Congenital neomelanocytic naevi appear in nearly 1% of all newborn infants. Most of these naevi are small; however, a special group of melanocytic lesions cover large areas of the body. These large lesions, termed giant hairy naevi (GHN), are of concern because they have the potential to transform into malignant melanomas, with an incidence of 1% to 25%.^1–8 The malignant melanoma that develops within other naevi typically begins at the epidermal–dermal junction, whereas in a GHN, malignant melanoma originates deep in the dermis.^1,7 Giant hairy naevi are relatively uncommon congenital malformations that are usually deeply pigmented, covered with a moderate growth of hair, and occur over large areas of the body. They represent developmental lesions arising from migratory arrest and proliferation of primitive neuroectodermal cells.⁹ The available evidence suggests that neoplasms arising in this setting are diverse in their morphological expression and variable in their clinical behaviour.¹⁰ Because most of these lesions are treated early, for cosmetic reasons and because of their malignant potential, the true incidence of malignant degeneration is unknown.

“Gene expression profiling by microarrays represents a valuable tool to identify new players in complex pathophysiological processes, such as melanoma”

Oligonucleotide or cDNA arrays have dramatically changed the study of the pattern of gene expression by enabling the simultaneous analysis of thousands of genes in a single experiment. For this reason, gene expression profiling by microarrays represents a valuable tool to identify new players in complex pathophysiological processes, such as melanoma.^11–13

In our study, gene expression profiles are described in GHN and the nearby normal skin of patients with GHN, in comparison with the normal skin of unaffected patients. Gene expression profiling in these lesions may provide insight into the pathogenesis of GHN, and perhaps into the biological potential of individual lesions.

METHODS

Patients and study design

Tissues from three patients with GHN and two patients with cleft lip and cleft palate (normal controls) were studied. The patients ranged in age from 2 to 22 years, and were admitted to our hospital for the removal of giant hairy naevi or for reconstructive surgery. Samples of tissue were taken during planned surgical procedures from GHN, from apparently normal skin near to the hairy naevus, and from the normal skin of patients admitted for the reconstruction of a cleft lip or cleft palate, in accordance with an IRB approved protocol. The samples were taken from the elbow, face, and back of the trunk. Samples were snap frozen and stored at −70°C for DNA microarray analysis and reverse transcription polymerase chain reaction (RT-PCR) analysis. Patterns of mRNA expression were compared between GHN and nearby normal skin, and between the normal skin near the GHN and the normal skin of control patients without such naevi. Figure 1 depicts an example of a GHN involving large areas of the trunk and legs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Child (2 years old) with a giant hairy naevus covering the back, legs, and abdomen.

Total RNA extraction

Total RNA was isolated from the frozen tissue samples by acid guanidinium thiocyanate/phenol/chloroform extraction using TRI® reagent (Molecular Research Center Inc, Cincinnati, Ohio, USA). This method was based on the single step method of RNA isolation described by Chomczynski and Sacchi.¹⁴ Samples were homogenised in TRI® reagent on ice and total RNA was extracted according to the manufacturer’s instructions. Purified RNA was measured by ultraviolet absorbance at 260 and 280 nm and stored in 25 μg aliquots at −70°C for DNA microarray hybridisation and analyses. The adequacy and integrity of the extracted RNA were determined by gel electrophoresis.

High density oligonucleotide array analysis

Probe labelling, hybridisation, and image acquisition were done according to the standard Affymetrix protocol. Briefly, 25 μg samples of purified total RNA were transcribed into cRNA, purified, and used as templates for the in vitro transcription of biotin labelled antisense RNA. The biotinylated antisense RNA preparation was fragmented, and placed in a hybridisation mixture containing four biotinylated hybridisation controls (BioB, BioC, BioD, and Cre). Eight samples (three GHN, three nearby normal skin of patients with GHN, and two normal controls) were then hybridised to eight Affymetrix gene chip arrays (HG-U95 Av2, from identical lots) for 16 hours. The arrays were washed and stained using the instrument’s standard eukaryotic GE wash 2′ protocol and antibody mediated signal amplification. The images were scanned and analysed with Afffymetrix GeneChip analysis suite 5.0. Images from each gene chip were scaled and adjusted to an average intensity value for all arrays of 1500. Scaled average difference values and absolute call data from each gene chip were exported to data files and used for statistical analysis.^15,16 A probe is “present” if its absolute call was “P” for at least two members of the group containing all samples. Otherwise, the gene was considered not expressed in the group using the Affymetrix algorithm. In vitro transcription and chip hybridisation were performed in collaboration with the UTMB Genomic Core Facility.

