-weighted image data set was acquired (parameters as above, 9 × 1 mm² slices, inter-slice gap=0.5 mm). Diffusion gradients were applied in three orthogonal directions and apparent diffusion coefficient (ADC) ‘trace' maps were calculated from the DW images obtained. Navigator echoes were used for motion correction (Ordidge et al, 1994). Mean ADC levels in cortex and striatum were calculated from one representative slice through the center of the striatum at each time point and values were expressed as percentage reduction from the contralateral side. Spin-echo T₁-weighted (T1WI) images (TR 500 milliseconds; TE 20 milliseconds) were acquired both before and 10 minutes after Gd injection to look for image enhancement owing to blood-brain-barrier (BBB) permeability. Lesion volumes were calculated from the ADC maps and also T2 and T1W images in a blinded manner and corrected for edema by multiplication of infarct volumes by the ratio of the volumes of contra- and ipsilateral hemispheres.

Tissue Processing

After decapitation, brains were quickly removed and rinsed with chilled (4°C) phosphate-buffered saline solution. For analysis of protein expression with immunoblotting, a block of tissue from the ipsilateral MCA territory was dissected on ice 4 mm from the anterior and 4 mm from the posterior pole, using a rat brain matrix. For analysis of mRNA expression with real-time reverse transcriptase polymerase chain reaction (RT-PCR), a small tissue sample (10 to 15 mg) was microdissected from the ipsilateral fronto-parietal cortex.

Western Blotting

For Western blot analysis, brain tissue (as described above) was homogenized on ice in whole cell lysis buffer, containing 10 mmol/L Hepes, 10 mmol/L NaCl, 4% (w/v) CHAPS (3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate), 5 mmol/L EDTA, 100 mmol/L iodacetamide, 100 mmol/L PMSF (phenylmethanesulfonylfluoride), 250 mmol/L sucrose and supplemented with a protease inhibitor cocktail (Roche Diagnostics, Burgess Hill, UK), using a glass/glass homogenizer. The homogenate was incubated on ice for 15 minutes, centrifuged at 4°C for 10 minutes at 1,000 g, and the supernatant was centrifuged again. The final supernatant was stored in aliquots at −80°C until determination of the protein content using the Dc Protein assay (BioRad, Hemel Hempstead, UK). For electrophoresis 25 μg of protein per sample was precipitated by methanol/chloroform, resolved by SDS-PAGE in 7.5% or 10% polyacrylamide slab minigels, transferred to polyvinylidene fluoride membranes and stained with Ponceau S to determine the total protein content per lane. Immunoblotting was performed after blocking with 5% (w/v) milk powder by probing with the following antibodies: HIF-1α (1:250, Abcam, Cambridge, UK), heme-oxygenase 1 (HO-1, 1:500, Stressgen, Ann Arbor, MI, USA), vascular endothelial growth factor (VEGF) (1:250, Novus, Newmarket Suffolk, England), endothelial nitric oxide synthase (eNOS) (1:250, Santa Cruz, Santa Cruz, CA, USA), erythropoetin (EPO) (1:200, Santa Cruz), and β-actin (1:5,000, Abcam) as second loading control. Immunoreactive bands were quantified using the ECL Plus Western Blot Detection Reagents (Amersham, GE Healthcare, Little Chalfont, Buckinghamshire, UK). Immunoreactive species were quantified by densitometry (UPC; VisonWorksLS (Version 6.0), Biospectrum AC, UVP, Cambridge, UK) and corrected for loading using the quantified data from the Poncau S staining and normalized for sham values.

