| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
EXPERIMENTAL STUDY |
1 Department of Molecular Medicine, Karolinska University Hospital CMM L8:01, SE-17176 Stockholm, Sweden, 2 Department of Surgical Sciences, Karolinska University Hospital P9:03, SE-17176 Stockholm, Sweden, 3 Department of Surgery, Center for Metabolism and Endocrinology, Karolinska University Hospital K54, SE-14186 Stockholm, Sweden, 4 Department of Oncology and Pathology, Karolinska University Hospital P1:02, SE-17176 Stockholm, Sweden and 5 Center for Genomics and Bioinformatics, Karolinska Institutet, SE-17177 Stockholm, Sweden
(Correspondence should be addressed to L Forsberg or C Larsson, Department of Molecular Medicine, Karolinska University Hospital CMM L8:01, SE-17176 Stockholm, Sweden; Email: Lars.Forsberg{at}cmm.ki.se or Catharina.Larsson{at}cmm.ki.se)
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Design and methods: Sporadic parathyroid adenomas and normal parathyroids were hybridized against HG-U95Av2 oligonucleotide arrays (Affymetrix) containing a total of 12 625 probe sets. Quantitative real-time PCR (QRT-PCR) was performed in a larger series of parathyroid adenomas, in order to con-firm the microarray results.
Results: Cyclin D1 and c-Jun showed increased expression in adenomas vs normal parathyroids by microarray analysis and QRT-PCR, suggesting an oncogenic role of these genes in parathyroid tumor development. At unsupervised hierarchical clustering, the adenomas fell into two groups: Group I adenomas were characterized by 11q13 LOH, while Group II adenomas lacked this abnormality. In addition, a t-test analysis identified largely non-overlapping genes with differential expression in the tumors subgroups; e.g. in Group I tumors the putative oncogene ENC 1 was found highly over-expressed vs Group II adenomas.
Conclusions: The microarray analyses revealed partly distinctive and partly common expression profiles in parathyroid adenomas with and without 11q13 LOH. In addition, approximately half of the under-expressed genes were mapped to chromosome 11, in agreement with a dose effect following loss of this chromosome.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
In sporadic forms of the disease, inactivation of HRPT2 is preferentially found in the malignant cases while MEN1 inactivation is mainly found in the benign form (3, 6–8). Furthermore, the Cyclin D1 gene as well as several genes related to calcium homeostasis have been suggested to be involved in the development of PHPT (9–11). Somatic deletion of chromosome 11 is the most frequent genetic aberration in parathyroid adenoma, and has been demonstrated by both comparative genomic hybridization (CGH) and loss of heterozygosity (LOH) techniques (12–14). This alteration is seen in a third of the tumors, and generally represents a quite specific chromosomal alteration accompanied by inactivating mutation of the MEN1 gene in the other allele (6, 15). In addition, losses of chromosomes 1p, 6q, 9p, 11p, 13q and 15q as well as gains of chromosomes 7, 16p and 19p are also demonstrated in sporadic hyperparathyroidism (12, 13, 16).
In the present study, we have characterized the global gene expression profiles in a series of sporadic parathyroid adenomas, in an attempt to obtain an improved picture of the genetic etiology behind parathyroid tumor development.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Seventeen parathyroid tumors (T1–T17) were collected from patients treated surgically for PHPT at the Karolinska University Hospital (Table 1
). Informed consent was obtained from each patient and the local ethics committee approved the study of the clinical material. All patients had a sporadic form of the disease without a personal or family history of MEN 1, HPT-JT or other familial forms of PHPT. The tumors were classified histopathologically as uniglandular parathyroid adenomas according to WHO criteria (17). Representative sections were analysed to ensure that the samples used for DNA and RNA extractions contained a sufficient proportion of tumor cells for the analysis (i.e. > 70%). Tumors T1–T8 were analysed by microarray, LOH and CGH, while tumors T1–T17 were all analysed by real-time PCR. The two biopsies from normal parathyroid glands (N1 and N2) were obtained from patients treated for nodular thyroid lesions when parathyroid glands had to be excised and reimplanted into the sternomastoid muscle for anatomical reasons. Both these patients had normal levels of PTH, serum calcium and thyroid hormones preoperatively. They were both verified as histopathologically normal, and were used as reference samples in the microarray and real-time PCR analysis.
