Delineating Chromosomal Breakpoints in Radiation-Induced Papillary Thyroid Cancer

Recurrent translocations are well known hallmarks of many human solid tumors and hematological disorders, where patient- and breakpoint-specific information may facilitate prognostication and individualized therapy. In thyroid carcinomas, the proto-oncogenes RET and NTRK1 are often found to be activated through chromosomal rearrangements. However, many sporadic tumors and papillary thyroid carcinomas (PTCs) arising in patients with a history of exposure to elevated levels of ionizing irradiation do not carry these known abnormalities. We developed a rapid scheme to screen tumor cell metaphase spreads and identify candidate genes of tumorigenesis and neoplastic progression for subsequent functional studies. Using a series of overnight fluorescence in situ hybridization (FISH) experiments with pools comprised of bacterial artificial chromosome (BAC) clones, it now becomes possible to rapidly refine breakpoint maps and, within one week, progress from the low resolution Spectral Karyotyping (SKY) maps or Giemsa-banding (G-banding) karyotypes to fully integrated, high resolution physical maps including a list of candiate genes in the critical regions.


Introduction
It is becoming increasingly clear that the pathogenesis of radiation-induced tumors is often distinctly different from that of spontaneous, non-radiation-induced tumors. Our research focuses on the physical mapping of proto-oncogenes related to tumorigenesis such as the neurothrophic growth factor receptor 1, NTRK1 (also known as trk-A) [1], the development of assays to detect chromosomal rearrangements leading to activation of oncogenes [2][3][4][5], and the mapping of translocation breakpoints in spontaneous cases of PTC as well as tumors in patients with a known history of either therapeutic or accidental exposure to ionizing radiation [6][7][8].
Fluorescence in situ hybridization is one of the most powerful tools to detect these genetic aberrations underlying the expression of chimaeric proteins [23][24][25]. Such proteins alter the signaling pathways in cells that have undergone neoplastic transformation [26][27][28]. Table 1 gives an overview of the most relevant RET/PTC rearrangements analyzed and described to date. Table 1. RET/papillary thyroid carcinoma (PTC) rearrangements.
Despite a high prevalence of mutations or rearrangements activating the rtks NTRK1 or RET, many phenotypically similar tumors do not show this abnormality. Adding a further level of complexity, in our studies of post-Chernobyl cases of PTC, only few tumors showed clonal abnormalities in 100% of metaphase spreads like the case S96T ( Figure 1) [7,8,25]. We hypothesize that these normal-looking metaphase spreads carry small submicroscopic lesions also known as cryptic translocations that are missed by the conventional methods of metaphase cell analysis, i.e., G-banding, whole chromosome painting (WCP) or SKY [8,[41][42][43].
Therefore, we feel that it is necessary to combine a variety of cytogenetic techniques to comprehensively describe all relevant aberrations. To further explore this, we utilized cell lines established from three cases of radiation-induced childhood thyroid cancer: S96T, as mentioned above, and S47T and S48T, as analyzed further in this publication. Table 2 gives an overview over clinical details and findings from G-banding and FISH studies in these cell lines. The RET/NTRK1 status of the cell lines used in this study has been published previously [17,34,45]. Now, if these oncogenic events arise from balanced intra-or interchromosomal rearrangements, gene copy numbers remain unchanged compared to normal diploid cells, and comparative genomic hybridization assays using either metaphase spreads [46], oligonucleotide (Nimblegen; Affymetrix) or bacterial artificial chromosome arrays [47,48] will fail to detect the abnormalities.
An additional complication in the definition of candidate genes for thyroid tumorigenesis is the great variety in levels of heterogeneity found in primary cell cultures and even established cell lines. Figure 2 illustrates this by presenting the results of our SKY analysis of case S47T, a childhood case of post-Chernobyl PTC [8]. Roughly half of the S47T metaphase spreads that we analyzed by SKY showed a balanced, reciprocal translocation t(5;7)(q23;p15). The other spreads did not show chromosome 7 material translocated to the der(5) (Figure 2, insert).
Similar challenges have been identified in previous publications analyzing PTC-associated rearrangements with and without exposure to ionizing radiation.
Thus, we have to accept that no single cytogenetic technique will reliably detect all potential aberrations found in the pathogenesis of radiation-induced (or indeed spontaneous) tumors.
In this communication we propose an algorithm utilizing a combination of cytogenetic techniques of increasing resolution to comprehensively, expeditiously and cost-effectively delineate chromosomal breakpoints in radiation-induced papillary thyroid carcinomas. By utilizing publicly available resources, our aim was the development of a replicable, targeted approach to breakpoint analysis which can be used by non-specialist laboratories worldwide. Figure 2. Spectral Karyotype analysis of the PTC cell line S47T. The arrowheads point at the abnormal chromosomes derived from the t(5;7)(q23;p15). The insert shows derivative chromosomes from a metaphase spread that did not show chromosome 7 material on the der(5).

