Cancer Research Targets Advances in Breast Cancer Research
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
Cancer Research 68, 122, January 1, 2008. doi: 10.1158/0008-5472.CAN-07-1939
© 2008 American Association for Cancer Research
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Correction (v68,p1245)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sabatino, M.
Right arrow Articles by Wang, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sabatino, M.
Right arrow Articles by Wang, E.

Cell, Tumor, and Stem Cell Biology

Conservation of Genetic Alterations in Recurrent Melanoma Supports the Melanoma Stem Cell Hypothesis

Marianna Sabatino1, Yingdong Zhao2, Sonia Voiculescu1, Alessandro Monaco1, Paul Robbins3, Laszlo Karai4, Brian J. Nickoloff5, Michele Maio6,7, Silvia Selleri1,8, Francesco M. Marincola1 and Ena Wang1

1 Infectious Disease and Immunogenetics Section, Department of Transfusion Medicine, Warren G. Magnuson Clinical Center, 2 Biometrics Research Branch, 3 Surgery Branch, and 4 Pathology Department, National Cancer Institute, NIH, Bethesda, Maryland; 5 Skin Disease Research Program, Loyola University, Chicago Medical Center, Maywood, Illinois; 6 Division of Medical Oncology and Immunotherapy, Department of Oncology, University Hospital of Siena, Istituto Toscano Tumori, Siena, Italy; 7 Cancer Bio-Immunotherapy Unit, Department of Medical Oncology, Centro di Riferimento Oncologico, Aviano, Italy; and 8 Department of Human Morphology, University of Milano, Milan, Italy

Requests for reprints: Ena Wang, NIH, Building 10, Room 1N224B, 10 Center Drive, Bethesda, MD 20892-1184. Phone: 301-451-8501; Fax: 301-402-1360; E-mail: ewang{at}mail.cc.nih.gov.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
It is generally accepted that human cancers derive from a mutated single cell. However, the genetic steps characterizing various stages of progression remain unclear. Studying a unique case of metastatic melanoma, we observed that cell lines derived from metachronous metastases arising over a decade retained a central core of genetic stability in spite of divergent phenotypes. In the present study, we expanded our previous observations comparing these autologous cell lines of clonal derivation with allogeneic ones and correlated array comparative genomic hybridization (aCGH) with gene expression profiling to determine their relative contribution to the dynamics of disease progression. aCGH and gene expression profiling were performed on autologous cell lines and allogeneic melanoma cell lines originating from other patients. A striking correlation existed between total extent of genetic imbalances, global transcriptional patterns, and cellular phenotypes. They did not follow a strict temporal progression but stemmed independently at various time points from a central core of genetic stability best explained according to the cancer stem cell hypothesis. Although their contribution was intertwined, genomic imbalances detectable by aCGH contributed only 25% of the transcriptional traits determining autologous tumor distinctiveness. Our study provides important insights about the dynamics of cancer progression and supports the development of targeted anticancer therapies aimed against stable genetic factors that are maintained throughout the end stage of disease. [Cancer Res 2008;68(1):122–31]


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
It is generally recognized that cancers originate from the clonal expansion of a single mutated founder cell (1). The veracity of this statement is critical because clonal derivation may affect the effectiveness of cancer therapies targeting genetic elements passed on to each generation by parental cells throughout the natural history of the disease. Experimental animal models (2) and clinical observations support this model (3); in melanoma, clonal derivation was documented in the transition from radial to vertical growth (4), in in-transit metastases (5), and in the progression from primary to metastatic disease (6). We recently characterized melanoma cell lines derived from metachronous melanoma metastases in a patient who experienced several recurrences over more than a decade (7). Mutational analysis of β-catenin gene, methylation of the human androgen receptor, karyotyping, and chromosomal comparative genomic hybridization (cCGH) documented a clonal origin of each cell line from a single ancestor. Although all cell lines originated from the same founder, each did not likely derive from the most recent previous metastasis because aberrations in each recurrence were sufficiently different to exclude a purely sequential process. Because subsequent recurrences did not build on the genomic alterations found in the earlier one, it was reasonable to postulate that there must have been a progenitor cell more genetically stable than the cells expanded from individual metastases, supporting the melanoma stem cell postulate (8–10). Moreover, a β-catenin mutation identified in all cell lines suggested that a stable mutation responsible for heightened self-renewal properties through activation of stem cell signaling pathways (11, 12) was carried by all the descendent cells. Interestingly, cells karyotyped from the same culture revealed differences evocative of ongoing chromosomal instability within the cell population isolated from each metastasis (10). It was nevertheless of interest to note that, in spite of the genetic entropy displayed by cells cultured from subsequent metastases, a conserved core of genetic alterations was maintained throughout the natural history of the disease that could have been inherited from the progenitor stem cell or resulted from a process of convergent evolution dictated by the structural organization of the genome. Indeed, shared chromosomal imbalances have been described in melanomas derived from different patients (4, 13–15). In the previous study, we applied cCGH that cannot detect single copy gains or losses involving regions smaller than 5 to 10 Mb. For the detection of smaller genomic imbalances, in the present study, we applied array CGH (aCGH) using a cDNA platform that provides direct evaluation of coding regions and allows parallel comparisons between genomic imbalances and transcriptional patterns (16–20). By using the same microarray platform for both aCGH and cDNA expression studies, Pollack et al. (17) found that changes in gene copy number were consistently associated with corresponding changes in expression of the affected genes. Similar observations were made by Nigro et al. (19) in glioblastoma samples. Most observations were performed at a single time point relevant to either primary or metastatic lesions, although no longitudinal long-term studies have been reported due to the rarity in which long-term recurrences occur in cancer. Thus, we assessed the weight that genetic imbalances bore on transcriptional programs during a period spanning more than a decade. Admittedly, by studying cell lines, information derived from the whole tumor may be missed. Further attempts to rescue DNA from fixed material from which the cell lines originated failed to yield material of sufficient quality for validation of the in vitro findings, limiting the breath of our work to an indirect analysis. However, karyotypic and genetic profile of cell lines are representative of persistent genetic alterations in vivo, as converging phenotypes are unlikely to randomly occur in vitro under steady-state culture conditions (14, 21–25).