Data analysis

Data analysis was carried out as described previously, including: chip validation, normalisation, filtering of data, and generation of gene lists by comparing (1) GHN and nearby normal skin, and (2) the normal skin near the GHN and normal skin of control patients without such naevi.¹⁵ The first step was to determine the degree of similarity or dissimilarity among sample transcription profiles by using model based expression analysis of oligonucleotide arrays.¹⁶ The next step was the elimination of genes that showed little variation across the samples, or that were absent in most of the samples. The first criterion was that the ratio of the SD and the mean of a gene’s expression across all samples was less than the threshold (of 0.85 and the upper limit of 8). Data were discarded if there was a large deviation in the number of present calls or if the correlation coefficient among the three chips within the group was less than 0.85. The second criterion required a gene to be called present in more than 80% of the arrays. We determined the presence (P) or absence (A) of each probe within the group according to the Affymetrix algorithm. A probe is “present” if its absolute call was “P” for at least two members of the group containing all samples. Otherwise, the gene was considered not expressed in the group. The primary goal was to identify genes with significant differences in expression between the test and control groups. The within group average of expression was calculated and comparisons were made between groups. Initially, comparisons were done by computing the expression fold difference for each gene and listing those that showed more than 1.5 fold increases or decreases in activity. Three groups were used for comparison: one group consisted of tissue from GHN (n  =  3), one consisted of nearby normal skin from patients with GHN (n  =  3), and one was normal control skin from patients with cleft palate (n  =  2). Paired comparisons were made between GHN and nearby normal skin, whereas unpaired comparisons were made between the uninvolved skin near to GHN and normal control skin. An entry was discarded as an outlier when its value was greater than 3 SD from the mean. Only significant differences at p < 0.05 were retained.¹⁷ Considering the number of samples in each group and its influence on the validity of the analysis, the power of the t test was computed. If less than 0.8, results were discarded even when p < 0.05. By using HG-U95Av2 Affymetrix arrays, about 6000 of the 12 625 genes present in the array were expressed in all the chips.

RT-PCR

The amount of total RNA extracted from GHN and nearby normal skin was measured by ultraviolet spectrophotometry and the RNA was stored at −70°C for RT-PCR analysis. The cDNA reaction and the PCR were performed with an optimised buffer and enzyme system (Titan™ One Tube RT-PCR system; Roche, Indianapolis, Indiana, USA). This system is designed to use avian myeloblastosis virus (AMV) reverse transcriptase for first strand synthesis and the Expand™ high fidelity blend of thermostable DNA polymerases, which consists of Taq DNA polymerase and a proof reading polymerase, for the PCR part. The reaction was carried out in a 50 µl volume containing 50–100 ng of total RNA, 1 μM of forward and reverse primers specific for hypoxia inducible factor 1 (HIF-1; GenBank Accession Number, U22431; forward: 5′-ACC CCA TTC CTC ACC CAT CA-3′ and reverse: 5′-TCC ACC TCT TTT GGC AAG C-3′), dopachrome tautomerase (TRP-2; GenBank Accession Number, AJ000503; forward: 5′-GGA GAA AAG TAC GAC AGA GAC AAG G-3′ and reverse: 5′-AGA AAA GCC AAC AGC ACA AAA AGA C-3′), and vascular endothelial growth factor (VEGF; primer pairs from R&D Systems, Minnesota, Minneapolis, USA); 1× PCR buffer with 4mM Mg²⁺, 0.2mM dNTPs, 5mM dithiothreitol solution, 5–10 units of RNAse inhibitor, and 0.05 U/µl of the enzyme mix (High Fidelity enzyme mix, reverse transcriptase, AMV in storage buffer). An initial RT step was performed at 50°C for 30 minutes and 94°C for two minutes for one cycle, followed by 35 cycles of denaturation at 94°C for 15 seconds, annealing at 55°C for 30 seconds, and extension at 72°C for 1.5 minutes, with a final extension step of 72°C for five minutes. In addition, primers for human β actin were used (R&D systems) as a “house keeping gene” to provide an internal marker for mRNA integrity within the experiment. PCR products were separated on (1.5% wt/vol) agarose gels, and visualised by means of ethidium bromide staining under ultraviolet light. Image capture and density analysis of bands were done with the SynGene™ gel documentation system (SynGene-Synoptics, Cambridge, UK) and the band intensity was expressed as the ratio of HIF-1 to β actin.¹⁸

Our study was conducted in compliance with the requirements for institutional review and informed consent at the University of Texas Medical Branch, Galveston, Texas, USA.

