Transcriptome Profiling Reveals Differentially Expressed Transcripts Between the Human Adrenal Zona Fasciculata and Zona Reticularis

Abstract

Context:

The human adrenal zona fasciculata (ZF) and zona reticularis (ZR) are responsible for the production of cortisol and 19-carbon steroids (often called adrenal androgens), respectively. However, the gene profiles and exact molecular mechanisms leading to the functional phenotype of the ZF and ZR are still not clearly defined. In the present study, we identified the transcripts that are differentially expressed in the ZF and ZR.

Objective:

The objective of the study was to compare the transcriptome profiles of ZF and ZR.

Design and Methods:

ZF and ZR were microdissected from 10 human adrenals. Total RNA was extracted from 10 ZF/ZR pairs and hybridized to Illumina microarray chips. The 10 most differentially expressed transcripts were studied with quantitative RT-PCR (qPCR). Immunohistochemistry was also performed on four zone-specific genes.

Results:

Microarray results demonstrated that only 347 transcripts of the 47 231 were significantly different by 2-fold or greater in the ZF and ZR. ZF had 195 transcripts with 2-fold or greater increase compared with its paired ZR, whereas ZR was found to have 152 transcripts with 2-fold or greater higher expression than in ZF. Microarray and qPCR analysis of transcripts encoding steroidogenic enzymes (n = 10) demonstrated that only 3β-hydroxysteroid dehydrogenase, steroid sulfotransferase, type 5 17β-hydroxysteroid dehydrogenase, and cytochrome b5 were significantly different. Immunohistochemistry and qPCR studies confirmed that the ZF had an increased expression of lymphoid enhancer-binding factor 1 and nephroblastoma overexpressed, whereas ZR showed an increased expression of solute carrier family 27 (fatty acid transporter) (SLC27A2), member 2 and TSPAN12 (tetraspanin 12)

Conclusion:

Microarray revealed several novel candidate genes for elucidating the molecular mechanisms governing the ZF and ZR, thereby increasing our understanding of the functional zonation of these two adrenocortical zones.

The zonal classification of the mammalian adrenal cortex, as seen in light microscopy, was first provided by Arnold in 1866 (1). He also coined the terms zona glomerulosa (ZG), zona fasciculata (ZF), and zona reticularis (ZR) for the three concentric zones. Since then, many researchers have demonstrated the functional relevance of these zones by providing their distinct roles in steroid hormone biosynthesis: ZG synthesizes mineralocorticoids and ZF produces glucocorticoids (2, 3). The human ZR is the site of biosynthesis of the 19-carbon (C₁₉) steroids dehydroepiandrosterone (DHEA) and DHEA sulfate (DHEAS) in the prepubertal, pubertal, and adult human (4–6). The adrenal C₁₉ steroid output that results from the expansion and differentiation of the adrenal zona reticularis in humans and some nonhuman primates is called adrenarche. The timing of adrenarche varies among primates, but in humans, serum levels of DHEAS are seen to increase at approximately 6 years of age (7, 8). Neither DHEA nor DHEAS are bioactive androgens, but they act as precursors for the production of more potent androgens, including T, in peripheral tissues that include prostate, adipose tissue, and skin (9).

