| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Biology and Genetics |
Department of Genetics, Institute for Cancer Research, [S. M. K., R. A. L.], and Departments of Pathology [R. H., A. E. S., J. M. N.] and Oncology and Radiotherapy [S. D. F.], University Clinic, The Norwegian Radium Hospital, N-0310 Oslo, Norway, and Department of Clinical Genetics, University Hospital of Lund, 22185 Lund, Sweden [P. Å.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Loss of DNA sequences from the short arm of chromosome 3 have been found in a number of different tumor types, including TGCTs (5, 6, 7, 8, 9, 10) . A candidate target gene for the observed 3p changes is the FHIT gene, which maps to 3p14.2 (11) . Homozygous deletions of FHIT, aberrant FHIT transcripts, and reduced Fhit protein expression have been identified in various cell lines and primary tumors (12 , 13) . Furthermore, replacement of FHIT has been shown to have a growth-inhibitory effect (14) . However, the suppressor designation of FHIT has been questioned because point mutations are rare (15 , 16) , splice variants are also found in normal tissue (17 , 18) , and Fhit fails to show growth-inhibitory effects in some cell lines (19 , 20) . The location of FHIT, in the unstable FRA3B region, has been suggested as the reason for the observed frequent deletions. On the other hand, additional functions of Fhit, apart from Ap3A binding and hydrolysis, have been revealed recently. Fhit interacts with tubulin (21) , associates with the ubiquitin-conjugating enzyme hUBC3 (22) , and promotes apoptosis (23) .
In the present study we examined 67 TGCTs to evaluate the importance of Fhit in the development of TGCT.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
For the AI analyses, DNA was isolated from freshly frozen tumor tissue, available from 62 of the patients. The frozen tumor specimens were evaluated for histological subtype(s) and percentage of tumor tissue by analysis of three frozen sections, taken at each end and in the center of the tumor piece. The frozen sections were stained with H&E and evaluated in a light microscope. The tumor specimens were classified according to the WHO recommendations (1) , and included 32 S samples (24 pure Ss and 8 S components from patients with combined tumors) and 30 NS samples. The visually estimated percentage of tumor cells for all 62 TGCTs was, on average, 73% (range, 17–100%). Among NSs, the average was 67% (range, 17–100%), and among S samples the average was 78% (range, 25–100%). Peripheral blood samples were collected from each patient and used as a constitutional DNA source for the microsatellite analysis. DNA from the tumor and blood samples was isolated using a standard phenol-chloroform extraction followed by ethanol precipitation.
In some cases the received frozen tumor piece was large enough to allow cutting off a piece for storage. This neighboring piece was, for each of 22 cases, used for FISH and mRNA analyses.
For the IHC analyses, 65 formalin-fixed, paraffin-embedded tumors, 32 S samples (26 pure Ss and 6 S components from patients with combined tumors) and 33 NSs, were analyzed. Fifty-eight of these cases corresponded to the fresh-frozen tissue pieces used for AI analyses.
AI Analysis.
Ten microsatellite loci mapping to 3p14.1-21.1 were analyzed, including 9 dinucleotide repeat markers (D3S1217, D3S1285, D3S1600, D3S1312, D3S1481, D3S1300, D3S1234, D3S1313, and D3S1295) and 1 tetranucleotide repeat marker (D3S1766). All primer sets were obtained from Research Genetics (Huntsville, AL). The location of all markers except D3S1481 and D3S1766 are available through the NCBI database.4
D3S1481 is located in intron 4, and D3S1300 and D3S1234 are located in intron 5 of the FHIT gene (25)
. The location of D3S1766 is found from the Genetic Location Database.5
Each locus (the microsatellite repeat with short flanking sequences) was amplified from template DNA (50–100 ng) by PCR, incorporating radioactive dCTPs during the PCR. The primer sets were run in multiplex or single PCRs according to previously described procedures (26)
. The products were separated on denaturing polyacrylamide gels and visualized by autoradiography.
AI, i.e., a skewed intensity ratio between the alleles in tumor DNA compared with constitutional DNA, LOH, i.e., complete loss of one allele in the tumor DNA, and microsatellite instability, i.e., the appearance of novel alleles in the amplified tumor DNA, were evaluated by visually comparing the pattern of the amplified tumor DNA with that of the amplified constitutional DNA. The results were independently evaluated by two of the authors and were reevaluated if agreement was not reached. All positive findings were confirmed by a second, independent PCR and electrophoretic run.