Results and Discussion
Where significant heterogeneity is observed in cultured cell lines, such as in the case of S47T, the possibility of contamination has to be considered. However, we exclude the possibility of a contamination of these 2 cell lines (S47T and S96T) based on the fact that all 10 out of 10 G-banded metaphases showed the identical translocation (Table 2). Therefore, the fact that individual metaphase spreads prepared from S47T showed two different der(7) chromosomes in subsequent passages of S47T must be due to a deletion event that followed the reciprocal t(5;7) translocation.
Instead of immunofluorescence characterization of cell lines, we performed comprehensive cDNA hybridization experiments. This elucidated DNA changes not visible by SKY or G-banding techniques. Results from these studies have been published [7,17,45].
To develop and validate our algorithm, we focused our attention on cell line S48T. Extensive G-banding analysis performed in the laboratories in Munich had indicated that primary cultures derived from case S48T carried multiple chromosomal abnormalities. The rearrangements were large in number and mostly unbalanced, which greatly complicated conventional karyotyping based on G-banding analysis ( Figure 3) [49]. The cloning of cell line S48T has been described previously [42].
Our Spectral Karyotyping analysis (SKY), shown in Figure 3 below the G-banding results, provided some additional clues to the origin of marker chromosomes.
Cell line S48T did not display signs of rearranged chromosomes 10, but a number of marker chromosomes carrying material from either chromosome 1 or 9 caught our attention. The long arm of chromosome 1 harbors the neurotrophic growth factor receptor kinase-1 (NTRK1) gene [1], which has been reported to be aberrantly expressed in various solid tumors among them post-Chernobyl PTC [9,50]. In all S48T metaphase spreads, we found several marker chromosomes containing genetic material from either chromosome 1 or 9. These common markers, three of which are derived from chromosome 1 ( Figure 4A) and four types derived from chromosome 9 ( Figure 4B), are shown in Figure 4.
Protein tyrosine kinases have been implicated in tumor initiation and progression [51][52][53]. In gene expression studies reported elsewhere, we were able to demonstrate that cell line S48T expresses the tyrosine kinase domain of NTRK-1 [44], which is normally located on the long arm of chromosome 1, band q12-21 [1] at position 156,830,671-156,851,642 bp in the UC Santa Cruz (UCSC) genome browser. For the analysis of chromosome 1 rearrangements, we pooled three individual BAC probes, since this has resulted in more reliable FISH signals [45,54,55]. Hybridization of a combination of a biotinylated probe DNA pool that maps close to NTRK1 at chromosome 1q12-21 (clones RP11-37N10, RP11-71P2 and RP11-315I20) and a digoxigenin-labeled probe pool comprised of probes RP11-262A11, RP11-299D6 and RP11-243J18 that bind close to non-muscle tropomyosin 3 (TMP3) (UCSC position 1: 154,127,780-154,155,725), a known translocation partner of NTRK1 in solid tumor cell lines [50,56], revealed complex translocation and genome amplification in line S48T ( Figure 5). Two derivative chromosomes each carried 1 copy of the ~10 Mbp region flanked by our probe pools (arrowheads in Figure 5), while a large marker chromosome contained about 2.5 copies (arrow in Figure 5).
The results shown in Figure 5 confirm comparative genomic hybridization results that indicated genomic amplification of the proximal long arms of chromosome 1 and chromosome 9 in S48T [42].  The abnormal staining pattern of the large marker chromosome (arrows in Figure 6 A,B) prompted us to investigate the distribution of centromeric heterochromatin in this cell line. Considered a rather rare event, some of the S48T metaphase spreads hybridized with the alpha satellite DNA probe showed not just one, but two large dicentric chromosomes (Figure 7, arrows).  Our strategy to rapidly map chromosomal breakpoints in metaphase spreads is based on hybridization of increasingly smaller BAC-derived DNA probe pools. Figure 8 shows chromosome 9-specific examples: the top in ( Figure 8A, B) shows the results obtained with normal metaphase chromosomes, whereas the bottom shows chromosomes in S48T. (B-C) Chromosome arm probes (CAP) provide first clues to the origin of markers. The arrow in the S48T metaphase in (C) points to the small insertion; (D) Chromosomal rainbow probes for chromosome 9 (CRB9) allowed us to narrow down the origin of the inserted material to chromosome 9, pools 10-11 (right) [42].
It should be noted that BAC-FISH is a very sensitive approach to detect translocations [57]. A single BAC clone is sufficient to highlight a small translocation as shown in the example in Figure 9. Here, the BAC clone set contained one sub-telomeric clone that had been assigned by mistake to chromosome 9ptel in one of the databases. As the hybridization experiments showed, this clone maps to the telomere on the short arm of chromosome 8 instead (Figure 9). . BAC-FISH results suggest a detection-sensitivity in the order of single BAC clones or translocated genomic regions in the order of a few hundred kb. The yellow arrows point at the signal generated by a chromosome 8ptel-specific BAC clone that was cohybridized with the chromosome 9 specific BAC CAP probe sets.
Once a minimal breakpoint interval defined by a single BAC clone or a contig of 2-3 clones is defined, genome databases can be consulted to search for candidate tumor-related genes. For the small insertion into the t(8;9;15) chromosome in S48T this approach is illustrated in Figure 10. This screen dump from the Genome browser web page at the University of California, Santa Cruz (UCSC), shows a region of roughly 1.5 Mbp, which was found inserted into the marker chromosome. Clones that were used in our hybridization experiments are included in the set of FISH mapped clones shown in this Figure (i.e., RP11-92C4, RP11-91D7) Interestingly, this region from the long arm of chromosome 9 contains the tumor growth factor (TGF) beta receptor 1 (TGFBR1) gene, which when mutated or duplicated, alters the transmission of the subcellular TGF beta signal and has been reported to cause a dominant disease phenotype [48,58]. While these findings do not support a notion that TGF beta duplications have a causal relationship to post-Chernobyl PTC, the observed gain might very well alter the cells" phenotype increasing their chances of survival and increased proliferation in the tumor microenvironment. Conversely, this metabolomic change might become a tumor"s Achilles heel in efforts to devise more efficient anti-tumor therapies.