Although some steroidogenic enzymes and cofactor proteins are common to all zones of the cortex, the zone-specific production of steroids results in part due to differential expression of key steroidogenic enzymes (10–13). The pathway leading to the synthesis of DHEAS is quite simple and requires only three steroidogenic enzymes, namely cytochrome P450 cholesterol side-chain cleavage (CYP11A1), CYP17 (a single enzyme catalyzing two biosynthetic activities: 17α -hydroxylase and 17,20-lyase), and steroid sulfotransferase (SULT2A1). It has been demonstrated that CYP11A1 is expressed in all zones of the adult human adrenal, whereas CYP17 is expressed in both ZF and ZR (14). Although 17α-hydroxylase activity is mandatory for production of the glucocorticoid, cortisol, in human adrenal ZF, both 17α-hydroxylase and 17,20-lyase activities are needed for C₁₉ steroid production in the ZR (15–17). Cytochrome b5 (CYB5), an allosteric regulator of CYP17, enhances the 17,20-lyase activity of CYP17 and is found to be most evident in ZR (14). SULT2A1 is also predominantly expressed in the cytoplasm of ZR (14). The pattern of CYB5 and SULT2A1 expression is thus consistent with the ability of ZR to produce DHEA and DHEAS. The hallmark of ZR is the low expression of type 2 3β-hydroxysteroid dehydrogenase (HSD3B2) (12, 18, 19). The relative lack of HSD3B2 expression/activity facilitates increased DHEA and DHEAS synthesis because HSD3B2 competes with CYP17 and SULT2A1 for pregnenolone and 17α-hydroxypregnenolone (20, 21). It was also recently demonstrated that the adrenal ZR is also able to synthesize the potent androgen T owing to the higher expression of type 5 17β-hydroxysteroid dehydrogenase (AKR1C3) in ZR as compared with ZF (22, 23).

Beyond steroidogenic enzymes and cofactor proteins, little is known about the differences in phenotypes of ZF and ZR. A handful of genes have been defined to have distinct expression patterns between these two zones using cDNA probe arrays for approximately 750 genes (6). However, the molecular mechanisms governing the distinct steroidogenic phenotype of the two zones have not been defined. In the present study, we have sought to identify the transcripts that are differentially expressed in the human adrenal ZF and ZR using visual microdissection, microarray, quantitative RT-PCR (qPCR), and immunohistochemistry. To better understand the biological aspect of the observed differences in gene expression, we also analyzed the same microarray data using gene ontology (GO) and pathway analyses. This approach has revealed several novel candidate genes for elucidating the molecular mechanisms governing the ZF and ZR, thereby increasing our understanding of the functional zonation of these two adrenocortical zones.

Materials and Methods

An expanded Materials and Methods section with statistical analyses is provided as a Supplemental File, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org.

Human tissue preparation

After informed consent from the family, human adult adrenal glands were obtained from cadaveric kidney donors that were transplanted at the Georgia Regents University (Augusta, Georgia). Use of these tissues was approved by the Institutional Review Board of Georgia Regents University.

Isolation of ZF and ZR tissue from human adult adrenals

Adrenal glands (n = 10) were trimmed free of fat and placed in culture medium DMEM/F14 medium (Gibco, Life Technologies), and ZF and ZR tissue was isolated by microdissection of the adrenal gland as previously described (6, 18).

RNA extraction and gene expression assays

RNA extraction and gene expression assays were performed as described previously (24).

Microarray analysis for human adrenal ZF and ZR cells

Total RNA extracted from ZF and ZR fragments was hybridized to an Illumina bead chip. The arrays were scanned at high resolution on the iScan system (Illumina). Results were analyzed by GeneSpring GX (version 12.1) software (Silicon Genetics) by customizing to the Illumina single-color analysis. To identify the differences among the two zones, a list of ZF and ZR markers was created by using transcripts satisfying the conditions including fold expression differences and statistical differences between the respective zones. Differences in the GO term were defined based on a list of genes having a 1.5-fold or greater and P < .05 difference in ZF and ZR expression using GeneSpring GX version 12.1 software. For the pathway analysis, a list of transcripts up-regulated in ZF and ZR was created (≥1.25-fold and P < .05).

Immunohistochemistry

Immunohistochemical analysis of lymphoid enhancer-binding factor 1 (LEF1), nephroblastoma overexpressed (NOV), solute carrier family 27 (fatty acid transporter), member 2 (SLC27A2) and tetraspanin 12 (TSPAN12) was performed using the streptavidin-biotin amplification method using a Histofine kit (Nichirei) or Envision+ kit (DAKO) as previously described (25).

Data analysis and statistical methods

Results are given as mean ± SEM where appropriate. Statistical analyses for microarray were done by a paired t test, followed by Benjamini and Hochberg false discovery rate correction. Significance was accepted at the level of probability of P = .0-.05. Statistical analyses for qPCR were done by paired t test followed by post hoc correction. Significance was accepted at the level of probability of P = .0-.05.

