• comparative genomic hybridization;
  • p53;
  • ras;
  • FHIT


  1. Top of page
  2. Abstract
  6. Acknowledgements

Two developmentally highly divergent nonmelanoma skin cancers, the epidermal squamous cell carcinomas (SCC) and the neuroendocrine Merkel cell carcinomas (MCC), occur late in life at sun-exposed body sites. To determine whether these similarities may indicate common genetic alterations, we studied the genetic profile of 10 MCCs and analyzed 6 derived cell lines and 5 skin SCC lines by comparative genomic hybridization (CGH) and molecular genetic analyses. Although the MCCs were highly divergent—only 3 of the 10 tumors exhibited common gains and losses—they shared gain of 8q21-q22 and loss of 4p15-pter with the genetically much more homogeneous SCC lines. In addition, 2 of 5 SCC and 2 of 6 MCC lines exhibited UV-B-type-specific mutations in the p53 tumor-suppressor gene and a high frequency (9/11) of CC[RIGHTWARDS ARROW]TT double base changes in codon 27 of the Harvey (Ha)-ras gene. Since 45% of the tumor lines were homozygous for this nucleotide substitution compared to 14% of the controls and in 1 MCC patient the wild-type allele was lost in the tumor, this novel polymorphism may contribute to tumor development. On the other hand, loss of 3p, characteristic for SCCs, was rare in MCCs. Although in 2 of 3 SCC lines 3p loss was correlated with reduced expression of the FHIT (fragile histidine triad) gene, the potential tumor suppressor mapped to 3p14.2 and 2 MCC lines with normal 3p showed aberrant or no FHIT transcripts. Taken together, in addition to the common UV-B-specific mutations in the p53 and Ha-ras gene, MCCs and SCCs also share chromosomal imbalances that may point to a common environmental-derived (e.g., UV-A) oxidative damage. © 2002 Wiley-Liss, Inc.

Skin squamous cell carcinomas (SCC) and Merkel cell carcinomas (MCC), although likely of different origin, belong to the group of nonmelanoma skin cancers (NMSC). SCCs are derived from epidermal keratinocytes and commonly manifest at the ages of 60–70. Being capable of local invasion and distant metastasis, SCCs can follow a rather aggressive clinical course and contribute substantially to NMSC mortality.1 Furthermore, due to immunosuppression, the number of SCCs significantly increased in organ transplant recipients,2 demonstrating that skin carcinomas are well controlled by the immune system.3, 4

Genetically, SCCs are so far best characterized by loss of heterozygosity (LOH), a technique to determine loci of potential tumor-suppressor genes. Using this technique, a distinct pattern of allelic loss was established for chromosomes 3p, 9p, 13q, 17p and 17q.5 In addition, in rare cases aberrant expression of the FHIT (fragile histidine triad) gene, a potential tumor-suppressor gene mapped to 3p14.2,6 was observed.7 Occasionally, also activating point mutations were found in the Harvey-ras gene, although their relevance for skin cancer is questioned.8, 9 The most frequent aberrations are UV-indicative mutations in the p53 tumor-suppressor gene, i.e., C[RIGHTWARDS ARROW]T transitions in CC sites with about 10% CC[RIGHTWARDS ARROW]TT double base changes. These aberrations also provided the first molecular proof for the causal role of UV-B radiation in SCC development.10

MCCs, also known as trabecular carcinomas11 are believed to be neuroendocrine skin tumors probably derived from neurotactile Merkel cells, which are located in the basal layer of the epidermis.12 The tumors mostly manifest in the epidermis of the head and neck and, albeit rare, are extremely aggressive with a local recurrency rate of 30–40% and a high incidence of regional lymph node or distant metastases. Up to 35% of patients ultimately die of their disease.13, 14 Organ transplant recipients are even more susceptible to develop MCCs and the mortality rate increases up to 56%.15

The characteristic epithelial and neuroendocrine features of Merkel cells and their carcinomas are well studied by ultrastructural and immunohistochemic methods.12, 16 However, less is known about genetic alterations. Despite a number of cytogenetic studies which questioned the prevalence of consistent aberrations, chromosomes 1 and 6 are believed to be most frequently involved in numerical or structural aberrations resulting in gains of chromosomal material.17, 18, 19 Comparative genomic hybridization (CGH) studies partly confirmed these findings but also demonstrated prevalence of loss of 3p, 10 and 17p as well as gain of 3q, 5p, 8q, 19 and X.19, 20 In agreement with this, loss of heterozygosity was described for chromosome 3p,21 and the FHIT gene was shown to be abnormally expressed in MCC.22 An additional study reported on mutations of the p53 tumor-suppressor gene.23 From the 4 mutations found in 15 tumors, 1 was a UV-B-indicative C[RIGHTWARDS ARROW]T transition, thereby providing evidence for UV-B radiation to be an important “carcinogen” also for the development of MCCs.