RESULTS

Histopathology

The diagnosis of GHN was confirmed histologically in each case. The dermal portions of the biopsies of the GHN were almost filled with moderate sized to small “naevus cells” with uniform ovoid nuclei, which are thought to represent immature melanocytic cells derived from the neural crest (fig 2A, B). Similar cells, which often were smaller and more elongated, extended to the deepest portion of the dermis. In the upper dermis, the cells had more cytoplasm, often forming rounded nests, and often contained finely granular brown cytoplasmic pigment. In some cases, nests of naevus cells were in contact with the dermal–epidermal junction. Multiple long hair follicles were present. The naevus cells made up just over 50% of the total number of cells. The other cells did not differ dramatically in their populations between GHN and normal skin, except for the macrophages, which were approximately two to fivefold more numerous in the naevi, and presumably more activated, by virtue of phagocytosis of melanin pigment. In general, the normal skin taken from patients with GHN was histologically identical to that from patients without GHN. There was the expected random variability in epidermal thickness and degree of pigmentation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

 Light micrographs of giant hairy naevus. (A) Cells with orderly nuclei and scant cytoplasm occupy the entire dermis. Melanophages can be seen beneath the epidermis. Haematoxylin and eosin stain (H&E); original magnification, ×100. (B) At low magnification, large regularly spaced hair follicles are apparent. The dermis appears to be cellular and there are foci of pigmentation in the papillary dermis. H&E stain; original magnification, ×40.

Gene expression patterns

After survey and statistical analysis of the 12 625 genes on each array, there was increased expression of 22 genes in the GHN tissue compared with the nearby normal skin, and a decrease in 73 genes (table 1). Thus, 0.75% of the genes were altered in GHN when compared with nearby normal skin. There were 227 genes that changed across all samples, as determined by microarray analysis, satisfying the filtering criteria as mentioned above. Categories of genes affected were those encoding structural proteins, proteins related to developmental processes, cell death associated proteins, transcription factors, growth factors, stress response modulators, and collagen associated proteins. A comparison between the unaffected nearby normal skin of patients with GHN and normal control skin generated a list of genes, 36 of which had increased expression and five of which had decreased expression (table 2).

Confirmation by RT-PCR

To confirm the findings using a different methodology, the degree of expression of three genes (HIF-1, TRP-2, and VEGF) found to be significantly altered in GHN compared with nearby normal skin was analysed by RT-PCR. The ratio of TRP-2 mRNA to β actin in GHN was 4.32 (SD, 0.06), compared with 0.94 (SD, 0.11) in normal skin (p < 0.05). The ratio of HIF-1 mRNA to β actin in normal skin was 2.32 (SD, 0.06), compared with 0.74 (SD, 0.01) in GHN tissue, and the ratio of VEGF to β actin in normal skin was 3.64 (SD, 0.15), compared with 0.82 (SD, 0.2) in GHN (p<0.05), which corroborate the microarray data.

DISCUSSION

Little information is currently available as to how changes in gene expression in the skin affect the development of GHN. New technology now allows us to investigate changes in gene expression using high density oligonucleotide microarrays in abnormal tissue such as GHN, in comparison with normal skin. Gene profiling may provide information on how these lesions develop and progress.