Results

Differentially expressed transcripts in ZF and ZR

Principal component analysis (PCA) showed that samples from the same group clustered together and thus indicated that the microarray data is highly reproducible (Figure 1A). Of the 47 231 probe sets on the Illumina array, the scatterplot showed that only 347 transcripts were significantly different by 2-fold or greater between the ZF and ZR (P < .05). On paired analysis, ZF was found to have 195 transcripts with a 2-fold or greater increase compared with ZR, whereas ZR had 152 transcripts with a 2-fold or greater higher expression than in ZF (P < .05) (Figure 1B). Some of these up-regulated transcripts could be involved in zone-dominant steroid hormone synthesis.

Figure 1.

A, PCA of microarray data. PCA three-dimensional scatterplot uses the expression of the all of the 47 231 independent transcripts on the normalized microarray platform [diamonds in red and spheres in green represent individual ZF and ZR samples (n = 10 each)]. B, Microarray scatterplot showing pooled data for ZF and ZR from 10 adrenals. Each spot indicates a unique probe with a total of 47 231 independent transcripts based on normalized microarray data. Spots outside the parallel lines represent transcripts with greater than 2-fold differences in expression. The transcripts involved in steroidogenesis are indicated as black circles.

Transcripts encoding genes involved in steroidogenesis in ZF and ZR

As expected, signal intensities of transcripts that are involved in both cortisol and C₁₉ steroid synthesis, namely steroidogenic acute regulatory protein, CYP11A1, 11β-hydroxylase, and CYP17 were highly and equally expressed in ZF and ZR (Figure 2A). P450 oxidoreductase was also expressed equally in the two zones, although the expression levels were relatively low.

Figure 2.

A, Heat map representation of microarray analysis for transcripts encoding genes involved steroidogenesis in ZF vs ZR (n = 10). Normalized log₂ signal intensity acquired from microarray analysis is represented by color (see color bar). Fold change was calculated as the average of normalized ZF signal intensities divided by that of normalized ZR signal intensities. Data are represented as mean ± SEM. FC, fold change. *, P < .01, by paired t test. B, Validation of microarray analysis by real-time qPCR for selected transcripts encoding genes involved in steroidogenesis in ZF vs ZR. CYP11A1 and CYP17 remain unchanged, whereas HSD3B2, CYB5, AKR1C3, and SULT2A1 were found to be significantly different between ZF and ZR. #, P < .01, §, P < .005, **, P < .001, ##, P < .05, by paired t test. Ten ZF-ZR pairs were used for the qPCR analysis.

The transcripts encoding genes involved in steroidogenesis that were found to be significantly different in the two zones were HSD3B2 (higher in ZF), and CYB5, SULT2A1, and AKR1C3 (higher in ZR) as shown by microarray (Figure 2A) and confirmed by qPCR (Figure 2B). Previous studies have also shown the zonal specificity of these genes by immunohistochemistry (12, 20). We also confirmed the low level distribution of CYP11A1 and CYP17 between the two zones by qPCR (Figure 2B).

Transcripts up-regulated in ZF

Most of the transcripts for proteins with the highest differential expression in ZF vs ZR have not been previously studied with regard to adrenal physiology and function (Figure 3A). In our list, a transcription factor involved in the Wnt signaling pathway, LEF1, was the most up-regulated ZF transcript when compared with ZR. This was followed by plasma membrane calcium ATPase isoform 3 (ATP2B3), NOV, and Purkinje cell protein 4 (PCP4). HSD3B2 was also one of the transcripts with the highest differential expression in ZF as compared with ZR. The relative signal intensities between the 10 human ZF and ZR samples were consistently uniform through the 15 transcripts and the respective zones, suggesting consistent visual dissection of the adrenal tissue.

Figure 3.