Although it was suggested that MCCs exhibit histopathologic features of small cell lung cancer, location, late onset, as well as UV-type-specific p53 mutations did point to a genetic relationship with skin SCCs. We, therefore, aimed to establish this correlation by directly comparing the genetic profiles of these 2 tumor types. To obtain a more comprehensive view, we combined the molecular cytogenetic screening method of CGH24 with a number of molecular genetic approaches. This allowed us to identify gross chromosomal changes as well as aberrations in specific genes of the same tumor cell populations.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Tumors and cell lines

Tumor material obtained from 10 patients with Merkel cell carcinoma (MCC-1 to MCC-10) was collected in the Department of Dermatology, Mannheim Medical School (University of Heidelberg-Mannheim). Clinical data are provided in Table I. Corresponding tumor cell cultures were established from 5 tumors (MCCL-2 to MCCL-6) similar as described for MCCL-1.25 In addition, 4 squamous cell carcinoma lines established from primary facial skin carcinomas—SCL-I,26 SCL-II,27 SCC-12 and SCC-1328—were investigated. For comparison, data from the previously analyzed SCC line MET-1 were also included.29, 30

Table I. Clinical Data from Merkel Cell Carcinoma Patients
Patient/tumorAgeSexTumor tissueTumor localizationTumor cell line established
  • 1

    Line MCCL-1 was established previously and karyotyped by conventional G-banding analysis.25

MCC-173MaleCutaneous metastatic tumorAbdomen (primary tumor at the back)MCCL-11
MCC-272MaleCutaneous metastatic tumorUnknownMCCL-2
MCC-469MaleRecurrent tumorThighMCCL-4
MCC-590MalePrimary tumorLeft clavicleMCCL-5
MCC-672FemalePrimary tumorLeft calfMCCL-6
MCC-878FemalePrimary tumorRight lower arm
MCC-975FemalePrimary tumorCheek
MCC-1073FemalePrimary tumorNeck

Cell culture conditions

Establishment and growth properties of the MCC cell cultures were previously described in detail.25 Briefly, fresh tumor samples were transferred into RPMI 1640 culture medium supplemented with 15% FCS and antibiotics. Cell suspensions were prepared by mechanical dissociation of the tumor material and seeded onto feeder layers of lethally irradiated (50 Gy) adult human dermal fibroblasts. The cultures were incubated for approximately 2 weeks at 37°C in an atmosphere containing 5% CO2. The tumor cells grew as loosely arranged floating aggregates of different sizes. For passaging, the cells were centrifuged, resuspended and plated at a split ratio of 1:3 onto new fibroblast feeder layers. Depending on the cell growth rate, cells were passaged every 1–3 weeks and were cryopreserved in liquid nitrogen at various passages. The SCC lines were cultivated as described.26 Cells were disaggregated routinely with 0.1% trypsin/EDTA solution and replated at a split ratio of 1:10. Preparation of metaphase spreads and Giemsa banding was performed using standard procedures.

Comparative genomic hybridization (CGH)

DNA was prepared from peripheral blood as well as the 6 MCC and 5 SCC lines according to standard protocols.31 Tumor material from MCCs was available either from cryosections (MCC-1, -2, -3, -9, -10) or from formalin-fixed paraffin sections (MCC-4, -5, -6, -7, -8). Areas containing representative tumor tissue with more than 50% tumor cells were determined microscopically on H&E-stained sections and were labeled on subsequent serial sections (25 μm). Tumor cells were dissected from labeled areas and collected in Eppendorf tubes. In the case of paraffin sections, the tissue was dewaxed 3× 10 min in xylene (45°C) and 3× 5 min in ethanol. After overnight incubation in 1 M sodium thiocyanate (37°C), tissue samples were washed twice in DNA isolation buffer (75 mM NaCl, 25 mM EDTA, 0.5% Tween 20) followed by overnight digestion with 1 mg/ml proteinase K (55°C) and 1 hr digestion with 200 μg/ml RNase A (37°C). DNA was phenol/chloroform extracted, ethanol precipitated and further purified by using the QIAEX II protocol (Qiagen, Hilden, Germany). CGH analysis, image acquisition and processing were performed as described previously.32