Real-Time Reverse Transcriptase Polymerase Chain Reaction

The ipsilateral cortex samples (10 to 15 mg) were used for RNA extraction. mRNA was isolated and enriched directly from the tissues using the RNeasy Lipid Tissue Kit (Qiagen, Crawley, West Sussex, UK) according to the manufacturer's protocol. RNA quality control was performed by optical density measurements and degradation was excluded after electrophoresis in a 1% agarose gel with ethidium bromide (10 mg/mL). cDNA templates were generated from 1 μg RNA using the QuantiTect Reverse Transcription with integrated removal of genomic DNA (Qiagen). Real-time PCR was performed using a BioRad iQ iCycler Detection System (BioRad Laboratories, Hemel Hempstead, UK) with SYBR green fluorophore in a 25-μL reaction with 50 ng cDNA. Protocols for each primer set were optimized and primer efficiency was tested using five serial 5 × dilutions of template cDNA. The primers used are summarized in Table 1. The protocols were as follows: 95°C, 10 minutes; 40 cycles of (95°C, 15 seconds; 60°C, 60 seconds). All reactions were performed in duplicate for every sample and melting curve analysis was performed after each run. β-Actin was used as housekeeping gene. Data were quantified using the 2^−δCT method and relative fold changes were corrected for β-actin and normalized for sham values.

Table 1. List of primers used for real-time RT-PCR.

Primer	Symbol	Sequence (5′-3′)
Heme oygenase 1	HO-1	Fwd: TCTGGTCTTTGTGTTCCTCTGT
		Rev: CAGAAGAGGCTAAGACCGCCT
Vascular endothelial growth factor	VEGF	Fwd: TTACTGCTGTACCTCCAC
		Rev: ACAGGACGGCTTGAAGATA
Endothelial nitric oxide synthase	eNOS	Fwd: TGGCAGCCCTAAGACCTATG
		Rev: AGTCCGAAAATGTCCTCGTG
Glucose transporter 1	GLUT-1	Fwd: GGTGTGCAGCAGCCTGTGTA
		Rev: GACGAACAGCGACACCACAGT
Erythropoetin	EPO	RT2 qPCR Primer Assay (SA Biosciences, Frederick, MD, USA)
Endothelin-1	END-1	Gene Detection Kit (Primer Design, Southampton, UK)
Pyruvate dehydrogenase kinase 1	PDK-1	Fwd: GGACTTCGGGTCAGTAAATGC
		Rev: TCCTGAAGAAGATTATCGGGAA
B-cell CLL/lymphoma 2	BCL-2	Fwd: GGAGCGTCAACAGGGAGATG
		Rev: GATGCCGGTTCAGGTACTCAG
BCL2/adenovirus E1B 19 kDa interacting protein 3	BNIP-3	RT2 qPCR Primer Assay (SA Biosciences)
Bcl2-associated X protein	BAX	Fwd: CCAAGAAGCTGAGCGAGTGTCTC
		Rev: AGTTGCCATCAGCAAACATGTCA
β-Actin	β-Actin	Fwd: TGCCCTAGACTTCGAGCAAGA
		Rev: CATGGATGCCACAGGATTCCATAC

Statistical Analysis

The primary outcome of the study was infarct volume at the end point of the experiments, that is 24 hours after pMCAO and 24 hours or 8 days after tMCAO. Sample size calculations were performed for each experiment, indicating a minimum of nine animals per group if a 50% reduction in infarct volumes and a s.d. of 40% is expected with an α- and β-error of 5%. However, due to predefined criteria, some animals had to be excluded from the analysis. Two animals did not show sufficient reduction of rCBF and one animal did not exhibit sufficient reperfusion during LD monitoring. Another animal of the pMCAO control group was excluded as its infarct volume was more than two s.d. smaller than the average of this group (the analysis including this animal is shown in Supplementary Figure 2), In addition, eight animals died prematurely, two animals suffered nonlethal subarrachnoid hemorrhage, and one animal was killed due to persistent fits. Thus, 6 to 10 animals per group were used for the final analysis, with a power still ⩾90% for each experiment. Analysis of variance with Dunnett's multiple comparison and paired as well as unpaired t-test or Kruskall–Wallis test with Dunn's multiple comparison and Mann–Whitney test were used where appropriate to assess statistical differences between groups. Mortality rates were compared by Fisher test or likelihood ratio. Pearson's correlation was used to assess the relationship between rCBF and infarct volumes. Statistical significance was declared if the two-sided P-value was <0.05. Summary data are presented as mean±s.e.m.