The criteria for determination of a SRO were based on definitions by Thorstensen et al. (27) . Briefly, the closest flanking marker on each side of a SRO must reveal heterozygosity in at least 10% of the total number of tumors exhibiting AI or LOH, i.e., in this study three tumors. In addition, at least two contiguous markers on each side of the SRO must show retained heterozygosity, although not necessarily in the same tumor.
FISH.
FISH was performed on interphase nuclei from each of 22 tumors. The nuclei were prepared as described previously (28)
. Briefly, frozen tissue samples were minced with a scalpel and submerged in methanol and acetic acid (3:1, v/v) fixative at -20°C for at least 15 min. The suspension was spun down, and the tissue pellet was dissociated to single cell suspension in 60% acetic acid at room temperature. After 2–3 min, preparations were made, left to dry overnight, and then stored at -20°C until use.
FISH was performed simultaneously with a biotinylated 3p telomere-specific probe (3pter; American Laboratory Supplies, Arlington, VA) and a spectrum orange directly labeled centromere-specific probe (CEP3; Vysis, Drowners Grove, IL) according to a modified manufacturer’s protocol. The slides were treated with RNase for 1 h at 37°C, denatured in 70% formamide with 2x SSC at 70°C for 2 min, and dehydrated through 1-min stepwise immersions in 70, 85, and 100% ethanol. The slides were dried and prewarmed on a hot plate at 37°C, and the probe solution (7 µl of hybridization buffer, 0.5 µl of CEP3 probe, and 1.5 µl of 3pter probe) was denatured for 5–10 min in a 70°C water bath. After application of probe solution and a coverslip (sealed), the slides were left overnight at 37°C in a prewarmed humidified box. The next day the slides were posthybridized by three 5-min immersions in 50% formamide-2x SSC at 45°C, three 2-min immersions in 4x SSC-0.05% Tween 20 at room temperature, and one 2-min immersion in 4x SSC at room temperature, followed by stepwise application of 100 µl of fluorescein-labeled avidin, 100 µl of biotinylated antiavidin, 100 µl of fluorescein-labeled avidin, and 10 µl of antifade solution with 4', 6-diamino-2-phenylindole. Between the four labeling steps, the slides were kept in a dark moisture chamber for 20 min, followed by washes in 4x SSC-0.05% Tween 20 and 4x SSC, each time at room temperature for 2 min. The fluorescence signals were evaluated in a fluorescence microscope (Zeiss Axioplan, Oberkochen, Germany). The numbers of centromeric and telomeric fluorescence signals were recorded from each of 
100 nuclei per tumor.
To compare the FISH results with those from AI analyses, we used the following criteria. If a nuclei showed two centromere and two telomere signals, it was interpreted as disomic for chromosome 3. If it presented three of each signal, it was interpreted as trisomic, and so forth. A tumor was classified as disomic if >50% of the nuclei were disomic and less than one-third of the nuclei showed three or more copies. If more than one-third of the nuclei showed three copies and the majority of the remaining nuclei were disomic, the tumor was classified as trisomic. The tumor was polysomic if the tumor could not be classified as trisomic and in sum one-third or more of the nuclei each presented three or more copies. If more centromere than telomere signals were recorded, this suggested loss of the whole or parts of the chromosome arm.
FHIT mRNA Analysis.
Seventeen of the TGCTs and normal testicular tissue from two of the TGCT patients were submitted to FHIT transcript analyses, described previously by Panagopoulos et al. (29)
. Briefly, mRNA was isolated from frozen tissue using Dynabeads oligo(dT)25 (Dynal AS, Oslo, Norway), followed by reverse transcription into cDNA and nested PCRs. The first and each of three nested PCRs were performed with 25 and 30 cycles, respectively. The following primer sets were applied: outer set FH107 and FH1033 (exons 1–10) and inner sets FH129 and FH984 (exons 1–10), FH380 and FH677 (exons 5–8), and FH380 and FH578 (exons 5–6). The nested PCR products were analyzed by electrophoresis through agarose gels, stained with ethidium bromide, and photographed. Thereafter the reverse transcription-PCR (RT-PCR) products were size-fractionated by electrophoresis through a low-melting point agarose gel, purified from the gel, and sequenced on the Applied Biosystems model 373A DNA sequencing system. The PCR mixtures, PCR programs, primer sequences, and sequencing are described in Panagopoulos et al. (29)
.
IHC.