PCR and DNA sequencing

Genomic DNA fragments from all cell lines were amplified by PCR for exons 5–8 of the p53 gene and codons 12/13 and 61 of the Ha-ras, Ki-ras and N-ras genes. Furthermore, genomic DNA fragments from the MCC lines were amplified for exons 1 and 2 of the CDKN2A gene. Amplification primers are specified in Table II. PCR was carried out in 50 μl sample volumes as follows: 5 min at 95°C followed by 33 cycles of 1 min at 95°C, 1 min at 53°C–62°C and 1 min at 72°C. PCR products were purified by gel electrophoresis, extracted from the agarose gel and directly sequenced according to the Thermo Sequenase radiolabelled chain terminator cycle sequencing protocol (Amersham Buchler, Braunschweig, Germany) using either the PCR primers or gene-specific nested primers (Table II).

Table II. Primers Used in This Study
GeneAmplified regionForward primer (5′ to 3′)Reverse primer (5′ to 3′)Tm (°C)Fragment length (bp)
  1. n.p., nested primer.

Genomic DNA

Reverse transcribed PCR (RT-PCR)

Poly A+ selected mRNA was extracted using the Oligotex direct mRNA kit (Qiagen, Hilden, Germany). cDNA was prepared from 100 ng of mRNA with random hexamer primers according to the cDNA synthesis protocol (Roche Diagnostics, Mannheim, Germany). CDNA-specific primers (Table II) were used to amplify a 494 bp CDKN2A fragment containing exons 1–3 and a 819 bp FHIT fragment containing the open reading frame (exons 4–10), respectively. For relative quantification, glycerolaldehyde-3-phosphate-dehydrogenase (GAPDH) was coamplified in the same reaction tube. PCR conditions were as follows: 10 min at 25°C, 60 min at 42°C, 5 min at 95°C and 5 min at 5°C. For cDNA amplification, PCR was performed for 33 cycles as described above with an annealing temperature of 62°C and 56°C, respectively. To avoid preferential amplification of the GAPDH transcript, PCR primers for GAPDH were added only after 19 cycles. PCR products were separated in a 1.5% agarose gel and stained with ethidium bromide.

Nonradioactive Southern analysis

Five micrograms of genomic DNA were digested with Eco RI, Bam HI and Hind III, resolved by 0.8% agarose gel electrophoresis and transferred to nylon membranes (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions. DNA was UV cross-linked to the membranes using a Stratalinker (Stratagene, La Jolla, CA). The membranes were incubated for 2 hr at 42°C in 50% formamide, 5× SSC, 0.1% N-lauroylsarcosine, 0.02% SDS and 2% blocking reagent (Roche Diagnostics). In a PCR reaction, the cDNA probe was labeled with digoxigenin-11-dUTP and diluted 1:3 in hybridization buffer (Roche Diagnostics). After overnight hybridization at 42°C, membranes were washed (2× 5 min in 4× SSC, 1% SDS, 2× 15 min in 2× SSC, 0.1% SDS and 2× 15 min in 0.1 SSC, 0.1% SDS, all washing steps at 68°C) and exposed overnight to Kodak X-Omat films.


  1. Top of page
  2. Abstract
  6. Acknowledgements

CGH profiles of MCCs are heterogeneous

In order to obtain an overall aberration pattern, we analyzed 10 MCCs (MCC-1 to MCC-10) by CGH. This resulted in 3 unexpected findings. First, the numbers of chromosomal changes differed remarkably in the individual tumors. In 4 MCCs, no or only 1 aberration was detected, whereas the other 6 tumors exhibited 4 up to 14 imbalances. Second, the aberration profiles were highly divergent with nearly all chromosomes being involved in genomic imbalances (Fig. 1, solid lines). Although the overall number of gains dominated, e.g., MCC-5 exhibited 11 gains and only 2 losses, we also identified 1 tumor with 8 losses and only 1 gain of chromosomal material (MCC-9), thus demonstrating a large genetic heterogeneity among these 10 MCCs. Third, we only detected few common (>2 tumors) aberrations. These were gain of 1q, 3q26-qter, 6p22-pter and 8q21-q22 and loss of 4p15-pter in 3 tumors (30%). Loss of 10q, 11q, 17p as well as gain of 5p, 12, 13q and X were seen in 2 MCCs (20%). With the exception of loss of 4p and gains of 12 and 13q, all other changes correlated with those described previously.19, 20 On the other hand, gain of 19 as well as loss of 3p and 13q, described as frequent aberrations previously,19 were not conspicuous in the MCCs analyzed here.