A, Heat map representation of microarray analysis for the 15 transcripts with the highest differential expression in ZF vs ZR (n = 10). Normalized log₂ signal intensity acquired from microarray analysis is represented by color (see color bar). Fold change was calculated as the average of normalized ZF signal intensities divided by that of normalized ZR signal intensities. Data are represented as mean ± SEM. FC, fold change. All transcripts showed a significance of P < .05. B, Quantitative RT-PCR and immunohistochemical analysis for ZF-expressed LEF1 and NOV. Validation of microarray analysis was done by real-time qPCR (a and d) and immunohistochemistry (b, c, e, and f) for transcripts with high differential expression in ZF vs ZR, namely LEF1 and NOV. Both LEF1 and NOV showed a higher expression in the outer ZF as compared with the inner ZF and ZR. Ten ZF-ZR pairs were used for the qPCR analysis. §, P < .005, **, P < .001, by paired t test.

We also performed qPCR to confirm the difference in zonal expression for two transcripts with high differential expression in the ZF (Figure 3B). The protein expression was also examined by immunohistochemistry. HSD3B2 was used as the positive control for positive ZF staining. LEF1 and NOV were clearly defined in ZG (data not shown) as well as the outer ZF but not in the inner ZF and ZR (Figure 3B). The degree of differences in the expression of LEF1 and NOV was found to be higher with qPCR than immunohistochemistry. The most obvious explanation would be to conclude that mRNA and protein values do not always agree, but comparisons between mRNA and immunohistochemistry are likely more complex. Although immunohistochemistry detects the functional product of the biomarker gene, it may be influenced by tissue fixation as well as potential lack of specificity and sensitivity of the antibody. On the other hand, qPCR may be a good indicator of mRNA levels. ATPase isoform 3 and Purkinje cell protein 4 were found to be highly expressed in the ZG as opposed to ZF and ZR by immunohistochemistry (data not shown).

Transcripts up-regulated in ZR

Other than CYB5, none of the other 15 ZR-dominant transcripts have been associated with adrenal function (Figure 4A). Fibrinogen-γ (FGG) chain was the most up-regulated transcript in the ZR followed by transcripts encoding two transporters, solute carrier family 13 (sodium-dependent citrate transporter), member 5 (transporting sodium citrate) and SLC27A2 (transporting fatty acids). In addition, the steroidogenic enzymes SULT2A1 and AKR1C3 were also found to be more highly expressed in the ZR compared with the ZF (∼5.5- and 3-fold, respectively, as found by qPCR; data not shown). We confirmed the zonal difference in expression for SLC27A2 and TSPAN12 in the ZR by qPCR (Figure 4B). Additionally, immunohistochemistry was performed for three transcripts (CYB5, SLC27A2, and TSPAN12) based on the availability of antibodies. CYB5 was used as the positive control for positive ZR staining. SLC27A2 and TSPAN12 were expressed preferentially in ZR (Figure 4B). Thus, we obtained consistent results from microarray, qPCR, and immunohistochemistry.

Figure 4.

A, Heat map representation of microarray analysis for the 15 transcripts with the highest differential expression in ZR vs ZF (n = 10). Normalized log₂ signal intensity acquired from the microarray analysis is represented by color (see color bar). Fold change was calculated as the average of normalized ZF signal intensities divided by that of normalized ZR signal intensities. Data are represented as mean ± SEM. FC, fold change. All transcripts showed a significance of P < .05. B, Quantitative RT-PCR and immunohistochemical analysis for ZR-expressed SLC27A2 and TSPAN12. Validation of microarray analysis was done by real-time qPCR (a and c) and immunohistochemistry (b and d) for transcripts with high differential expression in ZR vs ZF, namely SLC27A2 and TSPAN12. Ten ZF-ZR pairs were used for the qPCR analysis. §, P < .005, **, P < .001, by paired t test.

GO analysis of zonally expressed genes

GO analysis was done using 1187 probe sets that were significantly up- or down-regulated (≥1.5-fold, P < .05) in ZR (Table 1). A total of 11 GO classification terms were found to be significantly (P < .01) overrepresented among these 1187 transcripts: six terms were related to cellular components, one term to the molecular function category, and four terms to the biological process category. Interestingly, four of the 11 GO terms were associated with major histocompatibility complex (MHC) class II antigen processing. This trend suggests that the expression of MHC class II molecules likely differs in fundamental ways between the ZF and ZR.