Results

DMOG did not Change the Mortality or Basic Physiology After Middle Cerebral Artery Occlusion

In total, nine animals died before the end point due to complications of anesthesia, space-occupying cerebral ischemia, or epileptic fits. The mortality rates were not statistically different between control and DMOG groups (4 versus 4, and no significant differences were evident between pMCAO and tMCAO). There were no significant differences in rCBF values, measured using Laser Doppler Flow (LDF), between any of the groups during MCAO or during the first 20 minutes of reperfusion (Supplementary Figures 1A and 1B). Some animals developed fever up to 39°C after pMCAO due to substantial hypothalamic infarction, a phenomenon that was not observed after tMCAO. Overall body temperatures were higher in the control group as compared with the DMOG-treated group after pMCAO; however, this did not reach statistical significance. After tMCAO, body temperatures remained normal in all groups (Supplementary Figures 1A and 1B). Mean heart rates during MRI were similar in all groups (data not shown); the arterial blood gas parameters, pH, pCO₂, and pO₂ were within normal ranges and did not differ between time points or groups (Tables 2A and 2B).

Table 2A. Arterial blood gas parameters, before and 1.5 hours after pMCAO.

Time point	Group	Arterial blood gas parameter (mean±s.e.m.)
		pH	pCO2(mm Hg)	pO2(mm Hg)
Pre-pMCAO	Control	7.41±0.01	41±0.6	119±10
	40 mg/kg DMOG	7.40±0.01	42±1.5	117±8
	200 mg/kg DMOG	7.41±0.01	43±1.8	113±8
Post-pMCAO	Control	7.39±0.01	44±3.4	111±7
	40 mg/kg DMOG	7.41±0.02	42±3.1	112±8
	200 mg/kg DMOG	7.40±0.03	46±4.3	115±10

DMOG, dimethyloxalylglycine; pMCAO, permanent middle cerebral artery occlusion.

There were no significant differences between groups or time points.

Table 2B. Arterial blood gas parameters before and immediately after tMCAO.

Time point	Group	Arterial blood gas parameter (mean±s.e.m.)
		pH	pCO2 (mm Hg)	pO2 (mm Hg)
Pre-tMCAO	Control	7.39±0.01	46±1.9	118±4
	40 mg/kg DMOG	7.38±0.01	47±1.5	117±5
Post-tMCAO	Control	7.4±0.02	45±1.8	121±4
	40 mg/kg DMOG	7.38±0.01	45±1.0	120±4

DMOG, dimethyloxalylglycine; tMCAO, transient middle cerebral artery occlusion.

There were no significant differences between groups or time points.

DMOG Reduced Ischemic Injury and Improved Behavior After Both Permanent and Transient Middle Cerebral Artery Occlusion

MCAO induced focal reductions in ADC within the MCA territory after both pMCAO and tMCAO, although the spatial extent of the ADC reduction was considerably greater in the pMCAO animals (Figures 1A and 1B, 2A and 2B). Dimethyloxalylglycine reduced the extent of lesions, defined as areas of reduced ADC, across all pMCAO groups after 24 hours (P=0.003, analysis of variance). Interestingly, treatment with 40 mg/kg DMOG (265 mm³ mean difference to control, 95% confidence interval (CI): 102 to 427, P<0.05) offered more protection than with 200 mg/kg (194 mm³, 95% CI: 36 to 352, P<0.05, Figures 1A and 1B). There were no differences in edema within the groups after quantification of hemispheric volumes. Except at 24 hours in the striatum, mean ADC levels in the ipsilateral striatum and cortex of control animals were lower when compared with 40 or 200 mg/kg DMOG-treated pMCAO animals at all time points (Figure 1C). In keeping with this, behavior improved in these animals after DMOG treatment (P=0.008, Kruskall–Wallis test). Compared with the control group (8.5±0.4), DMOG-treated animals generally had a better neuroscore (40 mg/kg: 12.3±1, 200 mg/kg: 11±0.9); however, this reached statistical significance only in the 40 mg/kg group (P<0.05 versus control; Supplementary Figure 3A). We investigated whether this dose-dependent effect could be attributed to changes in CBF by recording ASL-derived rCBF MRI data at baseline, 1, 3, and 24 hours after pMCAO. The rCBF decreased in all animals compared with baseline values. At 1 and 24 hours, there was a trend toward improved rCBF values in the 40 mg/kg group compared with control and 200 mg/kg groups (P<0.1; Figures 1D and 1E). Decreased mean rCBF over time was strongly correlated with increasing infarct volumes in all animals and infarct volumes were exacerbated in animals in which the mean rCBF values remained below 30% of the contralateral side (Pearson's correlation r=−0.641, P<0.01, Figure 1F). In four animals from the 40 mg/kg group, the mean rCBF over time remained above the critical threshold of 30% from the contralateral hemisphere (P<0.01) as compared with none in any other group.