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Peripheral blood mononuclear cell preparation. Peripheral blood mononuclear cells (PBMC) were used as a reference for both aCGH and transcriptional analysis. For aCGH, PBMCs from a female donor who received leukapheresis at the Department of Transfusion Medicine were Ficoll gradient separated and used throughout the study. Validation of aCGH accuracy was done by obtaining six additional PBMCs from female and six PBMCs from male donors to confirm stability of gene representation in autosomes and sex-determined imbalances within the X and Y chromosomes. Total RNA obtained from pooled PBMC of six normal healthy donors was extracted, amplified, and used as a reference for transcriptional studies (26, 27).

Cell line origin and culture. Five melanoma cell lines were generated from distinct cutaneous melanoma metastases that progressively appeared in patient 888 during a period spanning longer than a decade (7). The metastases were excised for diagnostic or therapeutic purposes at the Surgery Branch, National Cancer Institute (Bethesda, MD). Other melanoma cell lines were generated from metastases from other patients (28). A375 cell line was purchased from the American Type Culture Collection. Cell lines were maintained in steady-state culture conditions using RPMI 1640 (Biosource International) supplemented with 10% fetal bovine serum (Biosource International). Cell lines were harvested by trypsin/versene (Biosource International) digestion (28). High-resolution HLA typing was performed by sequence-based typing (29, 30).

Array comparative genomic hybridization. Genomic DNA was extracted from short-term cultured tumor cells or PBMCs using Qiagen Mini kit. DNA labeling was conducted using BioPrime aCGH Genomic Labeling kit (Invitrogen; ref. 31). Test and reference DNA were, respectively, labeled with Cy5 (red) and Cy3 (green) and cohybridized to the cDNA microarrays printed at the Infectious Disease and Immunogenetics Section, Department of Transfusion Medicine, Clinical Center, NIH with a configuration of 32 x 24 x 23 and contained 17,500 elements (see Supplementary Data). The entire aCGH data set will be available online.9

Expression analysis. Total RNA was extracted from PBMCs or cell lines using RNA Mini or Midi kit (Qiagen) and amplified into antisense RNA as previously described (27, 32, 33). After amplification, the quality of antisense RNA was tested with Agilent Bioanalyzer 2000 (Agilent Technologies). Total RNA from PBMC pooled from six normal donors was extracted and amplified to serve as constant reference (27, 32). Test and reference RNA were, respectively, labeled with Cy5 (red) and Cy3 (green) and cohybridized to the cDNA microarrays. The entire microarray data set will be available online.10

Data accrual and statistical analysis. Fluorescence intensities generated by Cy5 and Cy3 probes hybridized onto the microarray slides were scanned at 10-µm resolution on a GenePix 4000 scanner (Axon Instruments) at variable photomultiplier tube voltage to obtain maximal signal intensities with <1% probe saturation. The fluorescence intensity and ratio data for Cy5 and Cy3 were initially analyzed using GenePix 4.3 software. Data on log base 2 scale was normalized with Lowess smoother method using BRBArrayTools developed by the Biometric Research Branch, National Cancer Institute.11 Further data preprocessing was done using web tool preP.12 Duplicate data points were merged by taking the average over the UniGene cluster IDs. Missing data were imputed using K-nearest neighbor imputation method with K = 15. Gene location was extracted according to UniGene cluster IDs from Ensemble and University of California at Santa Cruz using IDconverter.13 The preprocessed aCGH data were then segmented using the circular binary segmentation method implemented in ADaCGH14 to detect regions with abnormal DNA copy number. The smooth variables were set as defaults. The significance level for the test to accept change points was set to be 0.01 under 10,000 permutations. The cutoff for gain/loss calls was ±0.12 on the log base 2 scale, which is consistent with others' reports (4, 6, 34–36). Transcriptional data on melanoma cell lines were process by averaging results in replicate experiments for the same genes and aligned with aCGH data. Whole profile comparisons between aCGH and transcriptional data were performed using Spearman rank correlation. Gene annotations were extracted using IDconverter. Hierarchical clustering analysis was performed on BRBArrayTools using centered correlation as similarity metric and centroid linkage method: Cluster 3:0. Heat maps were generated using Eisen's TreeView software. Preferential chromosomal location of genes differentially expressed by patient 888 compared with allogeneic cell lines at different levels of significance (unpaired t test) was evaluated by Fisher's exact test (37) or by applying the BRBArrayTools gene set analysis in which the statistical significance (P value) of a gene set containing a given number (n) of genes is evaluated using the likelihood score (Kolmogorov-Smirnov) statistic permutation test for gene enrichment analysis (38). The test was separately applied to each gene set differentially expressed between patient 888 and the allogeneic cell lines within each chromosome or in areas of imbalances as detailed in Results.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Patient History
Details about the patient history were previously reported (7). Briefly, patient 888 had a congenital mole removed in December 1988. Over a period of more than a decade, the patient experienced several recurrences (Supplementary Table S1); from each one a melanoma cell line was established. Because we previously documented a clonal origin of the cell lines (7), we sought the opportunity to study the dynamics of genetic variation and its relationship with transcriptional changes in this case with such a uniquely prolonged course.

