Reduced Vascular Endothelial Growth Factor and Capillary Density in the Occipital Cortex in Dementia with Lewy Bodies

Abstract

In dementia with Lewy bodies (DLB), blood flow tends to be reduced in the occipital cortex. We previously showed elevated activity of the endothelin and angiotensin pathways in Alzheimer's disease (AD). We have measured endothelin‐1 (ET‐1) level and angiotensin‐converting enzyme (ACE) activity in the occipital cortex in DLB and control brains. We also measured vascular endothelial growth factor (VEGF); factor VIII‐related antigen (FVIIIRA) to indicate microvessel density; myelin‐associated glycoprotein (MAG), a marker of ante‐mortem hypoperfusion; total α‐synuclein (α‐syn) and α‐synuclein phosphorylated at Ser129 (α‐syn‐p129). In contrast to findings in AD, ACE activity and ET‐1 level were unchanged in DLB compared with controls. VEGF and FVIIIRA levels were, however, significantly lower in DLB. VEGF correlated positively with MAG concentration (in keeping with a relationship between reduction in VEGF and hypoperfusion), and negatively with α‐syn and α‐syn‐p129 levels. Both α‐syn and α‐syn‐p129 levels increased in human SH‐SY5Y neuroblastoma cells after oxygen‐glucose deprivation (OGD), and VEGF level was reduced in SH‐SY5Y cells overexpressing α‐syn. Taken together, our findings suggest that reduced microvessel density rather than vasoconstriction is responsible for lower occipital blood flow in DLB, and that the loss of microvessels may result from VEGF deficiency, possible secondary to the accumulation of α‐syn.

Introduction

Dementia with Lewy bodies (DLB) accounts for approximately 15%–20% of cases with dementia 23, 60. Pathological hallmarks of DLB include abnormal aggregates of α‐synuclein (α‐syn) within neuronal perikarya (Lewy bodies) and neurites (Lewy neurites). The predominant modification of α‐syn in Lewy bodies and neurites is phosphorylation of a single amino acid, Ser129 1, which promotes α‐syn aggregation and neurotoxicity 15, 42. DLB is characterized clinically by cognitive impairment (often of fluctuating severity), parkinsonian motor features and, in many patients, recurrent visual hallucinations 13. Cerebral hypoperfusion, or reduced cerebral blood flow (CBF), is a common feature of DLB and Alzheimer's disease (AD). Temporoparietal hypoperfusion can be demonstrated in both DLB and AD 27, 33 whereas occipital or parieto‐occipital cortex (BA17/18 and BA19) hypoperfusion is relatively specific for DLB 16, 20, 21, 33 and supports that clinical diagnosis 35.

Reduced CBF is not simply a concomitant of dementia. There is increasing evidence that it contributes to the development and progression of dementia. In a large study of CBF velocity in people without clinical dementia (the Rotterdam Study, 1730 participants aged 55 years and older), those who had greater CBF velocity at the outset were less likely subsequently to develop dementia 50. The interrelationship between CBF and development of dementia has been studied much more extensively in AD than DLB. Most cardiovascular risk factors (such as diabetes, hypertension, high cholesterol levels, atherosclerosis and obesity) are also risk factors for AD, emphasizing the interrelationship between vascular and neurodegenerative pathology and dementia 18, 19. In AD, hypoperfusion anticipates the dementia 9, 12, 14 and correlates inversely with cognitive performance. Mice overexpressing mutant human Aβ precursor protein (APP) show reduction in CBF well before the development of pathological or neurological abnormalities 39. Hypoperfusion is thought to impact on the disease through upregulation of APP and β‐secretase and increased production of Aβ. Conversely, the disease also influences CBF: Aβ peptides cause and enhance cerebral vasoconstriction 25, 40 which we have shown to be at least partly mediated through increased production of endothelin‐1 (ET‐1) and increased angiotensin‐converting enzyme (ACE) activity 37, 38, 43, 44, 45.