Figure 1.

Results of the magnetic resonance imaging experiments from the pMCAO study. (A) Examples of diffusion-weighted imaging data obtained from one representative animal per group; on diffusion-weighted images, infarcts can be detected as areas of hyperintensity and on the derived ADC maps as areas of hypointensity. Calculated ADC infarct volumes are shown in (B), with ^* indicating P<0.05 versus control. Mean ADC reductions after pMCAO are presented in (C) for the striatum and the cortex. The inlay indicates the regions of interest, which have been measured. ^*P<0.05 for both treatment groups versus control. (D) Representative CBF maps for each group obtained using ASL-PWI. The rCBF values were not significantly different between groups, but at 1 and 24 hours there was a trend towards higher rCBF values in the 40 mg/kg DMOG group compared with the other groups (E). ^#P<0.1. (F) The correlation between mean rCBF over time and final infarct volumes is significant for all animals. Four animals of the 40 mg/kg DMOG group had values above the critical threshold of 30% from normal (P<0.05). ADC, apparent diffusion coefficient; DMOG, dimethyloxalylglycine; pMCAO, permanent middle cerebral artery occlusion; rCBF, regional cerebral blood flow.

Figure 2.

Results of the magnetic resonance imaging experiments from the tMCAO study. (A) Representative examples of diffusion-weighted imaging data from the 24-hour reperfusion experiment. (B) Graph showing the calculated ADC infarct volumes with $ indicating P<0.01 versus control. (C) Representative examples of T2W and pc-T1W imaging data obtained from the 8-day reperfusion experiment. (D) Graph showing the calculated T2W infarct volumes and the pcTW volumes of BBB breakdown, with ^* indicating P<0.05 versus control. (E) Representative examples of the ASL-PWI CBF maps after tMCAO; hyperintensities represent hyperperfusion. (F) Graph showing the calculated rCBF values of the ipsilateral hemisphere, with $ indicating P<0.01 in a paired test. The inter-group comparison showed no significant differences. (G) Graph of rCBF/infarct volume ratios, with $ indicating P<0.01 versus control in an unpaired test. ADC, apparent diffusion coefficient; ASL-PWI; arterial spin labeling perfusion-weighted imaging; tMCAO, transient middle cerebral artery occlusion; rCBF, regional cerebral blood flow.

