Introduction

Neurofibromatosis type 1 (NF1) [MIM16220] is an autosomal dominant tumor predilection syndrome caused by mutations in the NF1 gene on chromosome 17q11.2 [1]. Its hallmark symptom is multiple benign dermal neurofibromas. The observed inter- and intrafamilial variability of several NF1 symptoms (e.g., neurofibroma number) has led to the hypothesis that they are modified by other genes [2]. Among the proposed candidates are genes influencing somatic mutation rates [3], genes involved in pathways regulated by neurofibromin, as well as steroid hormones and their receptors [4]. The rare ‘NF1 microdeletion syndrome’ is of special interest in the search for modifying genes. These patients carry a heterozygous microdeletion of NF1 and several contiguous genes [58] and often have a severe clinical phenotype including mental retardation, congenital heart defects, and an excessive number and early onset of dermal neurofibromas [9]. Fourteen protein-coding and two microRNA genes lie within the most prevalent 1.5 Mb deletion [1013], which arises from unequal meiotic crossover of two highly homologous low copy repeats flanking NF1 [14]. It seems likely that the haploinsufficiency of one or more of these genes contributes to the severe phenotype. Candidate genes for mental retardation, congenital heart malformation, and overgrowth have been identified among them [1517]. None has as yet been associated with dermal neurofibroma formation, thought to be triggered by second hit NF1 mutations in Schwann cells or their precursors [18] with the subsequent involvement of mast cells and fibroblasts [19, 20]. We hypothesize that interindividual variations in one of the genes of the microdeletion region also account for the variable neurofibroma number in classical NF1 patients. If so, this gene should be expressed in at least one type of neurofibroma precursor cell and presumably also in the neurofibroma itself. We therefore examined the expression of 13 of the 14 protein-coding genes in neurofibromas, peripheral nerve tissue, and cell types involved in neurofibroma formation. The SUZ12 pseudogene SUZ12P was excluded because of its extremely high homology to sequences on chromosome 19. The location and orientation of these genes together with the low copy repeats are depicted in Fig. 1.

Fig. 1
figure 1

Location, name, and orientation of our candidate modifying genes on human chromosome 17. The measuring staff shows the physical map coordinated on chromosome 17q11.2 in standard notation from the p-telomere to the q-telomere. Genes depicted above the measuring staff are transcribed in sense; genes below the staff are transcribed in antisense orientation (also indicated by the arrows). The grey boxes show the low copy repeats (LCR) where the breakpoints of the most common 1.5 Mb microdeletions are located

Materials and methods

Tissues and cells

Nine dermal neurofibromas from seven unrelated NF1 patients and plexiform neurofibroma tissue from four unrelated patients were received. Macroscopically normal peripheral nerve adhering to tumor was obtained in two cases. Two peripheral nerve samples (N. femoralis and N. ulnaris) were received from one NF1 patient. Samples were surgically excised and directly transferred to RNAlater™ (Ambion). Tissue of malignant nerve sheath tumors (MPNST) was obtained from two unrelated NF1 patients and flash frozen in liquid nitrogen. All patients gave informed consent. Primary human fibroblasts derived from the skin of NF1 patients and a healthy control and the NF1-MPNST cell line NFS-1 (kindly provided by Dr. Vincent Riccardi) were cultured in DMEM with 10% FCS. NFS-1, suggested to be of Schwann cell origin [18], shows 26% of the neurofibromin amount of control fibroblasts (J. Griesser, doctoral thesis), indicating an unstable protein. The human mast cell line LAD2 (kindly provided by Dr. Kirshenbaum, NIH, NIAID) was cultured as described [21].

LOH analysis

Two dermal neurofibromas from 4 NF1 patients were subjected to loss of heterozygosity (LOH) analysis as described [22, 23] with markers D17S907 (located within the schlafen family member 12-like gene (SLFN12L), 4.16 Mb downstream from NF1) and either IVs27TG24.8, D17S1800, or IVs27ac28.4 (located within NF1) (kindly provided by Dr. Kehrer-Sawatzki).

