Microarray-based gene expression profiling and DNA copy number variation analysis of temporal fossa arachnoid cysts

Arachnoid cysts (AC) are relatively common benign lesions, reported to be found in 1.1% of the adult population [1]. Typical symptoms in adults are headache, dizziness, seizures [2] and dyscognition [3]. Although ACs can be found throughout the central nervous system (CNS), they show a marked predilection for the temporal fossa (figure 1) [2, 4‐6]. In the majority of cases, skull indentations and a large corresponding temporal fossa, as seen on MRI or CT, suggest that ACs arise before the neurocranium is fully developed. Anatomically, ACs are formed by a splitting of the arachnoid mater (AM) [7‐9]. The size of ACs is classified according to three Galassi types [10]. Briefly, a type I cyst is small, biconvex, and located at the anterior temporal pole. A type II cyst involves the proximal and intermediate segments of the Sylvian fissure, and a type III cyst involves the entire Sylvian fissure, and has often a marked radiological mass effect.

The cyst wall is composed of non-neoplastic arachnoid cells, and it has been suggested that these cells could secrete cerebrospinal fluid (CSF) [11, 12]. The cyst fluid has a chemical composition similar to that of CSF [13‐15]. Thus, the frequently seen stable size of ACs might represent a steady state where fluid influx equals the efflux. However, until a proteomic profiling comparing CSF and AC fluid is performed, it remains questionable how similar these fluids are.

Given the likely congenital nature of ACs, it is possible that altered gene expression in neural crest cells at the time of leptomeningeal development may contribute to cyst formation. However, so far there has been very little research on the molecular biology of AC. Hence, putative candidate genes with a role in AC formation are yet to be found. Several methods are available for studying gene expression. As mRNA microarray has the advantage that it can be used to analyze a large number of genes simultaneously, it has evolved to be the method of choice in screening projects of the transcriptome.

Previously, we have characterized the molecular signature of intracranial meningiomas [16]. As these tumors derive from arachnoid cap cells, and normal arachnoid in sufficient quantities is difficult to obtain, we used samples from arachnoid cysts as control tissue. Therefore, it would be interesting to determine whether the gene expression signature of normal arachnoid membrane differs from that of AC membranes. If these signatures were similar, it would imply that AC tissue could be used as a justified source of control tissue in molecular biological studies of conditions involving the arachnoid, such as subarachnoid haemorrhage, meningiomas, and meningitis.

In the current study we have performed a molecular characterization of temporal fossa AC membranes. The aims were: 1) to determine the general gene expression profile of AC samples compared with that of normal AM tissue. In this way we aimed at finding candidate genes with a role in AC development; 2) to search for DNA regions with altered copy number variations (CNV) in AC tissue. The study showed marked similarities in the gene expression profiles of the AC and AM samples; however, we identified a small subset of differentially expressed genes and DNA alterations in arachnoid cysts compared to normal tissue.

In mice it has been shown that mutations in the SHROOM3 gene result in exencephaly, acrania, facial clefting, and spina bifida; all presumably due to failed closure of the neural tube [21]. The authors suggested that SHROOM3 is involved in the regulation of cytoarchitecture that is required for proper neurulation. Another study, has found expression of SHROOM3 in several human tissues including the brain [22]. The uniform down-regulation of SHROOM3 in the AC samples is an interesting finding that indicates that specific inactivation of SHROOM3 in foci of arachnoid cells may contribute to the development of cysts.

SOX9 is a transcription factor located at chromosome 17q24, and inactivation of one allele causes campomelic dysplasia (CD) [23]. CD patients typically have short, bowed long bones, craniofacial defects and female to male sex reversal. Interestingly, cystic hygroma and cystic kidneys have been reported in CD [24]. In the developing brain it is expressed in neural crest cells; the very same cells that differentiate into meninges, and animal studies have shown that inactivation of SOX9 in these cells results in craniofacial maldevelopment [25].

