Increased expression of ephrin A1 in brain arteriovenous malformation: DNA microarray analysis
- 1k Downloads
A number of previous studies have revealed the abnormal expression of various angiogenesis-related genes or products in brain arteriovenous malformation (AVM). To understand the molecular process of this disease, we analyzed gene expression profiles in brain AVM. Using a DNA microarray consisting of 17,086 genes, we identified differentially expressed genes in 5 brain AVMs from their draining veins, vessels retaining basic venous architecture. Not many genes were differentially expressed between the AVM nidus and the draining vein. When we applied an absolute cut-off value for normalized log2 (cy5/cy3 ratio) of 0.4, 19 genes were selected. Genes such as SOX8, TRIM2, FENA1 (ephrin A1), and AQP4 were upregulated, and genes such as I_1000105, KRT18, IGFBP7, EMILIN-2, and KRT14 were downregulated. Genes relating to angiogenesis, such as vascular endothelial growth factor and angiopoietin and other members of the ephrin family, were not differentiated. Among differentially expressed genes detected in this analysis, we focused on ephrin A1, a gene related to embryogenesis and angiogenesis. The expression of ephrin A1 was two and three to nine times higher than that of the draining vein and normal brain, respectively, using real-time reverse transcription-polymerase chain reaction. For the first time, here we report the increased expression of ephrin A1 in brain AVM, which may play an important role in the pathogenesis of AVM.
KeywordsAngiogenesis Brain arteriovenous malformation Ephrin A1 Gene expression Vein
Arteriovenous malformation (AVM) of the brain is defined as the direct communication between one or more arteries of the cerebral parenchyma and one or more draining veins without the intervention of a capillary bed [3, 15, 23]. Most brain AVM vessels are composed of large, thin-walled dilated channels displaying fibroblast proliferation and increased cellularity in the wall with focal deposition of elastic tissue layers. They are generally believed to arise when capillaries fail to develop in an area of the brain during early embryogenesis, resulting in abnormal communications between arteries and veins. The growth of blood vessels during embryogenesis and in adult organisms is tightly controlled, and this growth is thought to be mediated by soluble factors [13, 19, 27]. A number of previous studies revealed the abnormal expression of various angiogenesis-related genes or products in brain AVM [7, 8, 9, 10, 11, 14, 16, 27].
In order to understand the molecular process of this disease, we attempted to identify differentially expressed genes in brain AVMs as compared to draining veins, vessels retaining the basic venous architecture [3, 15]. We approached the problem by examining gene-expression profiles. DNA microarray technology nowadays makes it possible to simultaneously analyze the expression patterns of thousands of genes in a given tissue [12, 18, 22]. We here report the identification of ephrin A1, a gene involved in embryogenesis and angiogenesis, which is overexpressed in the nidus of brain AVM.
Materials and methods
Samples and clinical information
Clinical features of five patients with brain arteriovenous malformation
Rt posterior temporal
Rt posterior temporal
Lt anterior temporal
Frozen sections were placed directly in Trizol reagent (Invitrogen), homogenized for 2 min on ice, and RNA was isolated according to the manufacturer’s protocol. RNA was quantified based on absorbance at 260 nm. Total RNA was assessed for quality with an Agilent Lab-on-a-Chip 2100 bioanalyzer. Possible traces of genomic DNA contaminating the RNA preparations were removed by DNase I (Promega) digestion.
Labeling and hybridization
The targets for DNA microarray analysis were prepared according to the manufacturer’s instructions. Briefly, cyanine 3- and cyanine 5-labeled complementary RNA (cRNA) was synthesized followed by the production of double-stranded cDNA from 50–500 ng of total RNA with a Human 1A oligoDNA microarray kit (Agilent). After the purification of labeled cRNA, arrays containing 17,086 genes were hybridized with the fragmented cRNA targets at 60°C for 17 h, and then washed and dried with air. We repeated the experiments using the same sample set, but substituting dyes.
Scans were performed using an Agilent DNA microarray scanner (model G2565BA), and the expression value for each gene was calculated by Feature Extraction software. Data were normalized by the LOWESS normalization method. Only the genes deemed present on all slides were further considered. When both cyanine5 and cyanine3 dyes produced signal intensities lower than the cut-off [2.6× standard deviation (SD) of background], we excluded these genes from further analysis. We excluded a sample set when the correlation coefficient for each pair of color swapping was positive. We calculated the mean and SD of log2 (r1 × r2, where r1 and r2 are cy5/cy3 ratios for color swapping) for each pair of sample sets in the entire array, and eliminated pairs of elements whose log2 (r1 × r2) was greater than the SD from the mean. We selected genes whose absolute value for normalized log2 (cy5/cy3 ratio) was higher than the cut-off value determined in all sample sets examined [1, 25].