|
All tissue samples were collected immediately after surgery and stored at –70 °C until use. Total RNA was extracted from all 17 tumors (T1–T17) and the two normal parathyroid biopsies (N1 and N2) using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA), followed by purification with an Rneasy mini kit (Qiagen, Hilden, Germany). The RNA purity and quality was confirmed by analysis in an Agilent 2100 Bioanalyser (Agilent Technnologies, Inc, Paolo Alto, CA, USA). For eight of the adenomas (T1–T8), DNA was isolated in parallel with the RNA during the TRIzol extraction. Peripheral blood leukocytes were used as a source of constitutional DNA in the microsatellite analysis (T1–T8).
LOH analyses
The eight matched blood and tumor DNA samples from T1–T8 were genotyped for three microsatellite markers located at the MEN1 locus in 11q13: cen-(D11S4946/MEN1)- D11S4938-D11S4937-tel. For T2 and T3, LOH analyses of markers from the MEN1 region have been previously published (15). The markers were analysed using fluorescent detection in an ABI 377 laser fluorescent sequencer and the GeneScan 3.1 software (Applied Biosystems, Perkin Elmer Corp., Foster City, CA, USA).
Microarray experiments and data analyses
For eight of the tumors (T1–T8) and the two normal biopsies (N1 and N2), cDNA was synthesized from total RNA using a T-7 linked oligodT primer, and then used for cRNA synthesis with biotinylated UTP and CTP. After fragmenting, the labeled RNA was hybridized to HG-U95Av2 oligonucleotide arrays (Affymetrix Inc., Santa Clara, CA, USA) according to the protocol recommended by the supplier. After hybridization, the arrays were washed and stained with streptavidin–phycoerythrin (Molecular Probes, Inc., Eugene, OR, USA) in an Affymetrix fluidics station. The arrays were then scanned in an Affymetrix scanner and the expression values for each probe set (represented by 16–20 selected oligomers) were estimated using the Affymetrix Microarray Suite Software (MAS v5.0). To minimize discrepancies resulting from individual variations regarding sample preparations, hybridization conditions, staining or different oligoarray lots, all probe sets were scaled from each array to a target intensity value of 100 (using Affymetrix MAS v5.0), thus allowing interarray comparisons. Scaling factors (range 2.28–5.26) varied within the acceptable range i.e. <threefold. 3'/5' ratios for GAPDH and ß-actin were confirmed to be within acceptable limits (< threefold), and BioB spike controls were found to be present on all chips. In addition, the internal controls BioC, BioD and CreX were present in increasing intensity.
By using the statistical detection algorithm in Affymetrix MAS v5.0, we were able to separate the transcripts of particular probe sets that are reliably detected (present call), from the transcripts below the threshold of detection and therefore considered as absent. From the 12 626 probe sets on the array, control probes and genes scored as marginal or absent (not detected) were excluded, leaving 4436 present probe sets for the subsequent analyses.
Comparison files were created using the batch analysis tool in the MAS v5.0. These included files for each of the parathyroid adenomas (T1–T8), which were then individually compared with each of the normal parathyroids (N1and N2), thus yielding 16 comparison files. To identify differentially expressed genes between parathyroid adenomas and normal parathyroids, two different algorithms from the MAS v5.0 tool were applied. The change algorithm reported the probe sets to be unchanged, increased, decreased or marginal. The signal log ratio algorithm instead calculated an estimated log value for each probe set in each sample. The algorithms are described at http://www.affymetrix.com/support/technical/technotes/statistical_reference_guide.pdf. For both algorithms the signal values were imported into the GeneSpring 5.0 software tool (Silicon Genetics, Redwood City, CA, USA) to sort out genes with increased or decreased expression. Unsupervised hierarchical clustering analysis of the 4436 present probe sets was performed using Gene-Spring 5.0 software with the minimum distance set to 0.001 and the separation ratio set to 0.95. The comparisons of mean expression levels were compared in the two groups of tumors identified by clustering, i.e. T1–T4 with 11q13 LOH (Group I) and T5–T8 without 11q13 LOH (Group II). The comparisons were carried out using the expression values estimated by dChip analysis (www.dchip.org) (18). This was done to obtain more sensitive estimates of the low-abundance genes than were available through MAS v5.0. To avoid effects related to the wide variation at the upper end of the distribution of values, the logarithmic values were used. t-statistics were computed for each gene between the two groups of tumors identified by clustering. The false positive rate was estimated at several different thresholds for the t-score, by permutating the sample labels and re-computing all gene-wise t-scores. The t-score threshold was set so that the estimated number of false positives was half the number of genes with t-scores exceeding that threshold (i.e. the ‘false discovery rate’ was one half).