Little is known about the pathophysiology of reduced CBF in DLB. In this study, we have used autopsy tissue to examine possible mechanisms of hypoperfusion in the occipital cortex in a well‐characterized cohort of DLB cases. We did not observe an increase in ET‐1 or elevated production of angiotensin II (as measured by ACE activity), two potent vasoconstrictors, both of which are upregulated in the cerebral cortex in AD. However, we did demonstrate a significant reduction in vascular endothelial growth factor (VEGF), which plays a critical role in the formation and maintenance of blood vessels. The decline in VEGF was associated with a reduction in the level of the endothelial marker factor VIII‐related antigen (FVIIIRA), which we reported previously to correlate closely with microvessel density in the white matter 5 and shown here also to correlate with microvessel density in the contralateral hemisphere (as measured by computer‐assisted morphometry of FVIIIRA‐immunolabeled vessels in paraffin sections). The decline in VEGF in DLB correlated with a reduction in the ischemia‐sensitive protein myelin‐associated glycoprotein (MAG), a marker of ante‐mortem perfusion 6, and with the severity of Lewy body pathology in the affected cortex, as indicated by the levels of soluble and insoluble α‐syn. These data point to a distinct mechanism of cerebral hypoperfusion in DLB, reflecting loss of microvessels secondary to a decline in VEGF.

Methods

Study cohort

Brain tissue was obtained from the South West Dementia Brain Bank, University of Bristol. The study had local Research Ethics Committee approval. The brains had been separated midsagittally—the left hemisphere was sliced and frozen at −80°C and the right hemisphere fixed in formalin for paraffin histology and further detailed neuropathological examination. We used tissue from 12 patients with DLB (age at death − mean = 76.8 years ± SD 9.3, post‐mortem delay − mean = 27.5 h ± SD 10.3, five females and seven males) and 15 matched controls (age at death − mean = 81.3 years ± SD 9.3, Post‐mortem delay − mean = 34.3 h ± SD 17.5, three females and 12 males). Dementia had been assessed clinically according to Diagnostic and Statistical Manual for Mental Disorders, American Psychiatric Association (DSM) IV criteria, and DLB diagnosed according to consensus neuropathological criteria 34. All DLB cases had Braak stage 5 or 6 Lewy body pathology 11. Controls had no history of dementia and no neuropathological abnormalities apart from mild neurofibrillary tangle pathology (Braak tangle stage III or less) 10 and scattered diffuse Aβ plaques in some cases.

Tissue preparation

Approximately 200 mg of frozen occipital cortex (Brodmann area 17/18) was dissected and homogenized in sodium dodecyl sulfate (SDS) lysis buffer (5 M NaCl, 1 M Tris‐HCl pH 7.6, 1 mM phenylmethylsulphonyl fluoride, 1.7 mg/mL aprotinin and 10% SDS) in a mechanical Precellys 24 homogenizer (Stretton Scientific, Derbyshire, UK) with 2.3 mm ceramic beads (Biopic, Glasgow, UK). The homogenates were spun at 13 000 g for 15 minutes at 4°C and the supernatants removed and stored at −80°C. Total protein concentration was determined using Total Protein Kit (Sigma‐Aldrich, Dorset, UK) following the manufacturer's guidelines.

Immunocapture‐based fluorogenic ACE activity assay

Nunc maxisorp 96‐well plates (Fisher Scientific, Loughborough, UK) were coated with a monoclonal anti‐human ACE antibody (1 μg/mL; R&D Systems, Abingdon, UK) and left overnight at room temperature (RT). The plates were washed five times in PBS/0.01% tween‐20, blocked in 1% bovine serum albumin (BSA)/phosphate‐buffered saline (PBS) for 2 h at RT and washed again. Tissue homogenates diluted to 1 mg/mL total protein were added in duplicate to the wells and incubated for 2 h at RT with constant shaking. After another wash, the ACE inhibitor captopril (1 μM) (Enzo Life Sciences, Exeter, UK) was added to one of the duplicate sets of sample wells for 10 minutes and distilled water to the other, after which fluorogenic peptide substrate (10 μM) (ES005, R&D Systems, UK) diluted in activity assay buffer (100 mM Tris‐HCl pH 7.5, 50 mM NaCl and 10 μM ZnCl₂) was added and the plate was incubated for 3 h at 37°C in the dark. Fluorescence with excitation at 320 nm and emission at 405 nm was read at 3 h in a FLUOstar Optima plate reader. To (BMG LABTECH Ltd, Aylesbuty, UK) determine ACE‐specific enzyme activity, we subtracted the fluorescent signal after inhibition by captopril from that in the paired uninhibited wells. Each assay was repeated twice and the mean determined. ACE activity was interpolated from a standard curve produced by serial dilutions (39 pg/mL to 2500 pg/mL) of recombinant human ACE (R&D Systems, UK).