Genomic Imbalances
Definition of chromosomal gains and losses. The threshold for chromosomal region gains and losses was determined comparing female with male PBMC (Fig. 1A–C ). Six sex-matched female and six sex-mismatched male genomic DNAs were individually cohybridized against one normal female genomic DNA that served as a constant reference. A ratio of ±0.12 was determined as the threshold for at least one gene copy gain (+) or loss (–) corresponding to an exclusion of the 99th percentile of autosomal data while eliminating chromosome X ratios under the 15th percentile. Differences in gene copy number for chromosome Y could not be statistically assessed due to the small number of loci from this chromosome spotted in the array chip not allowing for sufficient statistical power.


Figure 1
View larger version (45K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Graphic illustration of aCGH smoothed data from normal donor and melanoma cell lines. A and B, representative examples of aCGH based on genomic DNA from female (A) and male (B) healthy donors cohybridized to a cDNA array platform together with a genomic DNA from a female donor used as constant reference in all experiments performed in this study. Log(2) ratios were plotted on the Y axis for all clones on the array according to their map position (X axis). A ratio of ±0.12 was determined as the threshold for at least one gene copy gain (+) or loss (–) corresponding to an exclusion of the 99th percentile of autosomal data while eliminating chromosome X ratios under the 15th percentile. The colors used for each target sequence are yellow (no significant DNA copy number change) and green (significant DNA copy number loss). C, the complete aCGH data for the 22 autosomes and the X chromosome averaging a series of six female-female and six female-male hybridizations using genomic DNA obtained respectively from six female and six male normal donor PBMCs. D, whole-genome aCGH profile for patient 888 melanoma cell lines. Log(2) ratios were plotted on the Y axis for all clones on the array according to their map position (X axis). A log(2) ratio threshold of 0.12 was used to determine clones gained (+0.12) or lost (–0.12). The colors used for each target sequence-specific dot outline areas of no significant DNA copy number change (yellow), significant DNA loss (green), and significant DNA gain (red).

 
Specificity of chromosomal imbalances in patient 888 melanoma cell lines. The previously defined threshold was used to identify chromosomal imbalances in cell lines derived from patient 888 and other patients (28, 39, 40). The imbalances observed in patient 888 cell lines are portrayed graphically in Fig. 1D and summarized in Table 1 .


View this table:
[in this window]
[in a new window]

  		Table 1. Genomic loss/gain in melanoma cell lines as obtained from aCGH analysis

 
All melanoma cell lines from patient 888 bore genomic imbalances in one or more DNA regions; some of them were similar supporting their common derivation from a unique ancestor. High concordance was observed between the previous data obtained by cCGH (7) and the present analysis based on aCGH (Fig. 2A and B ). A primary cluster of imbalances, including –10p, –17p, +3p, +7, and +17q, was confirmed by this study; linkage clustering applied to cCGH data had originally attributed this cluster to the hypothetical founder stem cell (7). Small deletions in chromosome 9p, 14q, and 16q were consistently seen in patient 888 cell lines by aCGH that had been less consistently seen by cCGH analysis (7). Moreover, unsupervised hierarchical clustering of patient 888 and allogeneic cell lines based on aCGH clearly segregated the former from the latter, strongly supporting patient specificity independent of disease stage (Fig. 2B). However, within patient 888 cluster, a separation was noted between 888-MEL and subsequent cell lines, suggesting that the biggest evolutionary step in the dedifferentiation process occurred early on. Although the global genetic make up of patient 888 cell lines was highly specific, commonalities were observed with allogeneic cell lines best explained by a convergent tendency of some imbalances to occur at higher rates in specific chromosomal regions as a reflection of a general pattern of chromosomal instability of human melanoma. Losses in chromosome 9, 10, and 17 and gains in chromosome 7 were respectively seen in three, five, five, and six of eight allogeneic cell lines. This suggests that these imbalances may bear particular significance in the genesis of melanoma as also observed studying NCI-60 cell lines (14). Interestingly, most of the common alterations identified by our study seem to be peculiar for melanoma as they were not frequently seen among the NCI-60 cell lines of other histology (14).


Figure 2
View larger version (39K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Conservation of a core of genetic alterations and segregation of patient 888 cell lines and others based on aCGH and gene expression profile. A, particular of linkage clustering among distinct genomic imbalances based on the binary, Euclidean distance comparing individual cell lines originating from patient 888. The red lined box includes the primary cluster of imbalances likely to be inherited by the founder stem cell. B, unsupervised hierarchical clustering of patient 888 and allogeneic cell lines based on smoothed aCGH data with ±0.12 threshold cutoff. Significant segments include regions in chromosome 3, 7, 10, and 17 (P < 0.01; see Table 1 for details). C, unsupervised hierarchical clustering of patient 888 and allogeneic cell lines based on gene expression data. Data represent the average of two experiments for each cell line and clustering was based on the complete unfiltered data set. D, multidimensional scaling analysis based on averaged gene expression data (a) and averaged aCGH data (b) for patient 888 (red) and other melanoma cell lines (blue).