CGH
CGH was carried out on tumors T1–T8 using methods previously described (13). For T2 and T3 the results have been previously published (13). A minimum of ten three-color digital images (DAPI, FITC and Spectrum Red fluorescence) were captured from each hybridization using an Axioplan 2 epifluorescence microscope (Carl Zeiss Jena GmbH, Jena, Germany) equipped with a Sensys charged-coupled-device camera (Photometrics, Tucson, AZ, USA) and analysed with the isis/CGH software (Metasystem, Altlussheim, Germany). Relative DNA sequence copy number changes were detected along the length of all chromosomes in each metaphase spread. Green to red ratios < 0.8 were scored as losses, ratios > 1.2 as gains and ratios > 2.0 as high-level amplifications. Heterochromatic regions in the centromeric and paracentromeric parts of some chromosomes, the Y chromosome, the short arm of the acrocentric chromosomes as well as the region next to the telomeres were not included in the evaluation. All tumor samples were hybridized and analysed on two different occasions, with concordant results.
Quantitative real-time PCR (QRT-PCR)
Total RNA was isolated from T1–T17 and N1 and N2 as described above. To avoid the presence of genomic DNA all samples were treated with DNAase after RNA extraction. Ten micrograms of RNA were reverse transcribed into cDNA using the TaqMan reverse transcription Kit (Applied Biosystems). Reverse transcription was performed in a thermal cycler at 25 °C for 10 min and 37 °C for 120 min. For five of the six genes analysed by TaqMan real-time PCR, Assays-on-Demand kits (Applied Biosystems) were available including Cyclin D1 (assay identification (ID) is Hs00233498_m1), c-Jun (Hs00277190_s1), ß-actin (Hs99999903_m1), TBP (Hs99999910_m1) and 18S rRNA (Hs99999901_s1). For each single sample, the master-mix included: 12.5 µl TaqMan 2 x Universal Master-Mix (Applied Biosystems), 8.25 µl RNAse-free water and 1.25 µl 20 x Assays-on-Demand gene expression assay mix. For the 36B4 gene (public ID M17885 [GenBank] ) no Assays-on-Demand are available, while the following previously published primers and probe were used (19): forward primer: 5'-AGATGCAGCAGATCCGCAT-3'; reverse primer: 5'-ATATGAGG AGCAGTTTCTCCAG-3'. The corresponding TaqMan probe 5'-AGGCTGTGGTGC-TGATGGGCAAGAAC-3' was purchased from MedProbe (catalogue number OL-0361-FAMDQ02). For 36B4, the TaqMan PCR Master-Mix was constituted of 12.5 µl TaqMan 2 x Universal Master-Mix, 7 µl RNAse free water, 1.5 µl primers (10 µM) and 1 µl 36B4 probemix (5 µM;) per sample to be analysed. For each sample and gene, 22 µl aliquots of the MasterMix and 3 µl (~15 ng) cDNA were transferred in duplicates to adjacent wells of a MicroAmp Optical 96-well reaction plate (Applied Biosystems) and covered with Optical Caps (Applied Biosystems). The plate was then placed in the ABI Prism 7700 Sequence Detector (Applied Biosystems) and the run commenced. The conditions for every TaqMan PCR reaction were: 2 min at 50 °C (stage 1), 10 min at 95 °C (stage 2) and 40 cycles of 15 s at 95 °C followed by 60 s at 60 °C (stage 3). When the run was completed, output data were analysed using the SDS software and exported into a Microsoft Excel file for analysis. The 36B4 housekeeping gene was used as an endogenous control to normalize the expression level of target genes. Standard curves for 36B4 and target genes were constructed using 1:10 serial dilutions of control cDNA prepared from the human neuroblastoma cell-line SK-NAS.