ET‐1 sandwich enzyme‐linked immunosorbent assay (ELISA)

ET‐1 protein level was measured using the QuantiGlo Chemiluminescent sandwich ELISA kit for human ET‐1 (R&D Systems, Minneapolis, MN, USA) following the manufacturer's guidelines, as described previously 44. Samples containing 1 mg/mL total protein were added in duplicate and incubated for 1.5 h at RT with agitation. Luminescence was read in a FLUOstar Optima plate reader. ET‐1 concentration was interpolated from a standard curve generated by serial dilutions of recombinant human ET‐1 (0.34 pg/mL to 250 pg/mL).

VEGF sandwich ELISA

VEGF protein level was measured using the Human VEGF Quantikine ELISA kit (R&D Systems, USA) according to the manufacturer's guidelines. The ELISA used a monoclonal mouse VEGF antibody as a capture antibody and a polyclonal biotinylated VEGF detection antibody. Tissue samples (1.0 mg/mL total protein) were added in duplicate and incubated for 2 h at RT with constant shaking. VEGF concentration was interpolated from a standard curve generated by a serial dilution of human recombinant VEGF (31.25 pg/mL to 2000 pg/mL). Absorbance was read at 450 nm in a FLUOstar Optima plate reader.

In pilot studies, we confirmed the post‐mortem stability of VEGF, by comparing measurements made in occipital cortex from brains (n = 3) with relatively short post‐mortem delays (6–10 h). We incubated the samples for up to 72 h at either RT or 4°C and found no correlation between incubation time and protein levels (data not shown) indicating stability of the protein up to 72 h.

FVIIIRA dot blot

FVIIIRA level was determined by dot blot analysis using a polyclonal rabbit anti‐human von Willebrand factor antibody, as previously described 2. Samples were diluted in Tris‐buffered saline (TBS) (1 in 800) and blotted onto nitrocellulose membrane (GE Healthcare, St. Giles, UK) for 1 h at RT. The membrane was blocked in 5% non‐fat dried milk protein (NFDMP) in TBS at 4°C overnight, washed, and then incubated for 1 h with polyclonal rabbit anti‐human von Willebrand factor, (0.3 μg/mL) (Dako, Glostrup, Denmark) at RT with agitation. After washing, the membrane was incubated with anti‐rabbit peroxidase‐conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA) in 5% NFDMP diluted in 0.3% TBS‐T for 1 h at RT with agitation. The membrane was washed and then developed using chemiluminescent ECL substrate (Millipore, Billerica, MA, USA) according to the manufacturer's guidelines. Image‐J was used to measure the integrated density of each sample. Serial dilutions of a standard reference brain tissue homogenate were used to adjust for any blot‐to‐blot variation. We previously demonstrated that FVIIIRA is stable under conditions of simulated post‐mortem delay for up to 72 h at 4°C or RT 2.

Validation of FVIIIRA dot blot for assessment of microvessel density

Dot blot measurements of FVIIIRA were compared with vessel density assessed in paraffin sections from the contralateral occipital cortex by computer‐assisted morphometry. Paraffin sections (7 μm in thickness) were labeled with polyclonal rabbit anti‐human von Willebrand factor (1.5 μg/mL) by a standard streptavidin‐biotin immunoperoxidase method, as previously described 2.

Measurements were made with the help of Image‐Pro Plus (Media Cybernetics, Marlow, UK), on randomly selected images captured under a ×20 objective on a Leica DM microscope (Leica Microsystems, Newcastle upon Tyne, UK) from 19 fields within the occipital cortex. An interactive measurement tool was used to trace along the FVIIIRA‐immunopositive microvessels within each field. The capillary density was expressed as the cumulative length of all microvessels divided by the area of the field, and the mean value was determined for the 19 selected fields in each section.

FIIIRA level, as measured by dot blot, was compared with vessel density, determined as described above, in a subset of 13 cases (six controls and seven DLB). In the seven DLB cases, we looked at the relationship between VEGF and vessel density.