thumbnail image

Figure 1. Summary of chromosomal gains and losses as detected by CGH. Solid lines indicate imbalances present in Merkel cell carcinomas (nos. 1–10) and dotted lines represent tumor cell lines derived from cutaneous squamous cell carcinomas (A, SCL-I; B, SCL-II; C, SCC-12; D, SCC-13; E, MET-1). Gains are shown on the right and losses on the left side of the chromosome ideogram. Asterisks mark the position of genes analyzed by molecular methods (p53, 17p13.1; N-ras, 1p13.2; Harvey-ras, 11p15.5; Kirsten-ras, 12p12.1; FHIT, 3p14.2; CDKN2A, 9p21).

Download figure to PowerPoint

MCC lines reflect the aberration profiles of the respective tumors

From the 10 MCCs, 6 were established as cell lines (MCCL-1 to MCCL-6).25 To determine their relevance as in vitro models, they were investigated by CGH as well. These analyses demonstrated that the individual profiles of the tumors were well maintained in the respective cell lines (data not shown). Additional changes, most likely caused by culture adaptation, were rare. One of the most obvious changes was a partial loss of chromosome 7 in MCCL-2, whereas MCC-2 tumor exhibited a balanced CGH profile for chromosome 7.

SCC lines more frequently share common aberrations

In the SCC lines, the overall number of imbalances was substantially higher than in the MCCs with 8 up to 17 imbalances per cell line. Furthermore and in contrast to the heterogeneous distribution of gains and losses in the MCCs, the 5 SCC lines exhibited a higher frequency of common aberrations (Fig. 1, dotted lines). In at least 3 of the 5 cell lines, gains affected 7p11-p14 (60%), 8q22-qter (60%), 11q13 (100%) and losses affected 3p14-q11 (80%), 8p (60%) and 9p23-pter (80%). As suggested from our previous comparison of SCCs and derived cell lines,30 overrepresentation of chromosomes 17 and 20 (in 3 and 5 SCC lines, respectively) were likely to be culture derived. Thus, while from those aberrations most frequently detected in the SCC lines, namely gain of 11q and loss of 3p, only loss of 3p was detected in 1 MCC, gain of 8q22 and loss of 4p15-pter (present in 3 MCCs and 2 SCC lines) were obviously characteristic for both SCCs and MCCs.

Aberrant FHIT transcripts are only found in MCCs

One potential tumor suppressor on chromosome 3p is the FHIT gene and it has been suggested that aberrant FHIT transcripts may play a role both in SCC and MCC.7, 22 To determine the relevance of abnormal FHIT expression in our cell lines, we analyzed FHIT transcription by semiquantitative RT-PCR, using coamplified GAPDH as a standard. Whereas 4 of the 6 MCC lines exhibited normal transcripts, MCCL-1 and MCCL-5 cells exhibited the 397 bp GAPDH but no 819 bp FHIT band. Both cell lines were further investigated by nested primer PCR. Also under these conditions, MCCL-5 cells did not reveal amplification products, suggesting that FHIT expression was completely lost. In MCCL-1 cells, on the other hand, we observed two transcripts, migrating at 550 bp and 290 bp (Fig. 2a). Direct sequencing proved that the 550 bp band represented wild-type FHIT transcripts with exons 4 to 10 and with that including the entire coding sequences (exons 5–9). In the 290 bp band, exons 5–7 were replaced by a 36 bp insert, causing loss of the initiation codon and consequently the absence of a translation product (Fig. 2b). Interestingly, both of these cell lines exhibited normal CGH profiles of chromosome 3p.

thumbnail image

Figure 2. Expression of the FHIT gene in MCC lines studied by RT-PCR analysis. (a) Expression of the FHIT gene by nested RT-PCR analysis. No amplification product was obtained for MCCL-5, whereas a 550 bp and a 290 bp transcript are shown for MCCL-1. mRNA isolated from primary keratinocytes of a healthy proband was used as a control for RT-PCR analysis. A 1 kb ladder served as DNA marker. (b) A schematic representation of both FHIT transcripts detected in MCCL-1 cells. The 550 bp fragment corresponds to the wild-type transcript and contains the coding exons 5–9, while in the 290 bp fragment, exons 5–7 are deleted and 36 bp are inserted instead.