As the lower dose exhibited a greater effect on infarct volumes and functional outcome, we investigated whether a single dose of 40 mg/kg DMOG would be sufficient to deliver protection in a tMCAO model. Owing to the clinical relevance, a postischemia treatment immediately after reperfusion was chosen. Indeed, at 24 hours but not 3 hours (42±27 mm³, 95% CI: −16 to 100, P=0.15) after reperfusion, ADC lesion volumes were significantly attenuated by 127±38 mm³ (95% CI: 45 to 208, P=0.0045) due to DMOG treatment (Figures 2A and 2B). This protection persisted for at least 8 days, and a significant reduction was evident not only in infarct volumes (72±30 mm³ reduction, 95% CI: 8 to 136, P=0.03) but also in the volume of BBB breakdown (87±40 mm³ reduction, 95% CI: 1 to 172, P=0.047) as measured with T2WI and pcT1WI, respectively (Figures 2C and 2D). There were no differences in edema within the groups as measured by quantification of hemispheric volumes at 24 hours and 8 days after tMCAO. Again, the significant reduction in infarct size was translated into an improved functional outcome after 24 hours (control 12.03 versus 40 mg/kg DMOG 14.65, P<0.001) and 8 days (control 13.8 versus 40 mg/kg DMOG 15.63, P<0.01) of reperfusion (Supplementary Figure 3B). The raw rCBF values, measured by ASL-PWI after tMCAO, did not differ significantly at 3 hours, 24 hours, and 8 days of reperfusion between the groups. However, there was a significant increase in rCBF from 84%±10% to 135%±9% between 3 and 24 hours in the DMOG group (P<0.01); this was not observed in the control group. Interestingly, the ratio of rCBF to infarct volume was significantly higher in the DMOG group after 24 hours (P<0.01), indicating a greater hyperperfusion despite smaller infarcts in this group (Figures 2E–G). Sham-operated animals showed no abnormalities on diffusion-weighted image, T2WI, or T1WI, showing symmetrical rCBF on ASL-PWI rCBF maps and a normal neuroscore.

Effects of DMOG on Hypoxia-Inducible Factor-1α Protein Levels in the Brain

After 24 hours pMCAO HIF-1α increased dose dependently with DMOG treatment (P<0.05) from 215±17 normalized immunoreactivity (ni) (control) to 291±36 ni (40 mg/kg) and 370±26 ni (200 mg/kg). When individually compared with control, HIF-1α was only significantly increased after 200 mg/kg DMOG (P<0.05; Figures 3B and 3C). After 24 hours of tMCAO, 40 mg/kg DMOG modestly, but not significantly increased HIF-1α (359±14 versus 464±21 ni, P<0.01), whereas after 8 days of reperfusion, no substantial difference was seen (309±36 versus 341±22, Figures 4A and 4B). However, all groups and time points were significantly different from sham levels (P<0.05).

Figure 3.

(A) Cortical mRNA levels of selected genes of interest after 24 hours of pMCAO as measured by RT-PCR. Values are normalized for sham levels and ^* indicates P<0.05 versus control. (B) Representative Western blots after 24 hours of pMCAO. (C) The calculated immunoreactivities after correction for PonceauS staining and normalization for sham values. ^*P<0.05 versus control. pMCAO, permanent middle cerebral artery occlusion.

Figure 4.

(A) Representative Western blots after 24 hours and 8 days of reperfusion in a 60-minute tMCAO experiment. (B) Calculated immunoreactivities after correction for PonceauS staining and normalization for sham values. ^*P<0.05 versus control. tMCAO, transient middle cerebral artery occlusion.

Effects of Dimethyloxalylglycine on Gene and Protein Expression of Hypoxia-Inducible Factor Targets in the Brain

As most of the cell protective properties of DMOG have been assigned to its inhibition of PHDs, we investigated whether the effects of DMOG on infarct size and functional outcome were associated with increased HIF and hence induction of neuroprotective target genes.

The expression levels of a large battery of genes that are known to be HIF inducible were examined 24 hours after pMCAO (Figure 3A). The greatest upregulation was seen in HO-1 (up to 4,663%±84% compared with sham) and VEGF (up to 3,953%±178% compared with sham). VEGF and eNOS were the only genes that were significantly upregulated by pMCAO plus DMOG treatment as compared with pMCAO alone (P<0.05 versus control and sham for both. treatments). Glucose transporter 1, EPO, and endothelin-1 also showed a graded upregulation with DMOG treatment; however, these changes only reached statistical significance when compared with sham (P<0.05) and not with control pMCAO. Pyruvate dehydrogenase kinase 1 showed no significant changes after either pMCAO alone or pMCAO plus DMOG treatment. Genes involved in apoptosis (B-cell CLL/lymphoma 2, Bcl2-associated X protein, and BCL2/adenovirus E1B 19 kDa interacting protein 3) showed only minor changes and, interestingly, BCL2/adenovirus E1B 19 kDa interacting protein 3 was consistently but not significantly downregulated in all groups when compared with sham (Figure 3A).