MAG and myelin proteolipid protein (PLP) direct ELISA

MAG and PLP levels in tissue homogenates were determined by direct ELISA as previously described 6. For measurement of MAG, brain tissue homogenates were diluted 1:10 in PBS, added in duplicate to clear 96‐well microplates (Fisher Scientific, Loughborough, UK), left for 2 h at RT with constant shaking, washed ×5 in PBS/0.01% tween‐20, blocked in 1% BSA/PBS for 2 h at RT and washed again. Mouse monoclonal anti‐MAG (Abcam, Cambridge, UK) diluted to 0.5 μg/mL was added for 2 h at RT. The plates were again washed and incubated for 20 minutes at RT with biotinylated anti‐mouse secondary antibody (Vector Labs, Peterborough, UK) diluted 1:500. After another wash, the wells were incubated with streptavidin‐horseradish peroxidase (HRP) (1:500) (R&D Systems, UK). Chromogenic substrate (TMBS substrate, R&D Systems, UK) was added in the dark for 30 minutes and the reaction then stopped with 2N sulphuric acid. The absorbance was read at 450 nM in a FLUOstar Optima plate reader. MAG concentration was interpolated from a standard curve of serially diluted (6.25 ng/mL to 400 ng/mL) recombinant human MAG (Abnova, Taipei City, Taiwan). We previously demonstrated that MAG is stable under conditions of simulated post‐mortem delay for up to 72 h at 4°C or RT 6.

Total α‐syn and α‐syn phosphorylated at serine 129 (α‐syn‐p129) sandwich ELISAs

Brain tissue homogenates (200 mg) were sequentially extracted in 1% NP‐40 buffer (140 mM NaCl, 3 mM KCl, 25 mM Tris ph 7.4, 5 mM EDTA, 2 mM 1,10‐phenanthroline, 0.1 M PMSF, 1.7 mg/mL aprotinin, NP‐40 detergent, 100 mL dH₂O) (soluble fraction) and then after centrifugation the pellet was solubilized in 6 M guanidine : HCl (insoluble fraction), as previously described for Aβ measurements in post‐mortem human brain tissue 7, 56, 57, 58.

Total α‐syn level was determined by sandwich ELISA. Mouse monoclonal anti‐α‐synuclein antibody (0.5 μg/mL) (BD Biosciences, Oxford, UK) was coated onto a NUNC Maxisorp 96‐well plate overnight at RT. The plate was washed in PBS/0.01% tween‐20 and blocked for 1.5 h in 1% BSA/PBS. Duplicate tissue samples diluted in PBS were incubated for 2 h at RT with constant shaking. The plates were washed and incubated for 2 h at RT with biotinylated polyclonal anti‐α‐syn (1 μg/mL) (R&D Systems, UK). After further washes, streptavidin‐HRP (1:200) (R&D Systems, UK) was added for 20 minutes. Chromogenic substrate (TMBS substrate, R&D Systems, UK) was added in the dark for 30 minutes and the reaction then stopped with 2N sulphuric acid. The absorbance was read at 450 nM in a FLUOstar Optima plate reader. Total α‐synuclein level was interpolated from measurements of serial dilutions (0.97 ng/mL to 62.5 ng/mL) of recombinant human α‐synuclein (rPeptide, Stratech, Suffolk, UK).

The level of α‐syn‐p129 was also determined by sandwich ELISA. Mouse monoclonal anti‐α‐syn antibody (BD Biosciences) at 0.5 μg/mL was coated onto NUNC Maxisorp 96‐well plates overnight at RT. The plates were washed in PBS/0.01% tween‐20 and blocked for 1.5 h in 1% BSA/PBS. Duplicate tissue homogenates diluted 1:99 in PBS were added for 5 h at RT with constant shaking. The plates were washed, biotinylated anti‐α‐syn‐p129 (0.8 μg/mL) (Abcam) was added and left at 4°C overnight. After further washes, biotinylated horse anti‐rabbit antibody (Vector Laboratories) diluted 1:1000 in PBS with 0.01% tween‐20 was added for 1 h at RT. After further washes, streptavidin‐HRP (1:200) (R&D Systems, UK) was added for 20 minutes. Chromogenic substrate (TMBS substrate, R&D Systems) was added in the dark for 30 minutes and the reaction then stopped with 2N sulphuric acid. The level of α‐syn‐p129 was interpolated from measurements of serial dilutions (3.25 ng/mL to 200 ng/mL) of recombinant α‐syn that had been phosphorylated at ser129 by incubation with casein kinase II (750 units, New England Biolabs, Hitchin, UK, one unit being defined as the amount of CKII required to catalyze the transfer of 1 ρmol of phosphate to 100 μM CKII peptide sequence RRRADSDDDDD in 1 minute at 30°C) for 1 h at 30°C in the presence of 200 μM ATP (New England Biolabs).