Cell Cultures and Preparation of Metaphase Spreads
Normal human control metaphase spreads were made from phytohemagglutinin-stimulated short-term lymphocyte cultures of blood obtained from a healthy male according to the procedure described by Harper and Saunders [59]. Acetic acid-methanol fixed lymphocytes were dropped on ethanol-cleaned slides in a CDS-5 Cytogenetic Drying Chamber (Thermatron Industries, Inc, Holland, MI) at 25 °C and 45-50% relative humidity.
The PTC cultures were established as described by Lehmann et al. and Zitzelsberger et al. [7,8]. All procedures followed protocols approved by the LBNL/UC Berkeley Institutional Review Board (IRB) Committee on protection of Human Subjects in Research regarding use of surplus surgical tissues for research. S48T lines were obtained from the tumor tissue of a 14 year old patient (7 years at time of exposure to elevated levels of radiation) undergoing surgery at the Center for Thyroid Tumors in Minsk, Belarus, following the diagnosis of Hashimoto"s thyroiditis and PTC. Initial chromosome preparations were carried out after an in vitro culture of S48T cells for 8-21 days. Later, clones were isolated by limiting dilution and cultured for more than 20 passages. After G-banding with Wright"s staining solution, karyotypes were recorded according to the International System for Human Cytogenetic Nomenclature [60].

Comparative Genomic Hybridization (CGH)
Comparative genomic hybridization [46] with DNA isolated from the primary culture as well as cell lines established from case S48T was performed following standard procedures as described for a case S42T [6]. In brief, genomic DNA was isolated from the primary culture as well as from cell lines and labeled with biotin-16-dUTP (Roche Applied Science, Indianapolis, IN, USA). Normal female reference DNA was isolated from peripheral lymphocytes of a healthy donor and labeled with dig-11-dUTP. After hybridization to normal metaphase spreads of a healthy donor, labeled DNA probes were detected with streptavidin-Cy2 or avidin DCS-FITC (Vector Inc., Burlingame, CA, USA) and anti-digoxigenin-Cy3/rhodamine conjugates. Slides were counterstained with 4",6diamidino-2-phenyl-indole (DAPI, Calbiochem, La Jolla, CA, USA) for chromosome identification. For CGH analysis, eight or more metaphases were analyzed. Averaged profiles were generated by CGH analysis software (Vysis, Downers Grove, IL, USA) from 10-15 homologous chromosomes and interpreted according to published criteria [61,62].