In analogy to the PCR experiments, several HIF-regulated targets were chosen for confirmation with Western blotting (Figures 3B and 3C). All proteins of interest were significantly altered after pMCAO when compared with sham (P<0.05), except for eNOS and EPO control. As for the RT-PCR findings, HO-1 levels were not different between the three experimental groups. Interestingly, eNOS protein levels were significantly upregulated after 40 mg/kg DMOG (213±24 ni versus 129±9 ni, P<0.05), but not after 200 mg/kg DMOG treatment when compared with control. VEGF increased significantly in both treatment groups (700±32 ni resp. 533±92 ni versus 158±7 ni, P<0.05), but only the higher dose significantly raised EPO levels from control (284±15 ni versus 137±12 ni, P<0.05).

As VEGF and EPO protein showed the greatest increases from controls after pMCAO and have previously been implicated in neuroprotection, we also measured their expression 24 hours and 8 days after 60 minutes of tMCAO (Figures 4A and 4B). Except for VEGF levels in the control group after 8 days of reperfusion, both proteins were significantly increased in all other groups when compared with sham. However, only VEGF showed a significant upregulation after 8 days in the DMOG group compared with controls (195±10 versus 114±10, P<0.05). Interestingly, the regulation pattern of VEGF and EPO was different from that of the pMCAO experiment. After tMCAO, EPO was more induced than VEGF, which was the opposite case to that after pMCAO.

Discussion

In this paper, we show that treatment with DMOG led to neuroprotection after MCAO as expressed by reduced infarct sizes and improved functional outcome. However, the protection after 24 hours pMCAO due to pre- and posttreatment DMOG was more potent with the lower dose of 40 mg/kg. Treatment with 200 mg/kg DMOG failed to deliver a significant effect on behavior after pMCAO, but was the only treatment that significantly increased HIF-1α levels compared with cerebral ischemia itself after 24 hours. Changes in rCBF did not entirely explain the observed DMOG effect, but rCBF after pMCAO was better preserved in the 40 mg/kg DMOG group than in any other group. To confirm the effect of the lower DMOG dose, we showed that a single treatment of 40 mg/kg after tMCAO was able to reduce infarct volumes and also the extent of BBB breakdown, which translated into improved behavior after both 24 hours and 8 days. rCBF after tMCAO at 24 hours recovered significantly better with 40 mg/kg DMOG than vehicle control. Genes associated with the observed neuroprotection were foremost eNOS, but also VEGF and EPO.

Originally explored as a CPH inhibitor, DMOG is now recognized as an inhibitor of many, but not necessarily all, 2-OG oxygenases, as well as numerous other enzymes. Inhibition of HIF PHDs has become an attractive target for the prevention and therapy of cardio- and cerebrovascular ischemic disease (Nagel et al, 2009). Pharmacological inhibitors of the PHDs and factor-inhibiting HIF have already shown neuroprotective properties in vitro and in vivo, by inducing a pro-survival and angiogenic gene response (Baranova et al, 2007; Siddiq et al, 2005). Several genes that have been implicated in neuroprotection are regulated by HIF-1α, such as eNOS, VEGF, and EPO (Semenza, 2007). However, it has also been shown that HIF-1α can upregulate apoptosis-related genes, like BCL2/adenovirus E1B 19 kDa interacting protein 3 (Baranova et al, 2007). Hence, HIF-augmenting therapies will have to find a way to balance beneficial versus potentially deleterious effects.