Comparison of the gene expression profile in GHN with the patient’s own nearby normal skin showed that there was a significant increase in the expression of 22 genes and a decrease in the expression of 73 genes. These genes can be categorised into functional groups. As examples, TRP-2 (GenBank accession number, AJ000503) and tyrosinase related protein 1 (TRP-1; GenBank accession number, X51420) were significantly raised. Both of these are recognised as useful markers of melanoma, because expression of these genes is restricted to cells of melanocytic origin. Tyrosinase protein or its mRNA can be detected even in poorly differentiated amelanotic melanomas,^19–21 as documented at the ultrastructural level.²² The relative abundance of tyrosinase mRNA in some undifferentiated amelanotic melanomas¹⁹ allows its use in PCR techniques as a valuable diagnostic option.²⁰ The same can be said for the determination of TRP-1, recognised by the commercially available antibody MEL-5, and TRP-2, which acts as dopachrome tautomerase. Expression of these genes and proteins is found predominantly in cells of melanocytic lineage. Thus, identification of the protein tyrosinases, TRP-1 and TRP-2, and structural melanosomal proteins, collectively called melanogenesis related proteins, provides a highly specific and sensitive tool for the positive identification of melanocytes. Pocket proteins are thought to govern the G1 to S phase transition mainly through their interaction with E2F transcription factors. E2F transcription factor (GenBank accession number, AL021154) expression was significantly decreased in GHN. E2F regulates the transcription of a large set of genes involved in all aspects of S phase initiation and progression, such as the mcm genes, cyclin E, proliferating cell nuclear antigen, and DNA polymerase α. Alterations in the cell cycle, DNA damage and repair, and tumour suppressor genes have been linked to the development of cancer. In GHN, two tumour suppressor genes were found to be altered. Deleted in liver cancer 1 (DLC1; GenBank accession number, AF035119) gene expression is increased, whereas secreted frizzled related protein 1 (hsFRP; GenBank accession number, AF056087) is decreased. HIF-1α (GenBank accession number, U22431) gene expression was decreased in GHN. HIF-1 is a transcription factor that functions as a master regulator of oxygen homeostasis. Over 40 HIF-1 target genes have been identified, which encode proteins that play key roles in crucial developmental and physiological processes, including angiogenesis/vascular remodelling, erythropoiesis, glucose transport, glycolysis, iron transport, and cell proliferation versus survival. HIF-1 is crucial for a variety of essential biological processes. HIF-1α expression and activity are regulated by major signal transduction pathways, including those involving phosphatidylinositol 3-kinase and extracellular regulated kinas/mitogen activated protein kinase. Although the number of patients reported in our study is small, the paired study design and the rigorous statistical comparisons used provide some confidence that the results are meaningful. A greater number of subjects are needed to support these differences; in addition, our study was limited by investigating only congenital giant naevi. Further studies on common small naevi would be of interest and might yield different findings.

“Although the number of patients reported in our study is small, the paired study design and the rigorous statistical comparisons used provide some confidence that the results are meaningful”

Comparison between the unaffected nearby normal skin of patients with GHN and normal control skin generated a list of 36 genes with increased expression and five with decreased expression. There were strong increases in melanoma associated gene (GenBank accession number, XM_056455) and SPARC-like 1 (SPARCL1; GenBank accession number, X86693; also known as MAST9 or hevin) expression. SPARC is strongly expressed in bone tissue, is distributed widely in many other tissues and cell types, and is associated generally with tissue remodelling—for example, it is abundant in tissues undergoing morphogenesis, mineralisation, angiogenesis, and pathological responses to injury and tumorigenesis. Experiments in vitro have provided evidence that SPARC disrupts cell adhesion, promotes changes in cell shape, inhibits the cell cycle, regulates cell differentiation, inactivates cellular responses to certain growth factors, and regulates extracellular matrix and matrix metalloprotease production. SPARC plays a role in angiogenesis, tumorigenesis, cataractogenesis, and wound healing.²³ The expression of genes for histone acetyltransferase p300 and cAMP responsive element binding protein (CBP; GenBank accession number, U65093) were increased. These proteins are required for the execution of crucial biological functions such as proliferation, differentiation, and apoptosis. Both proteins are believed to regulate the activity of a large number of general and cell specific transcription factors. One consequence of p300 depletion is transcriptional downregulation of the cyclin E gene, caused by deacetylation of histones at its promoter, leading to activation of senescence checkpoints in the cell cycle.²⁴

At the present time, the pathogenesis of GHN is not understood at the molecular level. There are clearly substantial differences between these giant naevi and the much more common small hereditary naevi, and between these and common naevi acquired during life. There are also differences in the biological behaviour of GHN among individuals, the most striking of which is the development of malignant melanoma in a small number of patients. It is hoped that the analysis of gene expression profiles may begin to provide data that will be useful in increasing our understanding of the biology of these lesions and the problems that they present.