Quantitative real time RT-PCR
To evaluate the expression level of the target genes, quantitative RT-PCR was performed using a Sequence Detection System, model 7700 (Perkin Elmer). Quantitative real-time RT-PCR primers were designed using Primer 3. A sense primer and an antisense primer were synthesized using sequences from a database search at NCBI for human ephrin A1 and GAPDH as follows: ephrin A1 sense primer: TGGAACAGTTCAAATCCCAAG; ephrin A1 antisense primer: CTCATGCTCCACCAGGTACA; GAPDH sense primer: CAGCCTCAAGATCATCAGCA; GAPDH antisense primer: ATCCACAGTCTTCTGGGTGG. Five serial dilutions of each total RNA sample (100, 50, 25, 12.5, and 6.25 ng total RNA) were analyzed for each target gene. Amplification reactions (25 μl) comprised the total RNA samples: 2× SYBR green PCR master mix, 40× MultiScribe reverse transcriptase and RNase inhibitor mix, and 200 nmol/l of each primer. The thermal cycling conditions were as follows: 30 min at 48°C for reverse transcription, 10 min at 95°C for the AmpliTaq Gold activation, 40 cycles of 15 s at 95°C for denaturation, and 1 min at 60°C for annealing and extension. CT values corresponded to the cycle number at which the fluorescent emission monitored in real time reached the threshold, set at 10 SD above the mean of the baseline emission calculated from cycles 5 to 15. The CT values decreased linearly with increasing target quantity and could be used as a quantitative measurement of the input target number. Total RNA concentrations from each sample were normalized by the quantity of GAPDH mRNA, and the expression levels of the target genes were evaluated by the ratio of the number of target mRNA to GAPDH mRNA.
Immunohistochemistry was performed using paraffin-embedded tissue. Heat-induced epitope retrieval was performed in 10 mmol/l citrate buffer. After blocking, the slides were then incubated with a primary antibody. The slides were incubated with a Histofine simple stain MAX-PO(Multi) (Nichirei) for 30 min at room temperature. The slides were washed again and incubated with detecting reagent. The slides were counterstained with hematoxylin. Primary antibodies were polyclonal rabbit anti-human ephrin A1 antibody (Zymed Laboratories). The sensitivity and specificity of primary antibodies were verified using a positive control (prostate) and a negative control (no primary antibody).
Six genes upregulated in the nidus of brain arteriovenous malformation (AVM) over the draining vein
SRY (sex-determining region Y)-box 8
Protein with very strong similarity to neural activity-related RING finger (mouse Trim2), member of the B-box zinc-finger family
Member of the protein-phosphatase 2C family
Adenomatous polyposis coli-like
Thirteen genes downregulated in the nidus of brain arteirovenous malformation (AVM) over the draining vein
Protein of unknown function, has high similarity to uncharacterized mouse Dact1
Protein with high similarity to keratin 18 (human KRT18)
Keratin 18 (cytokeratin 18)
Protein with high similarity to adipocyte enhancer-binding protein 1 (mouse Aebp1)
Williams-Beuren syndrome chromosome region 5
Insulin-like growth factor binding protein 7
Extracellular glycoprotein EMILIN-2
Cis-retinol-androgen dehydrogenase 7
Low affinity IIIa receptor for Fc fragment of IgG, neutrophil-specific phosphatidylinositol-linked membrane protein
Member of the DUF590 protein of unknown function family
Follistatin-like 1 (follistatin-related protein)
Retired, was a member of the ubiquitin carboxyl-terminal hydrolase family 2
Signal-intensity ratios of house-keeping genes and genes previously reported to be related to the pathogenesis of brain arteiovenous malformation (AVM)
Vascular endothelial growth factor (VEGF)
Fms related tyrosine kinase 1 (vascular endothelial growth factor-vascular permeability factor receptor)
Tyrosine kinase with immunoglobulin and epidermal growth factor homology domains
Activin receptor-like kinase-1
Matrix metalloproteinase 9 (gelatinase B, type IV collagenase)
Hypoxia-inducible factor 1 alpha subunit
The results of the DNA microarray analysis indicate that not many genes are differentially expressed between the AVM nidus and draining vein. When we applied an absolute cut-off value for the normalized log2 (cy5/cy3 ratio) of 0.3, only 28 genes were selected; when we applied 0.4, 19 genes were selected. Genes such as SOX8, TRIM2, FENA1 (ephrin A1), and AQP4 related to brain were upregulated. Among the downregulated genes, several genes were related to the extracellular matrix and keratin such as I_1000105, KRT18, IGFBP7, EMILIN-2, and KRT14. We focused on ephrin A1, a gene related to embryogenesis and angiogenesis, among differentially expressed genes detected in this analysis. We confirmed the upregulation of ephrin A1 in the nidus compared to the draining vein and the normal brain tissue using real-time RT-PCR. Our immunohistochemical study showed that ephrin A1 was highly expressed in astrocytes and neurons of the perinidal parenchymal tissue, but not in the endothelial or smooth muscle cells of the AVM vessels. Cells in normal brain tissue were not stained by ephrin A1 antibody.