In the fold-change analysis, the RNA level was standardized by correcting for the 36B4 level in the same sample. For all experimental samples (T1–T17), the quantity was determined from the standard curve and divided by the quantity of the mean normal quantity (N1/N2). Thus, the normal quantity becomes the 1 x sample, and all other quantities are expressed as a relative n-fold difference.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
. The findings and expression analyses are illustrated in Fig. 1 and Tables 2–5.
|
|
|
A total of 4436 of the 12 625 probe sets represented on the array gave a present call in at least eight of the ten arrays. This set of present genes was considered as informative and used in all comparisons in this study. However, for the analysis of the genes with decreased expression in the adenomas, the 4533 genes detected as present in both normal parathyroids were included in the analysis.
Differentially expressed genes between parathyroid adenomas (T1–T8) and normal parathyroids (N1 and N2) were identified by pair-wise comparison using change and signal log ratio algorithms in the Affymetrix MAS v5.0. The computing results from both analyses were then imported into the GeneSpring software tool, in order to identify probe sets with increased or decreased expression. A total of 224 probe sets showed increased expression in the adenomas vs normal parathyroids using at least one of the algorithms. The 35 probe sets (representing 31 genes) that were independently identified with increased expression in the adenomas using both algorithms are listed in Table 2. Several of the over-expressed genes are putative oncogenes involved in cell cycle control or transcriptional regulation, e.g. the Cyclin D1 gene and the c-Jun protooncogene. Similarly, 361 probe sets were detected with decreased expression in the adenomas by at least one algorithm. Eighty of these (representing 75 genes) were identified as decreased by both algorithms (Table 4). Noteworthy, CACNB2 and CXCL12 involved in calcium transport and homeostasis were present among these genes with decreased expression.
|
In addition, we investigated the microarray expression of genes selected because of their demonstrated or implicated relation to parathyroid functions (Table 3). The calcium-sensing receptor gene (CASR) showed a distinct pattern of decreased expression in all eight adenomas. The mean expression level was 0.61 (range 0.43–0.80) as compared with the mean expression of the two normal parathyroids (1.00). Similarly, a reduced expression was noted for the vitamin D receptor gene (VDR), with a mean reduction from 1.0 in the normal parathyroids to 0.47 (range 0.20–0.64) in the eight adenomas.
|
Similarities in the global gene expression patterns of normal and pathological parathyroids were studied by creating a hierarchical tree. The unsupervised hierarchical clustering analysis of the present probe sets (4436) revealed that the eight tumor samples fell into two distinct cluster groups (Fig. 1A). Group I was found to consist of the four tumors with 11q13 LOH (T1–T4) and Group II of the four tumors without 11q13 LOH (T5–T8). The expression profiles between the tumors in the two groups were then compared by direct t-test comparisons of the means between the two groups. The t-tests revealed 85 genes that showed significantly (P < 0.0032) differential mean expression between the two groups (Table 5). The results obtained from the two independent approaches (MAS v5.0: Fig. 1 and dChip: Table 5) support the distinction in expression between Group I and Group II adenomas.
As is evident from Table 5, approximately 50% of the genes with decreased expression in Group I vs Group II were located on chromosome 11 within and outside the q13 region (Fig. 1B). This indicated that the lower expression of some genes in Group I could be a result of low DNA dose, i.e. a reduction in DNA copies. In addition, no gene located on chromosome 11 was detected with increased expression when comparing the adenomas in Group I vs Group II. In order to detect the chromosome 11 status in the adenomas used for array analyses we complemented the LOH analysis with CGH. Of the eight adenomas hybridized onto microrarrays, tumors T1–T4 were by LOH performance identified with a deletion in 11q13 while T5–T8 had retention of the same chromosomal region. CGH results confirmed that the entire chromosome 11 was lost in three (T1–T3) of the four tumors in Group I. In T4, no loss of 11 was detected by CGH, suggesting that tumor T4 carries a balanced alteration or a small deletion of chromsome 11q13 beyond the resolution of CGH. The four adenomas in Group II (T5–T8) all showed retention of chromosome 11 by CGH. In addition, losses were detected in 1p32-ter (T4), 6cen-q22 (T4), 15q (T3, T8), 21q (T3, T8) and 22q (T8), while trisomies were observed for chromosomes 6, 8, 14 and 16 (T7).