In our experiments, a dose-dependent effect of pre- and posttreatment with DMOG in experiments with pMCAO was observed. Interestingly, the lower dose offered more protection than the higher dose and, surprisingly, only the higher dose significantly increased levels of HIF-1α after 24 hours. By contrast, RNA and protein levels of HIF targets VEGF and eNOS were similarly altered by DMOG treatment. VEGF RNA and protein increased significantly after both doses of DMOG. The same was the case for eNOS RNA, but protein levels were significantly higher only after 40 mg/kg DMOG. NO possesses vasodilatory, antiinflammatory, antithrombotic, and antiproliferative properties. Augmentation of NO production by eNOS, in contrast to neuronal nitrous-oxide synthase (nNOS), increases CBF (Endres et al, 2004). Mice lacking eNOS expression show a greater degree of hemodynamic compromise after MCAO and hence the product of eNOS activity (i.e., NO) may protect the brain after focal cerebral ischemia by improving blood flow within the penumbral zone (Lo et al, 1996). VEGF can have a dual function in the brain mediating both neuroprotection and BBB permeability through VEGF/VEGFR2/Flk1 and PI3K/Akt pathways (Kilic et al, 2006; Laudenbach et al, 2007; Wang et al, 2007). Recently, DMOG treatment of rat pups after cerebral ischemia led to increased BBB permeability and brain edema in association with raised levels of VEGF (Chen et al, 2008). In this severe hypoxic ischemia model, in contrast to pMCAO, reperfusion occurs and the vessel leakage effects of VEGF might outplay beneficial ones. Intrinsic and extrinsic EPO are known to be neuroprotective in vitro and in vivo (Digicaylioglu and Lipton, 2001; van der Kooij et al, 2008), although we only found a significant EPO stimulation by DMOG after pMCAO with the higher dose, next to HIF-1α increase. Hence, it is likely that the cumulative effects of eNOS, VEGF, and partly EPO are responsible for the improved outcome parameters after pMCAO (Fan et al, 2009). In particular, our data suggest that eNOS most likely mediated the modest improvements in rCBF with the lower DMOG dose after pMCAO. The strong correlation of CBF and infarct volume with an exacerbation of infarct volumes below a mean CBF value of 20% to 30% from normal is in line with very early work on CBF in this model (Buchan et al, 1992; Nagasawa and Kogure, 1989). Moreover, this enhances the critical importance of such a threshold in this model, wherein small changes in rCBF can translate in much bigger effects in outcome.

In agreement with the literature, HO-1 was hugely increased after pMCAO compared with sham animals (Fu et al, 2006); however, DMOG treatments did not have any further significant effects on mRNA or protein levels of HO-1, thereby indicating no association with protection. Pyruvate dehydrogenase kinase 1, a key regulator for glycolysis, did not change in the setting of pMCAO, although it has been established as an important mediator of resistance to muscle and liver ischemia in mice lacking PHD-1 (Aragones et al, 2008; Schneider et al, 2010). However, those tissues have a huge potential in upregulating their gylcolytic metabolism during hypoxia/ischemia, which is not the case for the brain, and it seems unlikely that increased glycolysis is a potential mechanism of PHD/HIF-related neuroprotection.