Download figure to PowerPoint

RT-PCR analysis of the SCC lines demonstrated normal-size FHIT transcripts for SCL-I, SCC-12 and SCC-13, whereas SCL-II and MET-1, both cell lines with a cytogenetic deletion of 3p14, exhibited the 397 bp GAPDH but no 819 bp FHIT fragment. By nested primer PCR, however, a 550 bp transcript was seen, indicating that expression of the FHIT gene was not completely abolished.30 Because this was the only transcript and its size corresponded to that of wild-type FHIT gene, 1 allele was obviously deleted and the other unaltered but only weakly expressed in these 2 SCC lines. Thus, while 2 MCC lines exhibited abnormal FHIT transcripts despite a normal distribution of chromosome 3p, the 3 SCC lines with loss of 3p expressed normal, although in part strongly reduced, levels of the FHIT transcripts.

UV-induced p53 mutations are found in MCC and SCC

A significant feature of skin cancer is the high frequency of p53 mutations and their UV-type signature. Since most of these mutations are found in exons 5–8 with the mutation hotspots for skin carcinomas in codons 177, 196, 247, 278 and 294,33 we performed sequence analysis of these exons with genomic DNA from the 6 MCC and 5 SCC lines. All amplification products had the expected size, thus excluding bigger insertions or deletions. Sequencing of the amplification products revealed mutations in 2 MCC lines (Table III). MCCL-5 cells carried a homozygous C[RIGHTWARDS ARROW]T transition in codon 278 (exon 8), causing an amino acid exchange from proline to leucine (Fig. 3a). MCCL-6 cells carried a heterozygous C[RIGHTWARDS ARROW]T transition in codon 179 (exon 5), causing an exchange from histidine to tyrosine (Fig. 3b).

Table III. Mutation Analysis of MCC and SCC Lines
Cell linep53 (exons 5–8)Harvey-, Kirsten- and N-ras (codons 12/13 and 61)FHIT mutationCDKN2A (exons 1 + 2) mutation
CodonBase changeAmino acid changeCodonBase changeAmino acid changeHarvey-ras codon 27 polymorphism
  • 1

    p53 mutation analysis confirms previously published data.34, 35

  • 2

    Data from the MET-1 cell line were previously published,30 but are included for comparison.–−, no mutation; n.d., not determined.

MCCL-1HomozygousDeletion exons 5–7, insertionDeletion  exon 2
SCL-I197C[RIGHTWARDS ARROW]TStopHeterozygousn.d.
thumbnail image

Figure 3. UV-induced p53 mutations in 2 Merkel cell carcinoma lines. (a) MCCL-5 cells show a homozygous C[RIGHTWARDS ARROW]T transition in codon 278, a mutation hotspot for skin carcinomas. (b) MCCL-6 cells exhibit a heterozygous C[RIGHTWARDS ARROW]T transition in codon 179. Point mutations are labeled by asterisks. Genomic DNA isolated from blood of a healthy proband served as a control. cDNA sequences are given for the mutated (Mut) and the wild-type (Wt) status.

Download figure to PowerPoint

From the 5 SCC lines, only 1 cell line, MET-1, was wild type for p53 (Table III).30 Two cell lines, SCL-II and SCC12, carried T[RIGHTWARDS ARROW]G transversions. SCL-II cells exhibited a mutation in codon 132, causing exchange from asparagine to lysine and SCC-12 cells, as described previously,34 a mutation in codon 216 causing an exchange from valine to glycine.34 SCL-I and SCC-13, on the other hand, exhibited UV-type p53 mutations. In SCL-I cells, 1 allele contained a CC[RIGHTWARDS ARROW]TT double base change in codon 196/197. The other allele contained a C[RIGHTWARDS ARROW]T transition in codon 197, causing a stop codon and thus leading to a truncated p53 protein (data not shown). In agreement with a previous report,35 SCC-13 cells carried a C[RIGHTWARDS ARROW]T transition in codon 258, causing an exchange from glutamine to lysine.