From the literature, none of the remaining differentially expressed genes are reported to have known functions linking them to cystogenesis or neural tube development. However, this does not imply that the genes have no function in AC formation as the biological function of many genes are poorly characterized and many genes may have different or multiple functions in different tissues at different stages during the life cycle. Therefore, more research is needed to explore the properties of these genes in humans. Hence, based on current but incomplete knowledge, the most promising candidate genes with a role in the formation of ACs are SHROOM3 and SOX9.

For validation of the differentially expressed genes, we selected four of them for qRT-PCR: ATP10D, BEND5, SHROOM3 which were significantly altered in the ACs (p = 0.005, p = 0.008, p = 0.005, respectively), and SOX9 which showed the same trend, but did not reach significance (p = 0.08). The expression of ATP10D, BEND5, SHROOM3 in the qRT-PCR study correlated significantly with that of the microarray study; i.e. a low expression found in the AC by microarray was similarly low by qRT-PCR. This adds internal validity to our microarray platform and thus to the results in general.

Microarray-based CNV analysis was used to study chromosomal aberrations in a detailed manner; with the platform used in our study, more than 946,000 copy number probes span the genome. Thus, submicroscopic DNA regions with duplications or deletions can be identified. Such regions might contain genes of importance for the formation of the cysts. Because of limited amounts of tissue only six AC samples were subjected to CNV analysis. In three of these cases we found a total of 14 CNVs so far not reported in the human genome: all of which were chromosomal gains (table 3). In our samples, the CNVs were located on chromosomes 7, 8, 10, 11, 12, 14, 15, 16, 17, and 19, and several genes are known to be located in these regions. The CNV results were matched against the publicly available Toronto database, as well as against a population based Norwegian sample of 361 individuals located in our laboratory. Although there were no mutual CNVs present in all the AC samples, this does not weaken their potential role. On the contrary, a growing body of evidence from studies on schizophrenia and autism suggests that different DNA alterations on different chromosomes may ultimately contribute to the development of the same phenotype [26‐28]. This illustrates the complexity of the genome and its regulation in health and disease, and suggests that similar mechanisms may underlie the pathogenesis of AC.

Gene expression research on diseases affecting the arachnoid membrane (e.g. meningiomas, post-SAH inflammation and meningitis) requires normal arachnoid tissue as control. Since normal arachnoid is more difficult to obtain than AC tissue in quantities sufficient for gene expression analysis, one could ask whether ACs could be used in this research as a substitute for normal arachnoid membrane. In the current study we have shown that the vast majority of genes had similar expression profiles in AC and normal arachnoid membranes. If adjusted for the nine differentially expressed genes, our results suggest that AC tissue may be used as control tissue in such studies.

A limitation of the present study is the low number of normal AM samples used as control tissue. At time of study we had only two AM samples sufficient for cDNA microarray and four AM samples for qRT-PCR analysis. Because of this we had to use strict criteria in the data analysis of the microarray results for the detection of differentially expressed genes. However, when we validated the genes with qRT-PCR and included additional AC and AM samples, the microarray findings were confirmed. As we also have shown, we found a significant correlation between the qRT-PCR and microarray results. Furthermore, all AC samples were supratentorial and the AM controls infratentorial. Thus, one might argue that the differential gene expression may rely on purely anatomical grounds. However, it is our opinion that structurally there is no difference between supra- and infratentorial arachnoid, and that if this were true many more genes would be differentially expressed.

In summary, this is the first study on the mRNA gene expression in intracranial AC membranes and we have shown that for the vast majority of genes, the expression profiles in AC and AM samples were similar. However, our study showed that a small subset of genes was differentially expressed in the AC samples compared to normal tissue. The differential expression level of these genes might be important for the development of ACs. We also found duplicated DNA regions. Particularly, the functional role of the altered gene expression for SHROOM3 and SOX9 must be further studied in model systems and the findings of duplicated chromosomal regions have to be verified in replication studies containing more samples.