The physiological HIF-1α response after transient MCAO seems to be biphasic, with an early peak (a few hours) being associated with pro-apoptotic genes and a later peak (after 24 hours) leading to a pro-survival gene expression (Baranova et al, 2007). At 24 hours after pMCAO, we did not find evidence of upregulation of pro-apoptotic markers, but we did find raised HIF-1α levels at both 24 hours and 8 days of tMCAO compared with sham. HIF-1α was also increased after 24 hours after DMOG treatment, but this did not reach significance when compared with the control group. Nonetheless, a single dose of 40 mg/kg DMOG delivered i.v. immediately after 60 minutes of tMCAO was able to offer protection that was evident at 24 hours and maintained up to 8 days of reperfusion. rCBF in the DMOG tMCAO group increased significantly to 35% greater than the contralateral hemisphere after 24 hours. Postischemic hyperperfusion has long been documented in animal stroke models and is the hallmark of efficient recanalization and reperfusion. However, there is no consensus of opinion as to the beneficial or detrimental effects of such hyperperfusion on the size of the eventual infarct (Marchal et al, 1999; Pan et al, 2007). Interestingly, in our experiments, it was associated with smaller infarcts and was specific to DMOG treatment. Recently, it has been shown that haplodeficiency of PHD-2, by upregulation of HIF-2 and VEGFR-1 and VE-cadherin, leads to improved tumor perfusion and oxygenation and inhibited tumor cell invasion, intravasation, and metastasis (Mazzone et al, 2009). We did neither measure HIF-2, VEGFR-1, nor VE-cadherin; hence, a role for these mediators in cerebral perfusion remains speculative, but VEGF was also significantly altered after tMCAO and DMOG treatment compared with controls. In contrast to the cerebral hypoxia and ischemia model of Chen et al., VEGF increased to similar levels in control and DMOG-treated groups 24 hours after tMACO. However, in DMOG-treated tMCAO animals, VEGF levels remained significantly elevated at 8 days after tMCAO, whereas levels had returned to baseline (sham) levels by this time point in control animals. This may indicate a positive role for VEGF in postischemic remodeling of the brain, rather than a detrimental effect by causing BBB breakdown (Sun and Guo, 2005).

There are some interesting confounds to the current study that lead to the conclusion that the biological effects seen in this model might not entirely be explained by HIF-mediated changes in gene expression. Protection was observed with only a moderate increase in HIF-1α, and despite a further increase in HIF-1α protein levels, as well as raised expression levels of VEGF, eNOS, and EPO, the higher DMOG dose failed to improve behavior. It is possible that a bigger sample size would reveal a protective effect of higher doses of DMOG on behavior or that an early peak in HIF-1α was missed. On the other hand, non-HIF-related effects of DMOG might also be responsible for the observed effects. For instance, it has been described that DMOG can activate antioxidant gene expression through the nuclear factor erythroid 2-related factor 2 pathways in microvascular endothelial cells. Nuclear factor erythroid 2-related factor 2 induces the expression of various antioxidant proteins with critical roles in the adaptive responses to oxidative stress (e.g., HO1 or glutathione peroxidase 2). Thereby, DMOG resulted in significant reduction of apoptosis in the postischemic myocardium (Natarajan et al, 2009). Moreover, DMOG can exert cell-toxic properties, especially at higher doses (Asikainen et al, 2005). DMOG is a nonselective PHD inhibitor expected to interfere with all 2-OG-dependent oxygenases affecting a number of important cellular processes such as DNA repair, carnitine biosynthesis, and collagen modifications, and most likely many other functions within the cell, including the Krebs cycle, which were not accounted for (Loenarz and Schofield, 2008). Of major importance, different levels of HIF might affect different targets: targets we may not know yet and some of which may not be beneficial. However, this should not necessarily stop us from finding therapeutic utility at this point. It is evident that the right amount of HIF is needed at the right time, as it has been shown that sustained hypoxia/ischemia and HIF activation may trigger negative feedback loops (Webb et al, 2009). The recent failure of the EPO trial in human stroke, however, may illustrate the theory that the upregulation of a whole protective cellular environment rather than the increase of one potentially protective downstream player is necessary (Ehrenreich et al, 2009).

In summary, we show a neuroprotective effect of DMOG pre- and posttreatment in a pMCAO model and posttreatment in a tMCAO model. Our data suggest that this effect was mediated by an induction of a pro-survival gene expression pattern and, at least partly, by improved rCBF, although a clear association with increased levels of HIF-1α at 24 hours and 8 days was not seen. Although DMOG may not be a suitable candidate drug for successful clinical translation, this study gives insights into the mechanisms underlying the concept of PHD inhibition for treatment of cerebral ischemia and it highlights the importance of assessing physiological variables, such as CBF, in studies that investigate neuroprotection. It also emphasizes the need for in vivo studies with selective PHD inhibitors and selective genetic PHD and HIF suppression, to assess the cell-tissue and disease-specific properties of the HIF–PHD axis.

The authors declare no conflict of interest.