SH‐SY5Y cell culture and oxygen‐glucose deprivation (OGD)

SH‐SY5Y cells and SH‐SY5Y cells stably transfected with full‐length human wild‐type α‐syn were routinely maintained in 42% vol/vol Ham's F12 nutrient mixture (F12) (Sigma‐Aldrich) and 42% vol/vol Eagle's minimum essential medium (MEM) (Sigma‐Aldrich), supplemented with 15% vol/vol fetal calf serum (FCS) (Sigma‐Aldrich), 2 mM l‐glutamine (Sigma‐Aldrich), 1% vol/vol non‐essential amino acids (NEAA) solution (Sigma‐Aldrich), 20 units/mL penicillin, 20 mg/mL streptomycin (Sigma‐Aldrich) and 250 ng/mL amphotericin B (Life Technologies, Paisley, UK) at 37°C in 5% CO₂ (21% O₂). SH‐SY5Y cells stably transfected with full‐length human wild‐type α‐syn were cultured in the presence of the antibiotic G418 (0.3 mg/mL) (Geneticin, Life Technologies).

For OGD experiments, the culture medium was exchanged for Dulbecco's modified Eagle's medium (DMEM) containing either high glucose (4500 mg/L) (Sigma‐Aldrich) or DMEM without glucose, both supplemented with L‐glutamine (0.584 g/L) and sodium pyruvate (Sigma‐Aldrich), in the absence of fetal bovine serum. The cells were maintained in DMEM containing either high glucose (4500 mg/L) or no glucose, under normoxic (5% CO₂/95% air at 37°C) or hypoxic (94% N₂, 5% CO₂ and 1% O₂) conditions for 6 h at 37°C. A subchamber culture system with a ProOx 110 oxygen controller (Biospherix, Lacona, NY, USA) was used to maintain cultured cells under hypoxic conditions at 37°C. The cells were washed in PBS and lysed in a proprietary cell lysis buffer (Sigma‐Aldrich) following manufacturer's guidelines and spun at 13 000 rpm for 15 minutes at 4°C. Cell supernatants were stored at −80°C until used.

In a separate experiment, SH‐SY5Y cells stably overexpressing human wild‐type α‐syn were grown on poly‐L‐lysine coated coverslips and exposed to the above experimental conditions for 6 h. Cells were fixed in 4% paraformaldehyde for 10 minutes, washed in PBS and blocked for 30 minutes in PBS with 0.01% triton‐100 and 5% horse serum (Millipore). The cells were incubated with anti‐α‐syn‐p‐129 (5 μg/mL) (Abcam) for 1 h at RT, washed in PBS, and incubated with donkey anti‐rabbit‐alexa‐fluor‐555 (1 in 1000) (Molecular Probes, Life Technologies) for 1 h in the dark. Immunofluorescent images were acquired using a confocal laser scanning microscope (magnification ×60).

Statistical analysis

The data were analyzed by unpaired two‐tailed t‐test, ANOVA with Bonferroni post hoc analysis and Pearson's correlation, with the help of SPSS version 16 (SPSS, Chicago, IL, USA). P‐values < 0.05 were considered statistically significant.

Results

ACE and ET‐1

We measured ACE activity and ET‐1 level to determine whether reduced blood flow might be caused by vasoconstriction resulting from increased ET‐1 or elevated angiotensin II production (both of which can be demonstrated in AD). ACE activity (Figure 1A) in the occipital cortex did not differ significantly between DLB and control brains (P = 0.14). ET‐1 level tended to be lower in DLB than controls (Figure 1B) but the reduction did not reach statistical significance (P = 0.08).

Figure 1.

The level of (A) ACE activity and (B) ET‐1 did not differ significantly between DLB and control brains. There was a trend toward a decline in ET‐1 in DLB (P = 0.08) and a non‐significant increase in ACE activity (P = 0.14).