Ras mutations are absent or rare in skin carcinomas

In addition to mutational inactivation of p53, mutational activation of the ras gene is also a characteristic, although less frequent, event in skin carcinogenesis. Both Harvey- (Ha) and Kirsten-ras (Ki-ras) were shown to be mutated in SCCs,36 whereas N-ras mutations were only observed in melanomas.37, 38 Since functionally relevant mutations are restricted to codons 12/13 and 61, we amplified exons 1 and 2 of the Ha-, Ki- and N-ras genes and analyzed the amplification products by direct sequencing. No mutations were found in any of the MCC lines (data not shown). Similarly, 4 of the SCC lines (SCL-I, SCL-II, SCC-13 and MET-1) were wild type for all codons of all 3 ras genes (data not shown). Only SCC-12 cells were mutated, showing a CC[RIGHTWARDS ARROW]TT double base change in codon 12 in 1 allele of the Ha-ras gene. This caused an amino acid exchange from glycine to leucine (Fig. 4).

thumbnail image

Figure 4. Sequence analysis of the Harvey-ras gene codon 12. SCC-12 cells show a UV-indicative heterozygous CC[RIGHTWARDS ARROW]TT double base exchange. The point mutation is labeled by an asterisk. Genomic DNA isolated from blood of a healthy proband served as a control. cDNA sequences are given for the mutated (Mut) and the wild-type (Wt) status.

Download figure to PowerPoint

Codon 27 polymorphism in the Ha-ras gene

Although sequence analysis only revealed 1 mutation in the activating codons of the ras genes, we detected a C[RIGHTWARDS ARROW]T transition in codon 27 of the Ha-ras gene in 5 of the 6 MCC lines and 4 of the 5 SCC lines (Fig. 5a). While 2 MCC- (MCCL-2 and -4) and 3 SCC lines (SCL-I, SCC-12 and SCC-13) exhibited 1 affected and 1 wild-type allele, 3 MCC- (MCCL-1, -5 and -6) and 1 SCC line (MET-1) were homozygous for the affected allele (Table III). Although this nucleotide exchange was at the third position of codon 27 and thus did not cause an amino acid substitution, the frequency of the homozygous status of the affected allele suggested a tumor-related selection for this nucleotide exchange. To investigate this further, we compared the ras patterns from tumors and corresponding blood samples. From the 4 different blood samples available to us, 3 (corresponding to MCCL-2, -4 and -6) revealed the same allelic distribution as the respective tumor cell lines. However, MCCL-5 was homozygous for the codon 27 polymorphism, whereas blood DNA from the same patient was heterozygous (Fig. 5b). In addition, we analyzed DNA from peripheral blood of 14 healthy donors. From these, 8 showed the affected allele. In 6 samples 1 allele and in 2 samples both alleles carried the base exchange (data not shown). Thus in contrast to 50% of the MCC lines, only 14% of the controls were homozygous for this codon 27 polymorphism.

thumbnail image

Figure 5. Sequence analysis of the Harvey-ras gene codon 27. (a) Codon 27 sequences and flanking bases are shown for MCCL-1 to MCCL-6. MCCL-2 and MCCL-4 cells exhibit a heterozygous C[RIGHTWARDS ARROW]T transition at codon 27, while MCCL-1, MCCL-5 and MCCL-6 cells are homozygous for this base exchange. MCCL-2 cells are completely wild-type for codon 27. (b) Codon 27 sequences of cell line MCCL-5 and genomic DNA isolated from blood of the same MCC patient. Whereas the blood DNA shows both the affected and the wild-type allele, the tumor DNA is homozygous for the base exchange. Point mutations are labeled by asterisks. Genomic DNA isolated from blood of a healthy proband served as a control. cDNA sequences are given for the mutated (Mut) and the wild-type (Wt) status.

Download figure to PowerPoint

Alterations of the CDKN2A gene are rare in MCC

A tumor-suppressor gene that is likely to be involved in the development of melanomas than SCCs is the cyclin-dependent kinase inhibitor CDKN2A.39 This gene encodes for 2 distinct proteins, the p16INK4A, which controls passage through the G1 cell cycle checkpoint,40 and the alternatively spliced p19ARF,41 which blocks MDM2 and thereby controls p53 activity.42 Sequence analysis of the CDKN2A gene revealed no base substitution in any of the 6 MCC lines. However, RT-PCR demonstrated lack of expression (lack of the 494 kb band) in the MCCL-1 cells. Although Southern blot analysis did not allow the detection of an abnormally migrating CDKN2A band, nested primer PCR showed no amplification products for exon 2, whereas in all other MCC lines 414 kb fragments were detected. Because of this deletion, expression of CDKN2A was abolished in the MCCL-1 cells (Fig. 6). Thus, our data confirm the recent findings by Cook et al.,43 who suggested that alterations of the CDKN2A gene are rare events in MCC.