 
Other chromosomal aberrations observed frequently in the NCI-60 melanoma cell lines (14) were not seen in patient 888; these included losses in chromosome 4, 11, and 18 and gains in chromosome 20, which were seen in five, five, three, and six of eight allogeneic cell lines. Finally, other chromosomal aberrations described in the NCI-60 melanoma cell lines were not confirmed by this study, including losses in chromosomes 13, 14, 16, and 21 and gains in chromosomes 2 and 3 (14). In summary, although patient 888 cell lines shared genetic traits with other allogeneic cell lines and with historical comparisons with the NCI-60 melanoma cell lines, their global profile portrayed a distinctive fingerprint that persisted throughout the natural history of the disease.

Dynamics of chromosomal alterations in patient 888. Significant losses in chromosomes 1p, 8p, 9p, 10p, 14q, 16q, 17p, and 22q and gains in chromosomes 3p, 7p, 7q, 8q, and 17q were detected in most of patient 888 cell lines (Table 1). Expectedly, the number of genomic regions with statistically determined genetic imbalances increased with time going from a total of 11 in 888-MEL to 23 in 1936-MEL. However, this trend was not consistently maintained because the last cell line 1962-MEL contained only 14 imbalances. Interestingly, this observation paralleled the strongly dedifferentiated phenotype of 1936-MEL, which displayed, among sister cell lines, the most degenerate aCGH profile (Fig. 2B), strong transcriptional deviation (Fig. 2C and D), total loss of pigmentation (Fig. 3B ), and an atypical non–spindle-type morphology in culture (Fig. 3C).


Figure 3
View larger version (50K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Melanoma cell line heterogeneity. A, relative expression of melanoma differentiation antigens by various cell lines. 888-MEL (b) and 1858-MEL (d–f) displayed the highest level of expression compared with 1290-MEL (c), 1962-MEL (g), and 1936-MEL (a), which revealed the lowest level of differentiation. 1936-MEL displayed a phenotype in culture quite different from its sister cell lines (characterized by lack of pigmentation; B) and a rounded morphology (C) in vitro compared with the spindle melanocyte–resembling morphology of the other cell lines.

 
The length of individual genomic imbalances also increased with time. Deletions of chromosome 9p well exemplify the dynamics of the extension of genomic alterations (see Supplementary Fig. S4). Losses of 9p progressively extended throughout the natural evolution of the disease from a limited region spanning 9p21-22, which surrounds the CDKN2A melanoma susceptibility locus (41, 42). This extension, however, did not occur in a strictly sequential fashion, once again emphasizing the relatively independent behavior of individual metastasis progression from the central ancestor. Interestingly, a two-sample t test (P < 0.01) applied to aCGH data to partition cell lines derived from early-stage tumors that responded to immunotherapy (888-MEL and 1290-MEL) from the later unresponsive ones (1858-MEL, 1936-MEL, and 1962-MEL) identified the genomic region around 9p21 as most representative of this watershed stage of the disease. This may be of interest because this area encodes several genes related to melanoma differentiation, immune responsiveness, and prognosis (41–43). Most genetic alterations followed a similar pattern of individual extension around a common center core (Table 1). The sporadic and nonsequential acquisition of additional genetic imbalances is exemplified by gains in chromosome 20q that were observed in 888-MEL, 1936-MEL, and 1962-MEL and in five of eight allogeneic cell lines but were missing in the intermediate sister cell lines.

Transcriptional Profiling
Each cell line used in this study was thawed, cultured, harvested, and processed for mRNA isolation, purification, and hybridization to the cDNA array chips twice. Transcriptional analysis was applied by averaging data from the two independent experiments to account for variability in mRNA expression during in vitro culture. Unsupervised clustering based on the complete data set applying center/centroid linkage analysis identified two main clusters: one containing all the cell lines originated from patient 888 and the other including only allogeneic ones (Fig. 2C). The former included also two allogeneic cell lines (938-MEL and the two 624-MEL clones) that clustered at a relatively close Euclidian distance from patient 888 cell lines. Among patient 888 cell lines, those derived from the same lesion clustered together. 888-MEL clustered closer to 1858-MEL, whereas 1290-MEL clustered at a higher Euclidian distance. Three different cell lines expanded following different procedures from the same 1858 metastasis clustered together. Similarly, two different passages of the 1290-MEL cell line (an early and a later passage) clustered together excluding the occurrence of clonal selection during prolonged cell culture that might have affected their genetic profile (25). The later cell lines 1936-MEL and 1962-MEL that grouped progressively away from the primary cluster still remained transcriptionally closer to the sister cell lines than the allogeneic ones. Strong similarity was observed between the global transcriptional differences observed by unsupervised clustering and the level of differentiation of each cell line based on the transcriptional expression of melanoma differentiation antigens (Fig. 3A); 888-MEL and 1858-MEL displayed the highest level of expression compared with 1290-MEL and the subsequent cell lines (44). As previously mentioned, 1936-MEL displayed a phenotype in culture quite different from its sister cell lines characterized by lack of pigmentation and a rounded morphology in vitro compared with the spindle melanocyte-resembling morphology of the others. The anaplastic behavior of this cell line had also been associated in our previous study (7) with an unusual level of chromosomal instability, high level of triploid/tetraploid chromosomal counts. A tight correlation between transcriptional and protein expression was observed comparing the present data with our previous analysis of melanoma antigen expression in the same cell lines by intracellular fluorescence-activated cell sorting analysis. With the exception of 526-MEL, the expression of the Melan-A and gp100/PMel17 (silv) correlated closely with previously described protein levels (28).