VEGF, FVIIIRA and MAG

To explore the potential mechanism of reduced CBF in occipital cortex in DLB, we measured the levels of VEGF, FVIIIRA (to indicate microvessel density) 5 and MAG (a myelin protein which is particularly sensitive to hypoperfusion) 6. We had expected VEGF level to be increased in the occipital cortex in DLB as a reaction to hypoperfusion. However, VEFG level was significantly reduced in the occipital cortex in DLB (P = 0.03) (Figure 2A), but did not change in the frontal, cingulate or parahippocampal cortex (data not shown), raising the possibility of increased rather than decreased perfusion of the occipital cortex. To investigate this further, we examined the level of FVIIIRA, as a marker of the number of endothelial cells and hence of microvessels. The level of FVIIIRA was significantly lower in DLB than in control occipital cortex (Figure 2B) (P = 0.04) but did not differ significantly between DLB and control tissue in the frontal cortex or thalamus (data not shown). To establish whether or not the level of FVIIIRA was a good indicator of microvascular density, we compared FVIIIRA level in cortex from 13 brains with the capillary length/section area in paraffin sections of contralateral occipital cortex from the same brains. The two sets of measurements showed a direct correlation (Figure 2C) (r = 0.633, P = 0.02). Comparison of capillary length/section area with the concentration of VEGF in the contralateral occipital cortex of seven DLB cases showed these measurements also to be directly correlated (Figure 3) (r = 0.898, P = 0.006), in keeping with the hypothesis that the loss of vessel density in DLB resulted from the reduction in VEGF. Finally, the concentration of VEGF in the occipital cortex showed a direct correlation with the concentration of MAG (r = 0.656, P = 0.02) (Figure 4), a protein that is highly susceptible to ischemia and the level of which therefore gives an indication of the adequacy of ante‐mortem perfusion 28.

Figure 2.

A . VEGF level, measured by sandwich ELISA, was significantly reduced in the occipital cortex in DLB. B . FVIIIRA level was also reduced in DLB compared with controls. C . The level of FVIIIRA correlated closely with vessel length/section area in paraffin sections from the contralateral cerebral hemisphere in a subset of cases assessed by morphometry. *P < 0.05.

Figure 3.

Vessel length/section area, assessed in paraffin sections from the contralateral cerebral hemisphere by morphometry, correlated positively with VEGF expression, as measured by sandwich ELISA.

Figure 4.

The level of VEGF, which was reduced in DLB, correlated positively with the concentration of myelin‐associated glycoprotein (MAG), an adaxonal myelin protein that is particularly susceptible to hypoperfusion in human brain tissue.

α‐synuclein, VEGF and MAG

To assess the relationship between reduced occipital blood flow and Lewy body pathology, we looked at the relationship between the levels of VEGF and MAG, and those of α‐syn. The levels of soluble α‐syn (P < 0.01) (Figure 5A) and insoluble α‐syn (P < 0.01) (Figure 5B) were, as expected, higher in the occipital lobe in DLB than controls. VEGF concentration correlated inversely with both soluble (r = −0.526, P = 0.007) and insoluble α‐syn (r = −0.454, P = 0.002) (Figure 5C,D), that is increased α‐syn is associated with reduced VEGF. MAG level also correlated inversely with soluble α‐syn (r = −0.40, P = 0.039) and insoluble α‐syn (r = 0.474, P = 0.014) (Figure 5E,F), suggesting that cortical perfusion falls as the concentration of α‐syn rises.

Figure 5.

The level of α‐syn in both soluble (A) and insoluble (B) extracts was significantly increased in the occipital cortex in DLB. Levels of VEGF (C,D) and MAG (E,F) correlated inversely with soluble and insoluble α‐syn level. **P < 0.01.

OGD increases neuronal α‐syn and α‐syn‐p129

We next modeled ischemia, as well as selective oxygen or glucose deprivation (GD), in SH‐SY5Y cells that had been transfected to overexpressing human α‐syn. The cells were cultured in the absence or presence of glucose, and under normoxic (5% CO₂/95% air at 37°C) or hypoxic (94% N₂, 5% CO₂, and 1% O₂) conditions for 6 h. α‐syn level tended to be higher in cells cultured in DMEM without glucose under normoxic or hypoxic conditions than in cells grown in medium with glucose under normal or hypoxic conditions, although the differences did not reach statistical significance (Figure 6A). The level of α‐syn‐p129 was significantly elevated in SH‐SY5Y cells cultured in the absence of glucose, under either normoxic or hypoxic conditions (Figure 6B) (P < 0.05 and P < 0.05, respectively), but did not change under conditions of hypoxia with normal glucose. Immunofluorescence confirmed an obvious increase in α‐syn‐p129 in SH‐SY5Y cells under conditions of GD and OGD (Figure 6C).