thumbnail image

Figure 6. Expression of the CDKN2A gene in MCC lines studied by RT-PCR analysis. Coamplification of CDKN2A and GAPDH cDNA. MCCL-1 cells exhibit normal expression of GAPDH as indicated by the 240 bp GAPDH cDNA band, while the 494 bp CDKN2A cDNA band is only detectable in the control. mRNA isolated from primary keratinocytes of a healthy proband was used as a control for RT-PCR analysis. A 1 kb ladder served as DNA marker.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

It is now well accepted that lifetime accumulation of DNA damage by sunburns is causal for the development of skin SCCs. For Merkel cell carcinomas, which also manifest at sun-exposed sites and develop late in life, the situation is less well established. We, therefore, assumed that a detailed genetic analysis would help to provide further clarification of the role of sun-related damage for MCC development and their relationship to SCCs.

A “fingerprint” for UV-B damage are C[RIGHTWARDS ARROW]T transitions and frequently CC[RIGHTWARDS ARROW]TT double base substitutions in the p53 gene.10 In agreement with this, we found p53 mutations in 4 of the 5 SCC lines, 2 of which were C[RIGHTWARDS ARROW]T transitions. In 1 SCC line even both alleles were affected. One allele exhibited a CC[RIGHTWARDS ARROW]TT double base change in codons 196/197 and the other one a C[RIGHTWARDS ARROW]T transition in codon 197 causing a stop codon. Whereas in most other cancers, e.g., colon cancer, a p53 mutation is generally followed by loss of the second allele,44 mutations in both alleles were already reported earlier for basal cell carcinomas.33 Our current data provide evidence that the same holds true for SCCs. This further substantiates the hypothesis that UV-B radiation gives rise to a mutation frequency high enough to generate even 2 hits in the p53 gene of the same cell. Mutations in the p53 gene seem to play a similar role in the development of MCCs. Two of the 6 MCC lines exhibited UV-B-indicative mutations; 1 affecting codon 278, a hotspot for p53 mutations in skin carcinomas.33 Thus in agreement with previous data, showing a similar mutation frequency for MCCs and MCC lines,23 our data strongly suggest that UV-B is an important mutagen for both SCCs and MCCs.

A second, although less frequent, aberration in SCCs is mutational activation of the Ha- or Ki-ras gene.8, 9, 36 Also here the predominant mutation is a UV-B-specific CC[RIGHTWARDS ARROW]TT double base change in codon 12 of the Ha- or Ki-ras gene. In line with the previous studies, we found a mutation in only 1 of the SCC lines in codon 12 of the Ha-ras gene. The MCC lines did not exhibit mutations in any of the ras genes. Although this is the first report and thus limited by sample number, our data may indicate that activating ras mutations are not important for MCC development. To our surprise, however, we detected a nucleotide exchange in codon 27 of the Ha-ras gene that, similar to the codon 12 mutation, carried a UV signature. This nucleotide exchange was detected in the 4/5 SCC and 5/6 MCC lines. Two findings support the hypothesis that this as yet undescribed polymorphism is linked to tumor progression. First, we detected a high frequency of this polymorphism in the tumor lines and its homozygous state was particularly frequent in the MCC lines (50%) compared to controls (14%). Second, by comparing DNA from MCCL-5 cells with blood from the same patient, we could demonstrate loss of the wild-type allele during tumor development. Future studies now need to confirm this correlation and also to establish functional evidence.

Although this nucleotide substitution does not cause an amino acid exchange (3rd base of codon 27) and, therefore, does not interfere with the function of the protein, we cannot exclude that it may indirectly affect the gene structure. Laken et al.45 recently reported on a germ-line mutation in the APC (adenomatous polyposis coli) gene of Ashkenazi Jews with a family history of colorectal cancer. This mutation created a small hypermutable region, leading to somatic truncating mutations in adjacent sequences. The authors suggested that some of the common polymorphisms occurring in cancer-predisposition genes, even if silent, may in part be responsible for the increased cancer risk by creating such mutation hotspots. Thus, it remains to be elucidated whether this Ha-ras codon 27 polymorphism belongs to this group of mutations and may be a marker for individuals with an increased risk of skin cancer.