Blocks with formalin-fixed, paraffin-embedded tumor tissue were available from 65 TGCTs. Tissue sections from these blocks were deparaffinized and stained using the biotin-streptavidin-peroxidase method (Supersensitive Immunodetection System, StrAviGen Multilink kit; Biogenex, San Ramon, CA) and OptiMax Plus Automated Cell Staining System (Biogenex). To unmask the epitopes, the tissue sections were placed in 10 mM citrate buffer (pH 6.0), microwaved twice (5 min each time), and then treated with 1% H2O2 for 10 min for blocking of endogenous peroxidase. The sections were incubated with the polyclonal rabbit anti-Fhit, ZP54 (Zymed Laboratories, Inc., San Francisco, CA), diluted 1:100 (5 µg IgG/ml), for 30 min at room temperature, followed by incubation with biotin-labeled secondary antibody (1:30 dilution) and streptavidin-peroxidase (1:30 dilution) for 20 min each. The sections were then stained for 5 min with 0.05% 3'3-diaminobenzidine tetrahydrochloride freshly prepared in 0.05 M Tris buffer (pH 7.6) containing 0.024% H2O2, and counterstained with hematoxylin, dehydrated, and mounted in Diatex. All dilutions of primary antibody, secondary antibody, and the streptavidin-peroxidase were made with PBS (pH 7.4) containing 1% BSA. A positive control was included in each experiment. Negative controls included replacement of anti-Fhit with normal rabbit IgG (5 µg IgG/ml). All controls showed satisfactory results.
The immunostaining was evaluated on a three-tiered scale for intensity (1, absent/weak; 2, moderate; 3, strong) and for extent of staining (percentage of positive tumor cells: 1, <10%; 2, 10–50%; 3, >50%) as defined by Greenspan et al. (30)
. The intensity and extent were multiplied to give a composite score (range, 1–9) for each tumor. All slides were scored by two of the authors (R. H., J. M. N.). Composite scores 
3 were defined as marked reduction or absence of Fhit protein expression (30)
compared with normal testicular tissue.
Statistical Analysis.
Pearson 
2 tests were performed to evaluate the differences observed in our results. However, when one of the expected values was <5, we applied the Fisher’s exact test. All tests were two sided. P 0.05 was interpreted as being statistically significant. Statistical analyses on differences between the main histological subgroups (i.e., Ss versus NSs) were performed for two alternatives of the S samples: 1, S samples, i.e., pure Ss and S components from combined tumors; and 2, pure Ss.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
AI Analysis.
Twenty-seven of 62 TGCTs (44%) showed AI at one or more of the 3p loci (Fig. 1A)
. The frequencies of AI at each locus ranged from 15 to 35% (Fig. 1B)
and did not differ significantly from each other. The majority of the changes observed were AI, whereas complete loss of one allele (LOH) was seen in five tumors, at two to seven loci. Eleven tumors (18%) displayed AI at all informative loci, whereas 16 tumors (26%) showed partial AI and were used to define one SRO. The SRO was flanked by D3S1312 and D3S1234. The distal side was determined by tumors 4, 108, 116, and 145, and the proximal side was determined by tumors 79, 87, 108, and 115. It should be noted that, in Fig. 1A
, samples 4, 108, and 116 indicate a more narrow SRO than the one described above, but according to our criteria at least three tumors must define each border. The SRO included parts of FHIT. Eleven tumors showed intrachromosomal alterations, as changes from AI to heterozygosity were observed at loci within FHIT or the two loci flanking FHIT. The importance of the FHIT region was further supported by the genotype pattern in three tumors, which showed alterations at all informative loci but each with intrachromosomal alterations, because increased change from AI to LOH was observed in this region. The most frequently altered loci were D3S1481 (32%), which is located in intron 4 of FHIT, and D3S1766 (35%).
|
Combined Evaluation of AI and FISH Results.
Among the 22 tumors analyzed by FISH, 9 revealed the constitutional genotype pattern in the 3p region examined. Three of these were disomic, whereas six were classified as trisomic in the FISH analyses. Of six tumors showing AI at all informative loci, three were trisomic, two were polysomic, and one was disomic. The remaining seven tumors displayed a mixed pattern of AI and constitutional genotypes; of these two were disomic, two were trisomic, two were polysomic, and one tumor had telomeric loss in 58% of the nuclei. Examples of AI and retained heterozygosity in tumor DNA and the corresponding telomere and centromere signals in interphase nuclei are shown in Fig. 2
.
|
). In four tumors, exon 2 was spliced to exon 7 (E2/E7), and thus the transcript lacked exons 3–6. One of these tumors, tumor 106, also contained an E2/E9 transcript. We did not succeed in sequencing the aberrant transcripts of two tumors (tumors 84 and 146). However, tumor 146 showed an aberrant gel band of slightly larger size than the E2/E7 transcript, and the aberrant band in tumor 84 was only slightly smaller than the normal transcript (Fig. 3)
.