Expression analysis

Total RNA was isolated using the RNeasy or the RNeasy Fibrous Tissue Mini Kit (Qiagen) according to the manufacturer’s protocol and treated with DNase I Amplification Grade (Invitrogen Life Technologies). One microgram thereof was reverse transcribed with the SuperScript™ First-Strand Synthesis system and random hexamers (Invitrogen Life Technologies). cDNA quality was controlled by amplifying part of the housekeeping gene RPII [24]. PCR conditions using 3 μl cDNA as template were 95°C for 2 min, followed by 35–40 cycles of 30 s at 95°C, 30 s at annealing temperature, 30 s at 72°C, followed by 5 min at 72°C. PCR products were separated on 1.5% agarose gels stained with ethidium bromide. Band intensity was measured with Kodak 1D Image Analysis Software and genes displaying less than 10% of the expression measured for the housekeeping gene were defined as not expressed.

The expression of CENTA2, RAB11FIP4, C17orf79, and UTP6 was also measured in a LightCycler 1.0 (Roche Applied Science) with the QuantiTect SYBR Green PCR mix (Qiagen) and 2 μl of diluted cDNA according to the manufacturer’s protocol. For analysis, the crossing point (C t) values were determined with the Fit Points method of the Lightcycler 3.5 software and a fluorescence threshold (F t) of 0.01. The initial template fluorescence (F 0) was calculated as described [25]. Normalization was carried out with the geometric mean of the reference genes G6PDH and TBP [24]. Primers, annealing temperatures, and product sizes are listed in Table 1.

Table 1 Primers used for the amplification of mRNA of putative NF1 modifying genes

Results

LOH analysis

A constitutive NF1 microdeletion could be excluded for the four NF1 patients tested (NF.326, NF.328, NF.329, and NF.331) as at least one of the two neurofibromas tested showed both NF1 copies. One neurofibroma of patient NF.326 showed a somatic deletion at both markers (peak reduction by approximately 44%) and was subjected to separate real-time expression analysis together with one NF.326 tumor without LOH. In the remaining three patients, investigation of transcribed polymorphic sites in genes lying within the deletion interval was not informative.

Expression analysis

Four of 13 genes (CENTA2, RAB11FIP4, C17orf79, and UTP6) were found to be expressed in all dermal neurofibromas (Table 2), all plexiform neurofibromas, peripheral nerve tissue (Table 3), and most cultured cells (Table 4). In all plexiform neurofibromas and most nerve tissue samples, six more genes (ATAD5, C17orf42, RNF135, EVI2B, SUZ12, and LRRC37B) were expressed. One gene, OMG, was not expressed in any sample. CENTA2, RAB11FIP4, C17orf79, and UTP6 were investigated in detail as candidate modifiers.

Table 2 Expression of candidate modifying genes in different tissues and cultured cells: expression in dermal neurofibromas
Table 3 Expression of candidate modifying genes in different tissues and cultured cells: expression in plexiform neurofibroma and nerve tissue
Table 4 Expression of candidate modifying genes in different tissues and cultured cells: expression in cell cultures

Real-time RT-PCR of candidate genes

Real-time RT-PCR of the candidate genes shows an upregulation of CENTA2 (significant; Mann–Whitney test, p-value 0.04), RAB11FIP4 (p-value 0.046), and UTP6 in dermal neurofibromas compared to peripheral nerve tissue (Table 5). No distinct change in regulation was found for C17orf79 where two samples showed a reduction, four samples were unchanged, and one showed an increase in expression.

Table 5 Expression of CENTA2, RAB11FIP4, C17orf79, and UTP6 measured by real-time RT-PCR

Direct comparison of one neurofibroma from NF.326 with LOH (comprising at least NF1 to SLFN12L and thus including RAB11FIP4, C17orf79, and UTP6) to one of the same patient without LOH revealed a reduced upregulation of RAB11FIP4 (310% to 113%) and a greater downregulation of UTP6 (45% to 12%) in the tumor with LOH. CENTA2 and C17orf79 showed no difference between the two samples (Fig. 2).