Despite rather divergent overall CGH profiles, SCC and MCC shared 2 regions of genomic imbalance—gain of 8q and loss of 4p (summarized in Table IV). Although gain of 8q was detected earlier in MCCs19 and was also seen in malignant melanomas where sun exposure is similarly implicated,46, 47 loss of 4p was not noticed before. Chromosome 8q gains, which most commonly involve the distal part, harbor the c-myc oncogene (mapped to 8q24). Moreover, extra c-myc copies were detected in primary and metastatic melanomas.48 Since c-myc also plays an important role in keratinocyte maturation49 and overexpression in the basal layer of transgenic mice resulted in a hyperproliferative epidermis,50, 51, 52 it is tempting to speculate that c-myc may be the candidate oncogene. Future studies will, therefore, aim to unravel its role in both SCC and MCC development. No obvious candidate tumor-suppressor gene is yet identified for loss of 4p. Using an approach recently applied to determine potential tumor-suppressor genes on chromosome 15,53 we will transfer chromosome 4 into both SCC and MCC cells and thereby determine their functional consequence on tumor growth.

Table IV. Summary of the Most Common Genetic Abnormalities in MCCs, MCC and SCC Lines
Tumors and cell linesGain of 1qGain of 6pLoss of 9pGain of 11qLoss of 4pGain of 8qLoss of 3pExpression of FHIT
  • 1

    Data from SCC line MET-1 have previously been published.30–+, presence of copy number change; −, no copy number change; n.d., not determined.

MCC-1+Aberrant transcript
MCC-5+++No transcript

One type of aberration that obviously differs between MCC and SCC is loss of 3p vs. aberrations in the FHIT gene, the potential tumor-suppressor gene on 3p14.2 (see Table IV). In agreement with our data, loss of 3p was already described earlier for SCCs5, 7, 54, 55 and is also frequently seen in freshly isolated tumors (Popp and Boukamp, unpublished data). The FHIT gene, on the other hand, was only marginally affected (strongly reduced expression due to loss of 1 copy of 3p) in the SCC lines as also confirmed by Rees et al.7 In MCCs, the situation may be different. While recent studies reported on frequent genomic (46%) or allelic (69%) losses of 3p in MCCs,19, 21 the FHIT status was not investigated in the same cells. In our CGH analysis, we only identified 1 tumor with loss of 3p (10%). However, 2 of the 6 MCC lines with normal copies of chromosome 3p exhibited aberrant FHIT transcripts. The FHIT gene encompasses FRA3B, the most active common fragile site in the human genome.56 Thus, it is not surprising that FHIT transcripts were detected, lacking 3 or more exons of the gene.22 Also 1 of our MCC cell lines, MCCL-1, revealed an aberrant FHIT transcript where FRA3B was involved. These cells had an insertion of noncoding FRA3B sequences possibly caused by loss of the FRA3B fragile site. The MCCL-5 line, on the other hand, did not exhibit any FHIT transcript indicating a homozygous deletion. These findings may suggest that loss of 3p and aberrations in the FHIT gene are induced by different mechanisms and are likely to cause independent phenotypic changes. Thus, the FHIT gene may be a relevant tumor suppressor for MCC; however, it does not seem to play a major role in SCC development.

In conclusion, by using a variety of molecular genetic approaches, we were able to dissect similarities as well as differences between the 2 nonmelanoma skin cancer types, the MCCs and the SCCs. Whereas UV-B-specific mutations in the p53 gene and codon 27 of the Ha-ras gene as well as gain of 8q and loss of 4p seem to be common for both tumor types, loss of 3p was characteristically found in the SCC cells. The MCCs, on the other hand, seemed more prone to aberrations in the FHIT gene. For these latter changes, the responsible mechanism as well as the potential carcinogens are still unclear. Oxygen radicals are able to cause DNA strand breaks and as a consequence allow chromosomal breakage—a prerequisite for gains and losses of chromosomal material. There is strong evidence that UV-A is able to generate such damage.57 In addition, we have previously shown that long-term increased temperature, characteristic for areas of severe sunburn, has the same effect.58 Thus, it is tempting to speculate that in addition to UV-B, UV-A and/or increased temperature are also involved in the development of both SCCs and MCCs.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank Ms. E. Tomakidi for her excellent cell culture assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft (Bo 1246/3-1 to PB and Mo 644/2-2 to IM), the Wilhelm Sander-Stiftung (98.013.1), the European Community (QLRT-1999-01084) and the “Verein zur Förderung der Krebsforschung in Deutschland“ (all to PB).


  1. Top of page
  2. Abstract
  6. Acknowledgements