Figure 6.

The levels of (A) α‐syn and (B) α‐syn‐p129 were increased in lysates of SH‐SY5Y cells overexpressing wild‐type α‐syn, when cultured for 6 h under conditions of glucose deprivation (GD) or oxygen‐glucose deprivation in comparison with the levels in cells grown under normal control (C) or hypoxic conditions with normal glucose (H) but only the increase in α‐syn‐p129 reached statistical significance (*P < 0.05). (C) Under the same experimental conditions, there was increased α‐syn‐p129 immunofluorescent labeling (red) in SH‐SY5Y cells cultured in the absence of glucose under both normoxic and hypoxic (oxygen‐glucose deprivation) conditions. The cell nuclei were stained with DAPI (blue). The confocal microscope images were acquired under a ×60 objective.

Neuronal production of VEGF is negatively modulated by α‐syn

To examine the possible association between α‐syn and VEGF expression, we cultured four different human SH‐SY5Y cell lines transfected with full‐length wild‐type α‐syn. The four cell lines varied in the level of overproduction of α‐syn (Figure 7A). A significant negative correlation between α‐syn levels and VEGF expression was observed within the non‐transfected and four transfected cell lines (r = 0.96, P = 0.008) (Figure 7B).

Figure 7.

A . Level of α‐syn (as measured by sandwich ELISA in cell lystates) in SH‐SY5Y wild‐type cells and SH‐SY5Y cells transfected with human wild‐type full‐length α‐syn (four transfected cell lines were studied: WT1, WT2, WT4 and WT7). B . The level of VEGF (measured by a sandwich ELISA) correlated inversely with that of α‐syn.

Discussion

Hypoperfusion of cerebral cortex in the parieto‐occipital region is a characteristic feature of DLB. Our study examined the potential mechanism which underlie reduced CBF in DLB and found that (i) the levels of ET‐1 and ACE, two potent vasoconstrictors that are upregulated in the cerebral cortex in AD, were unchanged in DLB (ii); VEGF, a potent pro‐angiogenic factor 41, 64, is reduced in DLB (iii); FVIIIRA level, a good surrogate marker of capilliary density, is reduced in DLB (iv); the reduction in VEGF and FVIIIRA is associated with ischemic damage (as evidenced by of the decline in MAG) and increased accumulation of α‐syn. We modeled ischemia and energy deprivation in neuronal SH‐SY5Y cells and showed that both GD and combined OGD elevated the production of α‐syn and α‐syn‐p129. We also showed that VEGF level is reduced in SH‐SY5Y cells overexpressing α‐syn. Together, these data indicate that reduced vascularity as a result of deficient VEGF expression is likely to contribute to cerebral hypoperfusion in DLB, that accumulation of α‐syn and α‐syn‐p129 is likely to contribute to deficient VEGF expression, and that cerebral hypoperfusion is likely to contribute to the accumulation of α‐syn and α‐syn‐p129.

Cerebral hypoperfusion is a risk factor for numerous neurodegenerative diseases including AD 17, 29 DLB 16, 33 and Parkinson's disease (PD) 20. Reduced CBF is a risk factor for dementia 50. Several studies showing a clear relationship between cardiovascular risk factors and dementia 18, 19. The evidence is strongest for AD but this probably reflects a lack of investigation in other neurodegenerative diseases. The mechanism and potential contribution of cerebral hypoperfusion to neurodegeneration and dementia in DLB is largely unexplored. In AD, hypoperfusion anticipates the dementia 9, 12, 14 and correlates inversely with cognitive performance 26. Mice overexpressing mutant APP show a reduction in CBF well before the development of pathological or neurological abnormalities 39. Cerebral hypofusion is thought to contribute specifically to the underlying disease processes in AD, through upregulation of APP and β‐secretase and increased production of Aβ 22, 31, 63, 65. Aβ peptides also cause cerebral vasoconstriction 25, 40 which we have shown to be at least partly attributable to increased production of the vasoconstrictor ET‐1 and of ACE (which catalyzes the production of another vasoconstrictor, Ang II) 36, 37, 38, 43, 44, 45.