Fig. 2
figure 2

Expression of CENTA2, RAB11FIP4, C17orf79, and UTP6 in dermal neurofibromas from patient NF.326 with and without LOH measured by real-time RT-PCR. Expression is shown relative to the mean of two NF1 +/− nerves, which was arbitrarily set to 100%. −LOH without loss of heterozygosity; +LOH with loss of heterozygosity

In contrast to the dermal neurofibromas, C17orf79 was upregulated in plexiform neurofibromas and even more so in MPNSTs (Table 5). Great discrepancies were found in some cases between the data from cultured cells and the data from tissue (Table 5). A reduced expression, interesting in view of our hypothesis, was found for RAB11FIP4 in fibroblasts and UTP6 in LAD2 cells.

Discussion

We hypothesize that one of the genes lying in the NF1 microdeletion region not only causes the high number and early onset of neurofibromas in the microdeletion patients but also accounts for the variation of neurofibroma number seen in classical NF1 patients. Accordingly, they would differ in level or activity of the putative modifying protein in cell types involved in neurofibroma formation. As only a difference in an expressed gene can have a modifying effect, we examined the expression of 13 genes of the NF1 microdeletion region in peripheral nerve tissue, dermal, and plexiform neurofibromas. The majority of cells in a neurofibroma are Schwann cells, with an unknown percentage of NF1 −/− Schwann cells. The other cell types are underrepresented which is why we also examined representative cell cultures such as primary fibroblasts, a mast cell line and, as representative for Schwann cells, an MPNST cell line.

Only four genes, CENTA2, RAB11FIP4, C17orf79, and UTP6, showed expression in all of the dermal and plexiform neurofibromas investigated as well as in nerve tissue, primary fibroblasts, and the MPNST cell line. They seem to function in basic cellular processes the disruption of which may promote aberrant growth. CENTA2 is a GTPase activating protein (GAP) for ARF6 [26], an ADP-ribosylation factor known to regulate tumor cell invasion [27]. The neurally expressed RAB11FIP4 is involved in the control of membrane trafficking during cytokinesis [28]. UTP6 influences apoptosome-mediated apoptosis via interaction with Apaf-1 [29]. C17orf79 (COPR5) binds arginine methyltransferase 5 (PRMT5) and thereby acts on the promoters of cyclin E1 (CCNE1) and the tumor suppressor gene NM23. COPR5 overexpression leads to reduced CCNE1 levels [30].

Real-time RT-PCR investigation of the candidate gene expression level revealed no clear downregulation in neurofibroma tissue compared to peripheral nerve. Two explanations for these findings are conceivable. First, it is possible that downregulation of a modifying gene occurs only during neurofibroma initiation and it is upregulated thereafter. Second, the level might only be reduced in one neurofibroma cell type but increased in others. Therefore, it seemed reasonable to also investigate the different cell types a dermal neurofibroma is comprised of. However, because of the sometimes strong discrepancies between the in vivo and in vitro data, these results have to be interpreted with caution. RAB11FIP4 was found to be reduced in NF1 +/− fibroblasts, UTP6 in the mast cell line. Increased expression of both genes was measured in the other cell types. This would support our second explanation mentioned above.

Direct comparison of a neurofibroma with LOH to one without LOH from the same patient showed the expected reduction in the neurofibroma with LOH only in the case of RAB11FIP4 and UTP6. Interestingly, CENTA2 and C17orf79 expression did not differ between both tumors. The LOH tumor carries an atypical deletion comprising at least NF1 to SLFN12L. As CENTA2 is located upstream of NF1, it might not be deleted, which would explain the unchanged expression. C17orf79, however, is located within the deleted region. In this case, upregulation of the remaining allele in the tumor with LOH is one possible explanation. Another possibility is higher expression of C17orf79 in one tumor cell type which has a growth advantage in the LOH tumor. Together with the finding of C17orf79 overexpression in plexiform neurofibroma and MPNST tissue, this points to an interesting function of this gene in tumor tissue.

In summary, CENTA2, RAB11FIP4, C17orf79, and UTP6 are candidates for modifying NF1 neurofibroma formation. Further studies, especially on protein activity in different NF1 patients, are necessary to elucidate their specific role in neurofibroma formation.