QRT-PCR analyses of selected genes
Two genes (Cyclin D1 and c-Jun), detected with increased expression in tumors vs normal parathyroid by the array analysis, were selected for QRT-PCR on a larger series of sporadic parathyroid tumors, including 17 adenomas T1–T17 (Table 1) and two normal parathyroid biopsies (N1 and N2). In these analyses, the quantity of the tumor cDNA was determined from the standard curve and divided by the quantity of the normal parathyroid. The 36B4 housekeeping gene was used as an endogenous control to normalize the expression level of target genes. Cyclin D1 thus showed more than twofold increased expression in 4/17 (24%) of the adenomas compared with the average expression of the two normal parathyroid samples (Fig. 1C). Similarly, c-Jun showed more than a twofold increased expression in 6/17 (35%) of the adenomas (Fig. 1D).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Comparison of parathyroid adenomas and normal parathyroid
The result of over-expressed genes in adenomas vs normal parathyroid identifies several putative oncogenes implicated in cell growth and transcriptional regulation, such as the Cyclin D1 gene and the c-Jun protooncogene (Table 2). Cyclin D1 over-expression is in agreement with the protein over-expression previously detected by immunohistochemistry in parathyroid adenomas (20, 21). A pericentric inversion with breakpoints at 11q13 and 11p15 is suggested to bring the Cyclin D1 oncogene under the influence of the 5'-regulatory sequences of the PTH gene (9), which results in dramatic over-expression of Cyclin D1. While this rearrangement has been described in a few tumors only, Cyclin D1 protein over-expression has been reported in 18–40% of adenomas by immunohistochemistry studies. Cyclin D1 promotes the G1–S-phase transition, probably by activating a cyclin-dependent kinase, but may also interact with other proteins and influence other pathways related to cell proliferation (22). In our study, over-expression of Cyclin D1 and c-Jun in the adenomas vs normal parathyroid tissue was confirmed by real-time PCR, which represents an accurate method for mRNA expression quantification. In the extended series of 17 adenomas, Cyclin D1 was over-expressed (>twofold) in 24% of the cases, which is in good agreement with the reported immunohistochemistry results (20, 21). Notably, in the four adenomas with Cyclin D1 over-expression, the level was largely increased compared with the normal parathyroid (4- to 36-fold), suggesting its relevance for tumor development.
The c-Jun gene, located in chromosomal region 1p32, has previously been shown to induce oncogenic transformation, probably due to several mechanisms: c-Jun is upregulated in many carcinomas and one suggested function (e.g. in primary hepatocytes and in liver tumors) is to antagonize the proapoptotic activity of p53 (23). Furthermore, c-Jun is a member of the same family of Jun transcription factors as JunD (24). Menin, encoded by the MEN1 gene in 11q13, interacts with the AP-1 transcription factor leading to a repression of the JunD-mediated transcription (25). Interestingly, menin was recently shown to uncouple JunD and c-Jun phosporylation from MAP kinase activation (26). Given the role of MEN1 in parathyroid tumor development, it is tempting to speculate that c-Jun could also play a role in this tumor type. In our array analysis, over-expression of c-Jun was confirmed by two independent probe sets (Table 2). Furthermore, real-time PCR showed a >twofold increased expression in 6/17 of the adenomas (35%) as compared with normal parathyroid.
QRT-PCR analyses was performed on two of the most interesting over-expressed genes and correlated well with the array results, thus supporting the accuracy of the array methodology applied. In the real-time PCR, the normalization to a housekeeping gene is currently the most acceptable method to correct for minor variations due to differences in RNA amounts or in efficiencies of reverse transcription. An ideal housekeeping gene should be expressed at an approximately constant level among different tissues and between, for example, tumor and normal samples. We found that the 36B4 housekeeping gene (coding for the acidic ribosomal phosphoprotein) is suitable for studies of normal and adenomatous parathyroid tissues.