Spectral Karyotyping Analysis (SKY)
Spectral Karyotyping is a molecular cytogenetic procedure to screen the entire human genome for interchromosomal translocations by hybridization of 24 different WCP probes mixtures to metaphase spreads. We applied SKY to case S48T and identified complex aberration patterns [8]. The SKY analyses followed essentially the recommendations of the manufacturer of the reagents and the SKY imaging instrumentation (Applied Spectral Imaging (ASI), Carlsbad, CA). Briefly, fixed cells on slides were pretreated with 50 µg/mL pepsin (Amresco, Solon, OH) in 0.01N HCl for 10 min at 37 °C before immersion in phosphate buffered saline (PBS) for 5 min. The slides were then incubated in paraformaldehyde (PFA) solution (1% in PBS) for 5 min, then in PBS for 5 min. After immersion in a 70%, 80%, 100% ethanol series for 3-5 min each step, the slides were air dried. Cells on slides were denatured for 5 min at 76 °C in 70% formamide (FA)(Invitrogen, Carlsbad, CA, USA)/2  SSC and then dehydrated in 70%, 80%, and 100% ethanol (2 min per step) before air drying.

Preparation of Locus-Specific DNA Probes (LSPs)
Our procedures for preparation of DNA probes from BAC/PAC clones [64,65] have been described in detail before [1,66,67]. Prior to the chromosome 9-specific FISH studies, 151 BAC clones from the Sanger Center 1 Mbp set [47] were re-arrayed on two 96-well microtiter plates (Table 3). Using information in publicly available databases (http://genome.ucsc.edu/ and http://www.ncbi.nlm.nih.gov/ gquery/gquery.fcgi), we selected additional BAC clones for the long arm of chromosome 1 from the Roswell Park Cancer Institute (RPCI) library RP11 [68] and for chromosome 9. A subtelomeric clone placed in position A1 on Plate 1, GS1-41L13, is not shown in Table 3. This BAC maps to the short arm of chromosome 8 ( Figure 9).    Individual clones were arranged so that the entire chromosome 9-specific clone set was contained on two 96-well plates in 15 rows termed "pools" with 9-12 clones per pool in individual wells ( Figure 11). This created pools "9-1" to "9-15", each of which covers a few megabase pairs (Mbp) of DNA on chromosome 9 roughly equivalent to chromosomal bands. Pools 9-1 to 9-5 (a total of 51 clones) and pools 9-6 to 9-15 (a total of 99 clones) map to the short and long arm of chromosome 9, respectively. The pool coverage ranges from 3.85 Mbp for pool 9-8 to 12.88 Mbp for pool 9-5. When large numbers of clones were grown, overnight cultures were done individually in 2 mL of Luria broth (LB) medium in 96 deep well plates (Beckman, City of Hope, CA). Fewer individual clones were grown overnight in up to 20 mL of Luria broth (LB) medium [69] containing 12.5 g/mL chloramphenicol (Sigma) and the DNA was extracted using an alkaline lysis protocol as described [70,71]. For preparation of DNA pools or "super-pools", i.e., combination of two or more pools, clones were grown individually and pooled prior to DNA extraction. Quality control and quantification of the DNA was typically done by agarose gel electrophoresis and fluorometry, respectively. Figure 11. Our BAC probe pooling strategy. Please note that the BAC clone in position A1 on Plate 1 was not used in the study of thyroid tissue described here.
In this communication, we will refer to the combination of all 150 BAC-derived DNA probes as whole chromosome painting (WCP) probe and call combinations of pools 9.1-9.5 and 9.6-9.15 "chromosome arm probes (CAP)" for chromosome 9p and 9q, respectively. To investigate chromosome 9 rearrangements in S48T with higher resolution, we labeled DNA extracted from 9p-specific clone pools and chromosome 9q-specific, adjacent pairs of pools with 5 different fluorochromes, and refer to these probes as "chromosomal rainbow probes (CRP)".

Conclusions
In many known instances, recurrent chromosomal rearrangements are not just random events in solid tumors, but become apparent once cells carrying these abnormalities gain growth advantages over other clones. Thus, knowledge regarding the physical location of translocation breakpoints, activation of proto-oncogenes or inactivation of tumor suppressor genes may provide crucial information for a better staging of tumors and/or the definition of treatment regimens for individualized anti-tumor therapy.
Technical approaches described in this communication outline rapid and thus cost-efficient ways to analyze a patient"s karyotype and reveal abnormalities within a matter of days. Utilizing resources that have been generated in the course of the International Human Genome Project, such as BAC libraries providing multi-fold coverage of the human genome, and avoiding the need for costly equipment, an average lab with basic instrumentation will now be able to perform and rapidly conclude high resolution physical mapping experiments of cancer genomes. makes any warranty, express or implied, or assumes any legal responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by its trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof, or The Regents of the University of California. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof, or The Regents of the University of California.