Ephrin A1 was originally identified as a tumor necrosis factor (TNF)-α-inducible gene in human umbilical vein endothelial cells and is expressed in the developing vasculature during embryogenesis [18, 22]. Ephrin A1 induces endothelial cell migration and capillary assembly in vitro and angiogenesis in the corneal pocket assay in vivo [4, 18]. These studies indicate that ephrin A1 is critical for normal blood vessel development and suggest that the molecule may also play a role in angiogenesis for pathological processes in adults. The increasing size of AVMs over time has been documented in the literature . Recurrent AVMs after negative angiograms have also been reported . It is well known that new shunts between arteries and veins grow after embolization of feeding arteries.
We compared our DNA microarray data of angiogenesis-related genes with previously published data. We found no differences in these genes between the AVM nidus and draining vein. The decreased expression of vascular endothelial growth factor receptor (VEGFR), angiopoietin 1 (Ang1) and tyrosin kinase with immunoglobulin and epidermal growth factor homology domains 2 (Tie2), together with an increased expression of Ang2, matrix metalloproteinase (MMP)-9, and tissue inhibitors of metalloproteinases (TIMPs) was reported [7, 8, 9]. Hashimoto et al. confirmed a general correlation between many of the previously published findings on abnormal angiogenesis-related protein expression in brain AVM and gene microarray analysis . Sure et al. recently indicated that in AVMs treated with embolization before surgery the expression of HIF1A and VEGF was significantly higher than in patients who did not have previous endovascular treatment . In our data set there was no significant difference of HIF1A and VEGF between AVM patients with and without embolization before surgery. The difference may be due to the different tissues used for control samples as comparisons to brain AVM. Previous studies all used normal cortex as a control, whereas we selected the draining vein as control, because most of the AVM nidus consists of vessels with a dilated venule-like structure, and the draining vein retains that basic venous architecture. Moreover, our samples are heterogenous in clinical features: the size of the nidus varies, two have a history of bleeding, embolization was performed in two. Gene expressions of the nidus and the vein may be affected. Obviously, it would have been better and more informative if the gene expression levels in the vessel and the brain component of the nidus, the normal vein and the tissue of the normal brain were compared individually.
Among the ephrin family, ephrin A1 was upregulated only in the nidus of brain AVM as compared to the draining vein. The ephrin A1 ligand is expressed predominantly in tumor cells, and ephrin A2 receptors are expressed primarily in a complementary fashion in tumor-associated vasculature . Treatment with soluble EphA2-Fc chimeric receptor results in decreased neovascularization in two different tumor types in vascular window assays, and EphA2-Fc of EphA3-Fc treatment results in decreased tumor vascular density and impaired tumor progression in vivo . Although the pathological state of brain AVM is completely different from tumor neovascularization, a similar approach may elucidate the pathogenesis of brain AVM and help in finding a new treatment for this disease.
Ephrin B2 is expressed in developing arterial endothelial cells, whereas ephrin B4 receptor has a complementary expression and is confined to venous endothelial cells. The ephrin B4 receptor protein and ephrin B2 are therefore considered to define the boundaries between arterial and venous endothelial cells [5, 6]. It was recently reported that endoglin and activin receptor-like kinase-1 (ACVRL1) are necessary for the maintenance of distinct arterial-venous vascular beds and that attenuation of the ACVRL1 signaling pathway is the precipitating event in the etiology of hereditary hemorrhagic telangiectasia, a disease associated with AVM and characterized by the inappropriate fusion of arterioles with venules [20, 24]. We did not detect differences in the expressions of these genes between the AVM nidus and draining vein.
In conclusion, we here report for the first time the increased expression of ephrin A1 in the nidus of brain AVM, which may play important roles in the development and maintenance of AVM.
This work was supported in part by a grant-in-aid for scientific research (C) from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (to H. K. and H.O.); the Japanese Ministry of Health, Labor, and Welfare (to H.K. and I.I.) and a Research for the Future Program Grant from the Japan Society for the Promotion of Science (I.I.). We thank Mr. Takashi Sakayori for his technical assistance.
- 9.Hashimoto T, Wen G, Lawton MT, Boudreau NJ, Bollen AW, Yang GY, Barbaro NM, Higashida RT, Dowd CF, Halbach VV, Young WL (2003) Abnormal expression of matrix metalloproteinases and tissue inhibitors of metalloproteinases in brain arteirovenous malformations. Stroke 34:925–931PubMedCrossRefGoogle Scholar
- 25.Yang IV, Chen E, Hasseman JP, Liang W, Frank BC, Wang S, Sharov V, Saeed AI, White J, Li J, Lee NH, Yeatman TJ, Quackenbush J (2002) Within the fold: assessing differential expression measures and reproducibility in microarray assays. Genome Biology 3:research 0062.1–0062.12Google Scholar
- 26.Yasargil MG (1987) Microneurosurgery. Vol IIA. Thieme, New York, pp 138–160Google Scholar