|
Normal and variant transcripts were seen in 5 tumors, 1 of 7 Ss, and 4 of 10 NSs. Lack of RT-PCR product with the FH129 and FH984 primer set was seen in 3 of 7 Ss and 3 of 10 NSs (Table 1)
.
|
.). Two tumors with concomitant CIS showed reduced Fhit expression in the CIS areas. These were CISs from patients with NS. The NSs, S samples, and pure Ss showed reduced or absent protein expression in comparable fractions, 19 of 33 (58%), 21 of 32 (66%), and 16 of 26 (62%), respectively. Different expression was observed among the components of teratomas including both mature and immature teratomas. The malignant nonepithelial teratoma components showed reduced expression in 87% (13 of 15) cases, whereas the malignant epithelial components showed reduced expression in 20% (3 of 15; P < 0.001).
|
Finally, reduced expression in the tumor tended to be linked to the presence of metastasis at the time of diagnosis (P = 0.13). Metastases were observed in 17 of 40 patients with reduced Fhit expression in the tumor, as opposed to 6 of 25 cases with normal expression. This tendency was seen both among Ss and NSs. Relapse was seen more often among patients with reduced Fhit than in those with normal expression, but this was not statistically significant.
Combined Evaluation of AI, mRNA, and Protein Expression of FHIT in the Same Tumors.
Seventeen tumors (10 NSs and 7 Ss; 6 pure Ss) were analyzed for FHIT alterations at the DNA, mRNA, and protein levels (Table 1)
. Changes at the DNA level were associated with variant transcripts or absence of PCR products with the FH129 and FH984 primers (P = 0.006). All 11 TGCTs presenting AI at loci within FHIT or breakpoints close to FHIT showed either variant transcripts or absence of the long PCR product, whereas among the remaining 6 tumors, which were heterozygous at FHIT loci, 2 showed aberrant transcripts and 4 had only normal-length transcripts. No association was found between aberrant transcript and protein expression. Reduced protein expression was seen both among tumors with aberrant transcripts (four of seven), tumors with absence of PCR products with the FH129 and FH984 primers (three of six), and tumors with only normal-length transcripts (two of four). Aberrant transcripts tended to be linked to nonseminomatous subtype and were found in 6 of 10 (60%) NSs, but in only 1 of 7 (14%) S samples (P = 0.13).
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The FISH results presented here confirmed previous cytogenetic data showing that chromosome 3 is usually present in two or more copies in TGCTs (5 , 7) . AI at all informative loci, detected in 18% of the TGCTs, may reflect low copy number gains, assuming that the observed loci presenting AI are in cis position, and that the AI observed at two consecutive loci also include the region between them. This is in line with the fact that five of six tumors displaying AI at all informative loci were trisomic or polysomic. However, six of nine tumors with retained heterozygosity at all loci were also trisomic. An interstitial deletion in one of three chromosomes can appear as unchanged, i.e., heterozygous, in the AI analysis. Alternatively, the number of trisomic nuclei may be too small to be detected by microsatellite analysis, or intratumor heterogeneity may exist. Previous comparative genomic hybridization studies have also found both gains and losses of whole or parts of chromosome arm 3p (32, 33, 34, 35) , and Southern blot analyses have revealed AI and/or LOH at 3p in 45–67% of the analyzed tumors (8, 9, 10) . Our results confirm that AI at 3p may reflect either gains or losses in TGCTs.
One SRO was determined in the present study. In these polyploid tumors, the importance of a SRO is unknown because the genotype pattern used to delineate the SRO may include both gains and losses. The SRO is, however, delineated from breakpoints (AI/LOH to heterozygosity) on the chromosome arm and is limited to parts of the FHIT gene.
The fact that the aberrant FHIT transcripts presented deletion breakpoints in boundaries of exons and introns indicated that they result from aberrant splicing rather than allelic deletions. The splice variants all lacked exon 3, and four of six lacked exons 3–6. The latter transcripts will not be translated because the start codon is located in exon 5. These splice variants have previously been reported in other tumors types and in non-neoplastic tissues (17
, 36, 37, 38, 39)
. However, the FHIT aberrant transcripts were often faint gel bands compared with the wild-type transcript. This is in contrast to our results, where the aberrant splice transcripts in most TGCTs appeared visually more abundant than the normal transcripts. In six tumors, no products were found by the primer set FH129 and FH984 (exons 1–10), although normal-length products were obtained by the primer sets amplifying exons 5–6 and exons 5–8. This indicates homozygous deletions in parts of the FHIT gene, although partial RNA degradation or suboptimal cDNA synthesis cannot be excluded. Two tumors (tumors 124 and 137) revealed only variant transcripts for the exon 1–10 fragment (Fig. 3)
but normal transcripts for exons 5–6 and exons 5–8 (Table 1)
. This observation can be explained by tumor heterogeneity, by the presence of normal cells, by the fact that smaller transcripts amplify more easily, or by homozygous deletions.