Correlation between Genomic Imbalances and Gene Expression Profiling
Cumulative comparison by Spearman rank correlation of aCGH with transcriptional data showed in several cases a significant or close to significant correlation (Table 2 ).


View this table:
[in this window]
[in a new window]

  		Table 2. Cumulative comparison of aCGH with transcriptional data

 
Analysis of individual segments was performed by identifying genes within each aCGH-defined region of imbalance and performing a two-sample t test in which the average intensity ratios of the genes belonging to the region were compared with the rest of the genes included in the arrays. In several cases, there was a good correlation between individual imbalances and transcriptional data particularly when the region included a sufficiently large number of genes to allow bigger statistical power.

Most correlations were direct because transcriptional changes matched the direction of the imbalance (increased ratios associated with amplifications and decreased with losses). However, few examples of consistent inverse correlation were noted: in particular, losses of chromosome 9 were correlated in patient 888 cell lines with increased average expression of the respective genes. Although this inverse correlation reached statistical significance only in two cell lines (Table 2), the same trend was consistently observed in all patient 888 cell lines and none of the allogeneic lines. Because aCGH is insensitive to structural chromosomal changes, such as translocations, and it represents a virtual gene-by-gene reconstruction of hypothetical chromosomes, this finding cannot be explained by structural rearrangements of the genome but it rather may represent a compensatory mechanism to allelic loss.

To test whether the observed correlations between aCGH and transcription exclusively determined the segregation of the sister cell lines from the allogeneic ones, we then pooled from the complete transcriptional data set genes encoded within the genomic regions of chromosomes 7, 9, 14, and 17 containing imbalances statistically associated with gene expression; these genes served as a basis for reclustering the complete set of autologous and allogeneic cell lines. As expected, this exercise clearly separated patient 888 from allogeneic cell lines. Surprisingly, however, the remaining genes not included in these regions could still segregate the sister cell lines away from allogeneic ones, suggesting that correlations between genetic imbalances and gene expression may participate in determining the genetic identity of each patient's cancer but are not the only factor.

To explain this finding, we tested whether the genes determining patient 888 individuality were randomly distributed across the whole genome or had preferential chromosomal locations. A two-sample t test was applied to identify genes differentially expressed between all of patient 888 melanoma cell lines (independent of stage of disease) and allogeneic cell lines at several levels of significance. Subsequently, a Fisher's exact test was applied in a two-by-two table to test if the number of differentially expressed genes in each chromosome was significantly higher than that of genes not differentially expressed (37). At a {chi}2 value of 3.841 (determined for a P < 0.05 when the degree of freedom is 1), only chromosome 17 displayed a significant enrichment of genes differentially expressed between patient 888 and other patients' cell lines. Similarly, the Kolmogorov-Smirnov statistic permutation test for gene enrichment analysis (38) showed a significance expression bias only in chromosome 17. This finding may reflect the large size of genomic imbalances affecting this compared with other chromosomes.

We then measured the relative contribution of genes contained in areas of genetic imbalance detectable by aCGH to the global transcriptional identity of autologous cell lines. For this purpose, regions with consistent chromosomal imbalances across the patient's history were defined those statistically significant in at least three of the patient cell lines (Table 1). The genes included in all these regions were pooled and tested against the rest of the gene expression data set. We calculated for each of the two so-defined groups the number of genes differentially expressed between patient 888 and the allogeneic cell lines at different level of significance (<0.001, <0.01, and <0.05, unpaired two-tailed t test). At all stringency levels, regions of imbalance were enriched with differentially expressed genes compared with the rest of the genome ({chi}2 value of 22.5, 20.5, and 11.7 for the three levels of t test significance for a {chi}2 P < 0.001 in all cases; Table 3A ). At decreasing stringency, the difference in the proportion of genes differentially expressed between patient 888 cell lines and allogeneic cell lines in the two regions decreased (4% versus 2% at a t test P < 0.001, 12% versus 9% at P < 0.01, and 25% versus 21% at P < 0.05). Similar results were obtained with the BRBArrayTools gene enrichment module (P < 0.001; Table 3B). Importantly, although the regions of imbalance were enriched with differentially expressed genes, they accounted only for a small proportion of the transcriptional signatures determining patient 888 cell line identity. In all cases, the total number of differentially expressed genes was lower compared with regions not containing imbalances and their contribution was limited to 26.8%, 20.9%, and 17.9% at the respective levels of significance. Thus, our study suggests that expression biases due to genetic imbalances contribute about 20% to 25% of the genetic fingerprint of individual patients' tumors and this balance persists throughout the end stage of disease.


View this table:
[in this window]
[in a new window]

  		Table 3. Expected over observed enrichment of genes determining patient 888 identity in regions of genomic imbalances

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
The natural history of cancer is likely determined by the accumulation in time of genetic alterations within a cell lineage (45). However, the recently proposed cancer stem cell hypothesis fine tuned this belief, suggesting that such alterations do not follow a linear progression from one tumor to the next but rather, like branches from a tree, stem independently from a common progenitor cell. This progenitor stem cell is responsible for preserving the genetic traits necessary for long-term survival, whereas its progeny follows an independent pattern of dedifferentiation in adaptation to tissue-specific requirements.