Expression of parathyroid-related genes
In the array analyses, Group I adenomas showed lower levels of PTH expression as compared with Group II adenomas. Since the PTH gene is located in 11p15, the lower expression might be explained by a reduced dose effect following loss of a chromosome 11 homologue in Group I adenomas. Two receptors of main interest for parathyroid tumorigenesis are CASR and VDR. Both these receptors showed a distinct pattern of under-expression in all adenomas as compared with the normal parathyroids (Table 3). Both CASR and VDR participate in the calcium homeostasis controlled by the parathyroid gland, and it is suggested that the reduced expression of CASR and VDR has a major impact on the altered calcium homeostasis in PHPT. Our array data regarding CASR expression (median 63%, range 43–80%) correlate well with the mRNA in situ data previously published which also showed reduced expression in adenomas vs normal parathyroid (reduced to median 64%, range 41–98%) (10). Furthermore, in our study the VDR expression was reduced in the adenomas (to median 47%, range 20–64%) compared with the expression in normal parathyroid tissue. A similar pattern of VDR under-expression in parathyroid adenomas was also reported by Carling et al. (27). The consistent reduced levels of both VDR and CASR expression are of interest given the recent demonstration that the CASR gene promoters have vitamin D response elements (VDREs) that confer transcriptional responsiveness to 1,25-dihydroxyvitamin D levels (28).
Expression profiles in parathyroid adenomas with and without 11q13 LOH
In order to display potential parathyroid adenoma subgroups, we performed an ‘unsupervised’ clustering analysis, in which no external input was used to guide the analysis process. We used the most common unsupervised analysis method, i.e. hierarchical cluster analysis (29). This method is used to search for patterns without any expectation concerning the number or type of groups that creates the dendogram, as illustrated in Fig. 1A. The resulting hierarchical tree demonstrated that the adenomas clustered together in two individual groups. Group I consisted of the four adenomas with 11q13 LOH and Group II of the four adenomas without 11q13 LOH (Fig. 1A). In addition, several genes were detected with a significantly (P < 0.0032) distinct expression profile in Group I and II (Table 5). Among the genes over-expressed in Group I can be noted ENC1 (5q12–q13.3), which showed 20-fold increase compared with Group II. This may reflect an important downstream effect. ENC1 is over-expressed in primary colon cancer, neuroblastoma and haircell leukemia (30). Functional in vitro assays suggest that the p53 inducible ENC1 is regulated by the ß-catenin/Tcf pathway, and that its over-expression stimulates proliferation and suppresses differentiation. The present series of tumors is too limited in number to draw conclusions about clinical phenotypes and expression profiles. However, it has been suggested that sporadic and MEN 1-associated parathyroid tumors with LOH of the MEN1 locus in 11q13 are larger than tumors without this abnormality (14).
Approximately 50% of the under-expressed genes in Group I are located on chromosome 11 suggesting an underlying gene dose effect (Table 5 and Fig. 1B). This hypothesis was substantiated by the demonstration of loss of the entire chromosome 11 by CGH. The results suggest that adenomas in Group I have partly similar and partly distinguishing expression profiles as compared with the Group II adenomas. The distinct expression profiles would suggest that adenomas with 11q13 LOH partly affect different pathways as compared with the adenomas without LOH. Nevertheless, another set of genes showed similarly differential expression in the entire adenoma group as compared with normal parathyroid. This may result from late downstream effects in common for tumor development of most adenomas, e.g. over-expression of Cyclin D1 and c-Jun. The common alterations could also represent secondary effects of the tumor status or altered calcium homeostasis. The reduced expression observed for, for example CASR and VDR, would preferentially be placed in this last category. PHPT is clearly a heterogenous disease both regarding clinical development and molecular genetic alterations. Although the biological effect of the altered expression pattern is difficult to survey, it can be speculated that partly different molecular pathways could lead to benign tumor development in the parathyroid gland.
In summary, we have shown that genetic subsets in parathyroid adenomas distinguished by the 11q13 LOH status are reflected in the gene expression profiles, but that commonly altered gene expression patterns are also evident. In the future, detailed analysis at the protein level will help us to understand the impact on tumorigenesis of the genes identified in this study.
|
Acknowledgements |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. Setzer, D. Lebrecht, and U. A. Walker Pyrimidine Nucleoside Depletion Sensitizes to the Mitochondrial Hepatotoxicity of the Reverse Transcriptase Inhibitor Stavudine Am. J. Pathol., March 1, 2008; 172(3): 681 - 690. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C Juhlin, C Larsson, T Yakoleva, I Leibiger, B Leibiger, A Alimov, G Weber, A Hoog, and A Villablanca Loss of parafibromin expression in a subset of parathyroid adenomas. Endocr. Relat. Cancer, June 1, 2006; 13(2): 509 - 523. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