Table 1.

GO Categories Significantly (P < .01) Overrepresented Among the 1187 Probe Sets Shown to Be Up- or Down-Regulated 1.5-Fold or Greater in ZR (P < .05)

GO Accession	GO Term	Corrected P Value	Probes
			Present	Total
Cellular Component
GO:0042613	MHC class II protein complex	2.26E-07	11	16
GO:0042611	MHC protein complex	6.74E-04	12	39
GO:0005576	Extracellular region	3.06E-04	134	1788
GO:0016020	Membrane	3.06E-04	381	6525
GO:0044421	Extracellular region part	0.002	67	754
GO:0044459	Plasma membrane part	0.005	106	1432
Molecular function
GO:0032395	MHC class II receptor activity	2.26E-07	10	13
Biological process
GO:0002504	Antigen processing and presentation via MHC class II	3.19E-06	11	20
GO:0006629	Lipid metabolic process	3.06E-04	69	735
GO:0046165	Alcohol biosynthetic process	0.002	13	51
GO:0044699	Single-organism process	0.003	349	6043

Pathway analysis of ZF vs ZR

A list of genes defined by microarray to be up-regulated in ZF and ZR (≥1.25-fold and P < .05) was exported for pathway analysis. There were 92 pathways found to be significantly different in the two zones (P < .01) (Supplemental Table 1). To determine the pathways up-regulated in either zone, a list of genes in the selected pathways was made and their zonal expression was compared (P < .05) (Supplemental Table 2). Pathways involving cell proliferation (namely, Wnt signaling, canonical Wnt signaling, and regulation of telomerase) were up-regulated in ZF. Those involving cell death (namely, Fox O family signaling, proteasome degradation, senescence and autophagy, apoptosis modulation and signaling, oxidative stress, and glutathione metabolism) were up-regulated in ZR (Supplemental Table 2). Similarly, steroid metabolism genes (namely, cholesterol biosynthesis, sulfation biotransformation reaction, metabolism of lipids and lipoproteins, and glutathione metabolism) were up-regulated in ZR.

Discussion

Using microarray, GO and pathway analyses, we have identified several novel candidate genes that could help in elucidating the molecular mechanisms governing the differences between ZF and ZR.

Our microarray analysis identified several steroidogenic genes that are known to be differentially expressed between ZF and ZR, namely HSD3B2, CYB5, AKR1C3, and SULT2A1. Several other studies have shown results consistent with our data (6, 12, 14, 16, 17, 22). Wang et al (6) showed that an aldo-keto reductase, AKR1B1, was among the highly expressed genes in ZR. In our study, the only highly expressed aldo-keto reductase was AKR1C3 (2.1-fold), which is known to convert androstenedione to T. This observation agrees with the results obtained by Nakamura et al (22). Pathways analyzed in our studies revealed that the sulfation biotransformation reaction pathway was higher in the reticularis, with two genes that were significantly up-regulated (namely, SULT2A1 and PAPSS2). The sulfation of steroids by SULT2A1 requires a sulfate donor, 3′-phosphoadenosine 5′-phosphosulfate (PAPS) (26). In humans, PAPS synthesis requires two isoforms of the enzyme PAPS synthase, PAPSS1 and PAPSS2 (26). Our pathway analysis of steroidogenesis showed that cholesterol and lipid metabolism is elevated in ZR. Glutathione metabolism was also elevated in ZR. Previously the high expression of the glutathione-S-transferases in the adrenal was shown to be attributed to their participation in the metabolism of peroxides and the subsequent presence of lipofuscin in ZR, thereby giving it a reddish brown color (27–29). Recently it was demonstrated that these enzymes also play a role in the metabolism of Δ5 steroids (28, 30). Another interesting transcript that was up-regulated in the ZR was SLC27A2 (also called FATP2), a fatty acid transport protein, which confers increased ZR fatty acid uptake and also exhibits fatty acyl-CoA synthetase activity (31). SLC27A2 has been shown to mediate the channeling of exogenous long-chain fatty acids into phosphatidylinositol (32, 33). The phosphatidylinositol thereby produced could participate in cell membranes or signaling mechanisms within the ZR cell. However, further studies are needed to test the physiologic role of SLC27A2 in the ZR.