Chromosomal aberrations have been described in human malignant melanoma using conventional cytogenetic methods (4) and associated with candidate predisposition loci (13, 15, 41, 42, 46–49). Most of the information is representative of a single time point, whereas little is known about the evolution of genetic alterations in time and how this may affect the evolving phenotype of cancer cells. This is due to the rarity in which long-term analysis of recurrent disease can be performed in this rapidly killing cancer. The model used here allowed the assessment of the evolution of metastatic melanoma throughout a decade of recurrences (7), confirming that a direct correlation exists between genetic background and transcriptional profile. This correlation is specific to each imbalance and does not change with time. However, such correlation could not be consistently predicted for all imbalances and, on occasion, inverse correlations could be observed for which a good explanation is lacking. Most importantly, transcriptional patterns related to genetic imbalances detectable by aCGH accounted only for a limited proportion (20-25%) of the functional signatures that determined the individuality of this patient's cancer natural history.

This study suggests that, in spite of a certain degree of entropic degeneracy restricted to each cell line, a strong core of genetic stability dominated the genetic make up of this tumor along the 12-year period that characterized her metastatic history. This stability insensitive to budding variations supports the cancer stem cell hypothesis (10). In this patient's case, the disease likely started with a relatively well-differentiated tumor expressing high levels of melanoma differentiation antigens, a melanocyte-like phenotype in culture, and limited genetic alterations (888-MEL). Whether this characteristic led the tumor immune responsiveness to treatment with a tumor-infiltrating lymphocyte (TIL) recognizing one of these antigens (tyrosinase) is impossible to test in a single case study. The second lesion, 1290-MEL, remarkably lost the expression of most tumor differentiation antigens, the melanocytic appearance in in vitro culture, and was genetically and transcriptionally less differentiated than 888-MEL and the subsequent cell line. Yet, it maintained the ability to respond to treatment with a TIL population reactive against a mutated protein not associated with the melanocytic lineage but likely essential for cancer cell survival (β-catenin; ref. 11). 1858-MEL was genetically and phenotypically closest to 888-MEL, although deriving from a metastasis that appeared a decade later and did not respond to immunotherapy. It is likely that the phenotype of these cell lines is representative of in vivo conditions as three independent cell lines prepared and maintained close proximity to each other after long-term culture. In spite of this high degree of differentiation, 1858-MEL completely lost surface expression of HLA class I molecules because of a genomic loss of the 15q21.1-22.1 region where the β2-microglobulin locus is situated.15 This represents an emblematic example of a molecularly defined escape mechanism that may play a dominant role in determining responsiveness to treatment (50, 51).

In spite of this phenotypic diaspora, a central core of genetic alterations was maintained through the end of the patient's life (Fig. 2A). It is possible that within this conserved core of genetic stability rests the interpretation of this patient's fate and the Achilles' heel of her cancer. Indeed, we believe that this in vitro documentation is representative of the evolution of this patient tumor. It is unlikely that the consistent alterations documented here were due to converging genetic artifacts related to in vitro culture because patient 888 cell lines maintained their individuality compared with allogeneic lines also grown in vitro for a long time.

In spite of the limited effect (25%) that genetic imbalances have on the transcriptional pattern of this patient's cancer, their global patterns correlated well with mRNA patterns and phenotypes; cell lines displaying enhanced level of genetic degeneracy (such as 1936-MEL) displayed also most severe transcriptional, translational, and phenotypic changes. This suggests that downstream cellular functions that cannot be directly accounted for by chromosomal alterations may be indirectly related through modulation of regulatory mechanisms, such as methylation patterns, mutational analysis, and characterization of microRNA.

In summary, the study of chromosomal imbalances based on aCGH confirmed our previous observations (7). Moreover, this study suggests that (a) both DNA- and mRNA-determined patterns influenced the individuality of this patients' tumors; (b) the individuality was maintained through the end stages of disease; (c) although their contribution is intertwined, DNA and mRNA are independent factors with genomic imbalances detectable by aCGH, contributing only 25% of the transcriptional traits determining autologous tumors singularity; (d) yet a striking correlation existed between total extent of genetic imbalances, global transcriptional patterns, and cellular phenotypes; and (e) genetic imbalances, transcriptional patterns, and phenotypes did not follow a strict temporal progression but stem independently at various time point from a central core of genetic stability best explained according to the cancer stem cell hypothesis. Thus, the present study may provide important insights about the dynamics of cancer progression and support the development of targeted anticancer therapies aimed against stable genetic factors maintained through the end stage that determine the individuality of each patient's disease.