The high frequency of TGCTs showing reduced Fhit protein expression is comparable to data from tumors of the lung, colon, breast, esophagus, kidney, and cervix (30 , 40, 41, 42, 43) . Lack of association of Fhit protein expression and changes at the DNA or RNA level have been observed earlier (44 , 45) . Although AI at loci within FHIT did not correlate with reduced protein expression, deletions not detected by the present analysis may have occurred elsewhere in the gene. Some TGCTs contained an increased copy number of chromosome 3, and thus most likely more copies of FHIT, and would be expected to show increased, rather than reduced protein levels. This apparent discrepancy might be explained by inactivation of FHIT through point mutations and or promoter methylation. Point mutations appear to be rare in FHIT (15 , 16) , but it has been suggested that 5' CpG island methylation represents an alternative pathway for silencing of the FHIT gene (46) . In fact, any regulatory mechanism, from gene transcription to protein stability, may be involved. Down-regulation of normal FHIT transcription could explain tumors with normal-length transcripts but reduced expression because our RT-PCR was qualitative, not quantitative. Higher protein expression than expected, according to observed normal and/or aberrant transcripts, may be explained by changes in the post-transcriptional regulatory mechanisms. In addition, a large amount of normal cells present in a few samples might have masked AI in the microsatellite analyses, whereas in the IHC analysis, the tumor parenchyma was identified by morphological examination. Certainly, intratumor heterogeneity may also explain observed differences because the analyses were performed in different, but adjacent tumor specimens (see "Materials and Methods").
The scoring for protein expression was based on both intensity of staining and percentage of stained tumor cells. A subset of tumors was classified as positive or negative depending on the amount of tumor tissue in the section showing moderate staining. For example, if 10–50% of the tumor tissue showed moderate staining, the tumor sample was classified as normal. Moderate staining could reflect loss of one allele in most cells and could be important, given a dose effect of the gene. Recent reports demonstrated haploinsufficiency for certain tumor suppressor genes, i.e., that reduced expression is sufficient to predispose for tumor formation (47)
. However, even when all tumors with composite scores 4 were included as reduced, in total 48 cases, there was no correlation between AI at one or more loci located within FHIT or at 3p and Fhit protein expression.
Reduced Fhit expression was observed in CIS, Ss, and NSs. We found no differences between the main histological TGCT subtypes, NSs and Ss, regarding Fhit protein expression. However, the malignant nonepithelial components in teratomas were associated with reduced expression. The reduced expression may provide these cells with a growth advantage, given a tumor suppressor effect of Fhit. Our clinical data may lend some support to this hypothesis because reduced expression of Fhit in TGCTs seemed to be associated with advanced disease. Further knowledge about the patient/tumor genotypes and phenotypes and about Fhit and its potential modulators may provide the means to individually tailor treatment and to predict the risk of relapse.
We conclude that the present results suggest that loss of Fhit function play an important role in the development of TGCTs.
|
FOOTNOTES |
|---|
1 Supported by grants from the Norwegian Cancer Society (to R. H., S. D. F, J. M. N., and R. A. L.). S. M. K. is a research fellow for the Norwegian Cancer Society. 
2 To whom requests for reprints should be addressed, at Department of Genetics, Institute for Cancer Research, The Norwegian Radium Hospital, Montebello, N-0310 Oslo, Norway.
3 The abbreviations used are: TGCT, testicular germ cell tumor; S, seminoma; NS, nonseminoma; CIS, carcinoma in situ; FHIT, fragile histidine triad gene; AI, allelic imbalance; FISH, fluorescence in situ hybridization; IHC, immunohistochemistry; LOH, loss of heterozygosity; SRO, smallest region of overlapping changes; RT-PCR, reverse transcription-PCR.
4 http://www.ncbi.nlm.nih.gov.
5 http://cedar.genetics.soton.ac.uk/public_html/ldb.html.
Received 7/17/01. Accepted 11/ 9/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
, AI; 
, constitutional homozygosity; 
, retained heterozygosity in tumor DNA; nd, not determined; 
, SRO. B, frequencies of AI at each locus, observed in TGCTs.