Adrenal development and zonation are generally thought to be driven by the centripetal migration theory. According to this theory, a common pool of progenitor cells located in the capsule gives rise to each zone by migrating inward, at first differentiating into ZG cells and then followed sequentially by the development of the ZF and ZR phenotypes; during this process, the cells senesce and die through an apoptotic mechanism (34, 35). As aging progresses, the size of ZR decreases, whereas that of the outer cortical zones increases (36, 37), thereby indicating that the adrenal tissue remodeling is attained by a balance between cell proliferation in the ZG and ZF, and apoptosis in the ZR (38, 39). Our pathway analyses also indicated that cell proliferation was up-regulated in the ZF, whereas pathways related to cell death were up-regulated in the ZR.

Wnt/β-catenin signaling pathways have been shown to play a very important role in adrenal development and homeostasis, in humans as well as mice (40–44). In humans, dysregulation of Wnt signaling has been found in a subset of sporadic adrenocortical adenomas and carcinomas (45, 46). In mice, the subcapsular region of the adrenal, which differentiates into ZG, shows higher β-catenin expression as well as more active Wnt/β-catenin signaling (43). In human adrenals, although β-catenin was detected throughout the entire cortex, activation of β-catenin, as shown by cytoplasmic and nuclear accumulation, was seen in ZG (45). It has also been proposed that Wnt4, a factor in the Wnt signaling pathway, may be needed for proper formation and function of ZG and that its deficiency decreases aldosterone production in mice (42). In humans, ZG was shown to exhibit higher expression of DKK3, a gene from the dickkopf family that acts as a morphogen in Wnt signaling. However, other Wnt signaling-related proteins like frizzled 2 and dishevelled showed no significant difference between ZG, ZF, and ZR (44).

Our pathway analysis showed that components of the canonical and noncanonical Wnt signaling mechanism, namely LEF1, Axil/conductin (AXIN)-2, cyclin D (CCND)-1, and CCND2 were up-regulated in ZF and ZG. LEF1 is a transcription factor that initiates the transcription of Wnt-responsive target genes involved in proliferation, differentiation, and survival, including c-myc and CCND1 (40, 47, 48). AXIN2 is a scaffold protein forming a part of the β-catenin destruction complex glycogen synthase kinase-3β/AXIN/adenomatous polyposis coli and is known to mediate the degradation of excess β-catenin and thereby maintain homeostasis (49). CCND1 and CCND2 are members of the highly conserved cyclin family involved in cell cycle regulation by positively coordinating cell cycle transition from the G₁ phase to the S phase (50), thus promoting cell proliferation. Telomerase and the shelterin complex control adrenocortical homeostasis by the maintenance of telomeres in the proliferative adrenal subcapsular cells (51–53). Telomerase extends telomeres by addition of TTAGGG nucleotide repeats to the 3′-terminus of the chromosome, whereas shelterin protects the telomere from DNA damage (54). Interestingly, our pathway analysis also determined that some of the components of the shelterin complex, namely, TERF1 (telomeric repeat binding factor), TERF2, and TINF2 (TERF1-interacting nuclear factor), are significantly elevated in ZF. Thus, it can be speculated that along with ZG, cell proliferation may also occur in ZF to a limited extent. TSPAN12, a member of the phylogenetically ancient tetraspanin family, which is involved in norrin- but not Wnt-mediated frizzled 4/β-catenin signaling (55) is up-regulated in the ZR. TSPAN12 has been previously shown to play a role in promoting tumor growth (56), and it can be thus speculated that whatever little cell proliferation occurs in the ZR is norrin and not Wnt induced. Also, it is interesting that NOV, a growth-regulatory protein, was up-regulated in the ZG and ZF. It has been identified as a definitive zone marker of the human fetal adrenal and has been shown to inhibit growth while promoting differentiation (57). It has also been demonstrated that NOV is down-regulated in childhood adrenocortical tumors (58). It would therefore be interesting to identify a role for NOV in the ZG and ZF, as suggested by our immunohistochemistry studies.