   Acknowledgments
 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


   Footnotes
 
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

9 Gene Expression Omnibus series 8777, http://www.ncbi.nlm.nih.gov/geo/. Back

10 Gene Expression Omnibus series 8778, http://www.ncbi.nlm.nih.gov/geo/. Back

11 http://linus.nci.nih.gov/BRB-ArrayTools.html Back

12 http://prep.bioinfo.cnio.es/ Back

13 http://idconverter.bioinfo.cnio.es/ Back

14 http://adacgh.bioinfo.cnio.es/ Back

15 Manuscript in preparation. Back

Received 5/24/07. Revised 8/16/07. Accepted 10/16/07.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Sidransky D. Screening for clonal genetic alterations. Eur J Cancer 1995;31A:1127–9.[CrossRef]
  2. Iannaccone PM, Weinberg WC, Deamant FD. On the clonal origin of tumors: a review of experimental models. Int J Cancer 1987;39:778–84.[Medline]
  3. Fang D, Nguyen TK, Leishear K, et al. A tumorigenic subpopulation with stem cell properties in melanomas. Cancer Res 2005;65:9328–37.[Abstract/Free Full Text]
  4. Bastian BC, LeBoit PE, Hamm H, Brocker EB, Pinkel D. Chromosomal gains and losses in primary cutaneous melanomas detected by comparative genomic hybridization. Cancer Res 1998;58:2170–5.[Abstract/Free Full Text]
  5. Wiltshire RN, Dennis TR, Sondak VK, Meltzer PS, Trent JM. Application of molecular cytogenetic techniques in a case study of human cutaneous metastatic melanoma. Cancer Genet Cytogenet 2001;131:97–103.[CrossRef][Medline]
  6. Balazs M, Adam Z, Treszl A, Begany A, Hunyadi J, Adany R. Chromosomal imbalances in primary and metastatic melanomas revealed by comparative genomic hybridization. Cytometry 2001;46:222–32.[CrossRef][Medline]
  7. Wang E, Voiculescu S, Le Poole IC, et al. Clonal persistence and evolution during a decade of recurrent melanoma. J Invest Dermatol 2006;126:1372–7.[CrossRef][Medline]
  8. Rasheed S, Mao Z, Chan JM, Chan LS. Is melanoma a stem cell tumor? Identification of neurogenic proteins in trans-differentiated cells. J Transl Med 2005;3:14.[CrossRef][Medline]
  9. Grichnik JM, Burch JA, Schulteis RD, et al. Melanoma, a tumor based on a mutant stem cell? J Invest Dermatol 2006;126:142–53.[CrossRef][Medline]
  10. Grichnik JM. Genomic instability and tumor stem cells. J Invest Dermatol 2006;126:1214–6.[CrossRef][Medline]
  11. Rubinfeld B, Robbins P, el Gamil M, Albert I, PorfiriE, Polakis P. Stabilization of β-catenin by genetic defects in melanoma cell lines. Science 1997;275:1790–2.[Abstract/Free Full Text]
  12. Reya T, Clevers H. Wnt signalling in stem cells and cancer. Nature 2005;434:843–50.[CrossRef][Medline]
  13. Nelson MA, Radmacher MD, Simon R, et al. Chromosome abnormalities in malignant melanoma: clinical significance of nonrandom chromosome abnormalities in 206 cases. Cancer Genet Cytogenet 2000;122:101–9.[CrossRef][Medline]
  14. Roschke AV, Tonon G, Gehlhaus KS, et al. Karyotypic complexity of the NCI-60 drug-screening panel. Cancer Res 2003;63:8634–47.[Abstract/Free Full Text]
  15. Curtin JA, Fridlyand J, Kageshita T, et al. Distinct sets of genetic alterations in melanoma. N Engl J Med 2005;353:2135–47.[Abstract/Free Full Text]
  16. Hyman E, Kauraniemi P, Hautaniemi S, et al. Impact of DNA amplification on gene expression patterns in breast cancer. Cancer Res 2002;62:6240–5.[Abstract/Free Full Text]
  17. Pollack JR, Sorlie T, Perou CM, et al. Microarray analysis reveals a major direct role of DNA copy number alteration in the transcriptional program of human breast tumors. Proc Natl Acad Sci U S A 2002;99:12963–8.[Abstract/Free Full Text]
  18. Furge KA, Dykema KJ, Ho C, Chen X. Comparison of array-based comparative genomic hybridization with gene expression-based regional expression biases to identify genetic abnormalities in hepatocellular carcinoma. BMC Genomics 2005;6:67.[CrossRef][Medline]
  19. Nigro JM, Misra A, Zhang L, et al. Integrated array-comparative genomic hybridization and expression array profiles identify clinically relevant molecular subtypes of glioblastoma. Cancer Res 2005;65:1678–86.[Abstract/Free Full Text]
  20. Pinkel D, Albertson DG. Array comparative genomic hybridization and its applications in cancer. Nat Genet 2005;37 Suppl:S11–7.[CrossRef][Medline]
  21. Macville M, Schrock E, Padilla-Nash H, et al. Comprehensive and definitive molecular cytogenetic characterization of HeLa cells by spectral karyotyping. Cancer Res 1999;59:141–50.[Abstract/Free Full Text]
  22. Masramon L, Ribas M, Cifuentes P, et al. Cytogenetic characterization of two colon cell lines by using conventional G-banding, comparative genomic hybridization, and whole chromosome painting. Cancer Genet Cytogenet 2000;121:17–21.[CrossRef][Medline]
  23. Roschke AV, Stover K, Tonon G, Schaffer AA, Kirsch IR. Stable karyotypes in epithelial cancer cell lines despite high rates of ongoing structural and numerical chromosomal instability. Neoplasia 2002;4:19–31.[CrossRef][Medline]
  24. Kawai K, Viars C, Arden K, Tarin D, Urquidi V, Goodison S. Comprehensive karyotyping of the HT-29 colon adenocarcinoma cell line. Genes Chromosomes Cancer 2002;34:1–8.[CrossRef][Medline]
  25. Vogl A, Sartorius U, Vogt T, et al. Gene expression profile changes between melanoma metastases and their daughter cell lines: implication for vaccination protocols. J Invest Dermatol 2005;124:401–4.[CrossRef][Medline]