We have explored the potential mechanism underlying cerebral hypoperfusion in DLB. One of our main findings was a reduction in VEGF in DLB in the occipital cortex. VEGF is a pro‐angiogenic 41, 63 neurotrophic factor 49, 51, 52 and is elevated in AD 30, 54. VEGF expression is regulated by HIF1A 47. The reduction in VEGF expression in DLB was disease‐specific: a similar reduction was not observed in the occipital cortex of a subset of AD cases (data not shown). It was also region‐specific: VEGF level was not reduced in DLB in midfrontal cortex, cingulate gyrus or parahippocampal cortex (data not shown), regions less affected by cerebral hypoperfusion. VEGF declines with age 48 but is clearly further reduced in DLB. Capillary vessel density was also reduced in DLB, a reduction that was disease‐ and region‐specific, and correlated with VEGF. The pathogenesis of reduced blood flow differs from that in AD, in which ACE and ET‐1 are upregulated and vasoconstriction seems to play a major role: in response to Aβ and mediate thought to impair CBF. In contrast, ET‐1 level and ACE activity were unchanged in DLB.

To our knowledge, this is the first report of vascular abnormalities in the cerebral cortex in DLB. The findings in the cortex differ from those reported on post‐mortem examination of the substantia nigra in human PD 59 and a primate model of PD 4, both of which showed increased vascularity and an increase in the number of cells that could be immunolabeled with antibody to VEGF. While the apparent discrepancy could conceivably reflect differences between the pathogenesis of PD and DLB, it is noteworthy that these studies focused on end‐stage substantia nigra, a region characterized by marked neuronal loss, microglial inflammation and reactive astrocytosis. In contrast, the occipital cortex is relatively well preserved in DLB. Furthermore, the previous studies relied on semi‐quantitative immunohistochemistry to assess VEGF expression and blood vessel density in the substantia nigra. Indeed, Wada et al 59 found the level of VEGF in the striatum to be unchanged when measured by ELISA; the comparative immunohistochemical data were not shown. One limitation of our study was the inability of the ELISA to discriminate between the different forms of VEGF (VEGF‐A, ‐B, ‐C and ‐D), although VEGF‐A is most abundant within the brain 27.

We found the reduction in VEGF in DLB to correlate with the increase in α‐syn and with a reduction in MAG, a myelin protein that is particularly sensitive to ischemia. We previously demonstrated that chronic ante‐mortem hypoperfusion (eg resulting from cerebrovascular disease) causes a significant reduction in the concentration of MAG in human brain tissue 6. To explore the interrelationship between cerebral hypoperfusion, VEGF production and Lewy body pathology, we used SH‐SY5Y cells that overexpress α‐syn to model Lewy body disease 24 and either GD or OGD to model the effects of an energy deficit or ischemia on neurons. Both GD and OGD (but not hypoxia alone) resulted in elevated levels of α‐syn and α‐syn‐p129. These findings are consistent with previous in vivo and in vitro studies. Impairment of cell metabolism, particularly GD, was shown to increase α‐syn accumulation and aggregation in SH‐SY5Y cells 8, and transient focal ischemia increased α‐syn immunoreactivity in the gerbil hippocampus 28, 61 and α‐syn level in mouse cerebral cortex 62. Transient ischemia also promoted oligomerization and fibrillation of α‐syn in [A30P]‐α‐syn transgenic rats 55. Acidosis and free‐radical stress (associated with OGD) were reported to induce α‐syn Ser129 phosphorylation and aggregation in SH‐SY5Y cells 32, 46, 53, 55, 62. In further experiments designed to examine the relationship between VEGF and α‐syn production, we showed that overexpression of α‐syn in SH‐SY5Y cells caused a decline in VEGF, the extent of decline correlating with the level of expression of α‐syn. This finding too conforms to a previous report of a twofold reduction in VEGF mRNA in neurons overexpressing α‐syn 3. These findings raise the possibility of a vicious cycle in accumulation of α‐syn leads to cerebral hypoperfusion, which promotes α‐syn production and Ser129 phophorylation and accelerates the progression of disease.