Several studies have demonstrated that ZR is the zone of cell aging and senescence (36, 37, 59, 60). However, Wolkersdorfer et al (61) have shown that the highest apoptotic index was detected in the outermost zones of the human adrenal cortex, predominantly ZG. Sasano et al (39) studied cell proliferation by Ki67 immunostaining and apoptosis by a 3′-hydroxy nick end labeling technique and demonstrated that Ki67 immunoreactivity was predominantly observed in ZF, whereas cortical cells positive for nick end labeling were present in ZR and, in some cases, in ZG. The increased occurrence of cell aging in ZR has also been histologically shown by the increased presence of lipofuscin in ZR (62–64). The formation of lipofuscin was attributed to the high expression of the glutathione-S-transferases in ZR (28, 30), which was also confirmed by our pathway analysis.

One possible mechanism contributing to ZR apoptosis is the higher expression of MHC class II molecules in ZR cells, thereby priming them for eradication by the immune system (65). Our study demonstrated that several genes encoding components of the MHC are higher in ZR including HLA-DMA, HLA-DPB1, HLA-DRB3, HLA-DRB6, HLA-DRB4, and HLA-DRB6 (data not shown). Our GO analysis also showed that the expression of MHC class II molecules differed in fundamental ways between ZF and ZR. Earlier studies have confirmed that human leukocyte antigen (HLA) class II molecules are restricted to ZR cells (6, 66, 67). It has been shown that MHC class II expression and the maturation of the human adrenal cortex occur concomitantly (67). Thus, it has been proposed that MHC class II molecules facilitate interactions between cells in the ZR and immune cells, such as lymphocytes, that may regulate adrenal androgen production (68). Other studies have shown a related increase in the ZR-specific T lymphocytes during aging (69). The expression of the MHC class II antigens on ZR cells may also result from the effect of cytokines released by macrophages or other immune cells within the adrenal gland (70). However, Khoury et al (66) have shown the presence of expression of the MCH class II antigens on ZR cells in culture, although they were mixed with ZF cells, which remained MCH class II negative. These results suggested that the presence of MHC class II molecules on ZR cells was an inherent property of ZR rather than the effect of cytokines released by immune adrenal cells. Furthermore, a direct cell-to-cell contact has also been demonstrated between MHC class II molecules expressing ZR cells and T lymphocytes (71). The presence of the HLA antigens could play a role in differentiation or alternatively make the ZR more susceptible to apoptosis via immune attack.

Our pathway analyses shed more light on the cell death occurring in ZR. We have shown that ZR shows an increase in certain components of the cell death machinery including pathways involving Fox O family signaling (FoxO4), proteasome degradation (UCHL1 [ubiquitin carboxyl-terminal esterase L1] and PSMB8 [proteasome (prosome, macropain) subunit, beta type 8]), senescence and autophagy (ING2 [inhibitor of growth family, member 2] and lysosomal-associated membrane protein-2), and apoptosis signaling (TNF-related apoptosis-inducing ligand). FoxO4, a member of the forkhead transcription factor, has been found to promote apoptosis by negatively regulating the cell cycle (72) and inducing proteasome activity (73). UCHL1 is a ubiquitin-protein hydrolase involved in both the processing of ubiquitin precursors and ubiquitinated proteins and has been shown to prevent the degradation of free monoubiquitin in lysosomes (74), whereas PSMB8 is a component of the proteasome (75). Likewise, ING2 has been shown to be involved in cellular senescence (76). The role of lysosome-associated membrane protein-2 in the maintenance of the lysosome and that of TNF-related apoptosis-inducing ligand in inducing apoptosis has also been well characterized (77, 78).

In summary, our study helped to identify transcripts that are differentially expressed in the human adrenal ZF and ZR and provided a broad understanding of pathway differences in ZF and ZR. Detailed analysis of the newly defined gene differences should help to clarify the mechanisms leading to functional differences between the ZF and ZR.