  26. Jin P, Zhao Y, Ngalame Y, et al. Selection and validation of endogenous reference genes using a high throughput approach. BMC Genomics 2004;5:55.[CrossRef][Medline]
  27. Wang E. RNA amplification for successful gene profiling analysis. J Transl Med 2005;3:28.[CrossRef][Medline]
  28. Cormier JN, Panelli MC, Hackett JA, et al. Natural variation of the expression of HLA and endogenous antigen modulates CTL recognition in an in vitro melanoma model. Int J Cancer 1999;80:781–90.[CrossRef][Medline]
  29. Adams SD, Barracchini KC, Chen D, et al. Ambiguous allele combinations in HLA class I and class II sequence-based typing: when precise nucleotide sequencing leads to imprecise allele identification. J Transl Med 2004;2:30.[CrossRef][Medline]
  30. Adams S, Robbins FM, Chen D, et al. HLA class I and II genotype of the NCI-60 cell lines. J Transl Med 2005;3:11.[CrossRef][Medline]
  31. Lieu PT, Jozsi P, Gilles P, Peterson T. Development of a DNA-labeling system for array-based comparative genomic hybridization. J Biomol Tech 2005;16:104–11.[Medline]
  32. Wang E, Miller L, Ohnmacht GA, Liu E, Marincola FM. High fidelity mRNA amplification for gene profiling using cDNA microarrays. Nat Biotechnol 2000;17:457–9.
  33. Feldman AL, Costouros NG, Wang E, et al. Advantages of mRNA amplification for microarray analysis. Biotechniques 2002;33:906–14.[Medline]
  34. Balazs M, Carroll P, Kerschmann R, Sauter G, Waldman FM. Frequent homozygous deletion of cyclin-dependent kinase inhibitor 2 (MTS1, p16) in superficial bladder cancer detected by fluorescence in situ hybridization. Genes Chromosomes Cancer 1997;19:84–9.[CrossRef][Medline]
  35. Tirkkonen M, Tanner M, Karhu R, Kallioniemi A, Isola J, Kallioniemi OP. Molecular cytogenetics of primary breast cancer by CGH. Genes Chromosomes Cancer 1998;21:177–84.[CrossRef][Medline]
  36. Urzua U, Frankenberger C, Gangi L, Mayer S, Burkett S, Munroe DJ. Microarray comparative genomic hybridization profile of a murine model for epithelial ovarian cancer reveals genomic imbalances resembling human ovarian carcinomas. Tumour Biol 2005;26:236–44.[CrossRef][Medline]
  37. Gargalovic PS, Imura M, Zhang B, et al. Identification of inflammatory gene modules based on variations of human endothelial cell responses to oxidized lipids. Proc Natl Acad Sci U S A 2006;103:12741–6.[Abstract/Free Full Text]
  38. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50.[Abstract/Free Full Text]
  39. Cormier JN, Hijazi YM, Abati A, et al. Heterogeneous expression of melanoma-associated antigens (MAA) and HLA-A2 in metastatic melanoma in vivo. Int J Cancer 1998;75:517–24.[CrossRef][Medline]
  40. Wang Z, Marincola FM, Rivoltini L, Parmiani G, Ferrone S. Selective human leukocyte antigen (HLA)-A2 loss caused by aberrant pre-mRNA splicing in 624MEL28 melanoma cells. J Exp Med 1999;190:205–15.[Abstract/Free Full Text]
  41. Ohta M, Berd D, Shimizu M, et al. Deletion mapping of chromosome region 9p21-p22 surrounding the CDKN2 locus in melanoma. Int J Cancer 1996;65:762–7.[CrossRef][Medline]
  42. Gursky S, Olopade OI, Rowley JD. Identification of a 1.2 Kb cDNA fragment from a region on 9p21 commonly deleted in multiple tumor types. Cancer Genet Cytogenet 2001;129:93–101.[CrossRef][Medline]
  43. Mandruzzato S, Callegaro A, Turcatel G, et al. A gene expression signature associated with survival in metastatic melanoma. J Transl Med 2006;4:50.[CrossRef][Medline]
  44. Wang E, Panelli MC, Zavaglia K, et al. Melanoma-restricted genes. J Transl Med 2004;2:34.[CrossRef][Medline]
  45. Vogelstein B, Kinzler KW. The multistep nature of cancer. Trends Genet 1993;9:138–41.[CrossRef][Medline]
  46. Brose MS, Volpe P, Feldman M, et al. BRAF and RAS mutations in human lung cancer and melanoma. Cancer Res 2002;62:6997–7000.[Abstract/Free Full Text]
  47. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer. Nature 2002;417:949–54.[CrossRef][Medline]
  48. Pirker C, Holzmann K, Spiegl-Kreinecker S, et al. Chromosomal imbalances in primary and metastatic melanomas: over-representation of essential telomerase genes. Melanoma Res 2003;13:483–92.[CrossRef][Medline]
  49. Garraway LA, Widlund HR, Rubin MA, et al. Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature 2005;436:117–22.[CrossRef][Medline]
  50. Ferrone S, Marincola FM. Loss of HLA class I antigens by melanoma cells: molecular mechanisms, functional significance and clinical relevance. Immunol Today 1995;16:487–94.[CrossRef][Medline]
  51. Marincola FM, Jaffe EM, Hicklin DJ, Ferrone S. Escape of human solid tumors from T cell recognition: molecular mechanisms and functional significance. Adv Immunol 2000;74:181–273.[Medline]



This article has been cited by other articles:


Home page
 Cancer Res.Home page
Correction: Melanoma Stem Cell Hypothesis
Cancer Res., February 15, 2008; 68(4): 1245 - 1245.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Correction (v68,p1245)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sabatino, M.
Right arrow Articles by Wang, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sabatino, M.
Right arrow Articles by Wang, E.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online