Isolation of Definitive Zone and Chromaffin Cells Basedupon Expression of CD56 (Neural Cell AdhesionMolecule) in the Human Fetal Adrenal Gland

MARCUS O. MUENCH, JENNIFER V. RATCLIFFE, MIKIYE NAKANISHI, HITOSHI ISHIMOTO, AND

ROBERT B. JAFFE

Department of Laboratory Medicine (M.O.M.) and Center for Reproductive Sciences (J.V.R., M.N., H.I., R.B.J.), Departmentof Obstetrics, Gynecology and Reproductive Sciences, University of California, San Francisco, San Francisco, California94143

The cortex of the human midgestation adrenals comprises athin layer of cells, the definitive zone (DZ) that surrounds thelarger, inner fetal zone (FZ). CD56 expression was observed byimmunohistochemistry on DZ cells and isolated groups ofcells within the FZ. CD56 mRNA expression also was detectedamong DZ cells but not selected sections of FZ cells isolated bylaser capture microdissection. Flow cytometric analysis ofdispersed adrenal cells indicated CD56 expression on a subsetof adrenal cells lacking expression of hematopoietic (CD45�

and CD235a�) and endothelial (CD31�) cell markers. TheCD56�CD31�CD45�CD235a� cells were isolated by discontin-uous-gradient centrifugation and fluorescence-activated cell

sorting. The purified cells were enriched for DZ cells based onexpression of mRNA for metallopanstimulin-1, nephroblas-toma overexpressed gene, and 3-�-hydroxysteroid dehydro-genase. P450c17 mRNA expression also was detected among asubset of CD56� cells consistent with expression of low levelsof this protein in some DZ cells. The presence of a subpopu-lation of chromaffin cells among the CD56� population alsowas shown by the expression of chromogranin A mRNA. Thesefindings indicate that CD56 expression can be used to isolateDZ and chromaffin cells to further study their functional anddevelopmental properties. (J Clin Endocrinol Metab 88:3921–3930, 2003)

HUMAN TISSUES COMPRISE a large number of dif-ferentiated cell types that are integrated to form the

structure and perform the specialized functions required ofeach tissue. Because many mature cells have lost the capacityto grow, replacement of mature cells that have died occursin most tissues in response to developmental changes, thenormal aging process, disease, or damage. Progenitors havebeen characterized in most tissues that have a limited ca-pacity for growth and differentiation but are capable of re-newal of specific cell types within tissues (1–6). Stem cells,long known to exist for tissues such as the hematopoieticsystem, have been recently identified in many tissues (7–9).Stem cells are characterized by an extensive proliferativecapacity and capable of differentiating into many cell typesincluding, in some cases, cells foreign to their tissue of origin.Molecular and cytogenetic techniques are now revealing theprocesses that occur as stem cells and progenitors becomespecialized, fully differentiated cells. To characterize theseevents, markers for each cell type in the system are necessary.

Our interest is in understanding the developmental pro-cesses inherent in the human fetal adrenal gland. In the earlystages of organ development, some fetal organs do not re-capitulate the structural or functional organization of the

adult. The human adrenal gland is exceptional in that thefetal structure persists throughout gestation and its func-tional adult organization does not occur until after birth. Thehuman fetal adrenal also has diverged markedly from othermammals, with only some higher primates and few otheranimals (armadillo, sloth) having a similar adrenal architec-ture. Because the fetal adrenal gland is necessary for main-tenance of intrauterine homeostasis, induction of enzymes inorgans essential for extrauterine life and possibly involved inthe initiation of parturition, elucidating fetal adrenal devel-opmental biology, is of particular importance.

The human fetal adrenal gland arises from the coelomicridge. Quite rapidly, three cell types are apparent: 1) a thinlayer encapsulating the gland; 2) two to three cell layers ofsmall, tightly packed basophilic cells; and 3) a large innerportion (�80% by volume) of large cells with lipid-filledvacuoles. These layers are the capsule, definitive zone (DZ)and fetal zone (FZ), respectively. Later in gestation, the tran-sitional zone (TZ) will become apparent between the DZ andFZ (10). From early in development (6–8 wk), FZ cells ex-press P450 side-chain cleavage enzyme (P450scc) and P45017�-hydroxylase/17–20 lyase (P450c17). This complement ofsteroidogenic enzymes permits synthesis of large amounts ofdehydroepiandrosterone and dehydroepiandrosterone sul-fate. The cells lack 3-�-hydroxysteroid dehydrogenase (3-�HSD), however, and thus are not capable of synthesizingother steroid products independently. Rather, the dehydro-epiandrosterone and dehydroepiandrosterone sulfate areconverted to estrogens by the placenta in a unique cooper-ative effort between the placenta and fetus, the fetoplacentalunit. In addition to P450c17 and P450scc, 3-�HSD and

Abbreviations: Chr-A, Chromogranin A; DZ, definitive zone; FACS,fluorescence-activated cell sorting; FITC, fluorescein isothiocyanate; FZ,fetal zone; GAPDH, glyceraldehyde-3-phosphate dehydrogenase;3-�HSD, 3-�-hydroxysteroid dehydrogenase; LDLR low-density li-poprotein receptor; lin�, lineage�; mAb, monoclonal antibody; Mps-1,metallopanstimulin-1; NovH, nephroblastoma overexpressed gene; PE,phycoerythrin; PI, propidium iodide; P450scc, P450 side-chain cleavageenzyme; TZ, transitional zone.

0021-972X/03/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 88(8):3921–3930Printed in U.S.A. Copyright © 2003 by The Endocrine Society

doi: 10.1210/jc.2003-030154

3921

P450c11 are expressed during the latter part of gestation inthe TZ and DZ, and aldosterone synthase is expressed in theDZ close to term.

In contrast to the FZ, before the third trimester, the DZ cellsexpress only P450scc and do not have the capacity to producesteroids (11). This has led to the hypothesis that this thin rimof cells may represent a progenitor population in the humanfetal adrenal gland. We believe that these cells proliferate anddifferentiate first into cells of the FZ and later in gestation intocells of the TZ and FZ. Finally, just before birth some DZ cellstake on characteristics of the adult zona glomerulosa andproduce aldosterone. This hypothesis of DZ cells serving asprogenitors is generally accepted yet lacks direct evidence.Recently, we used laser capture microdissection to isolate DZcells (11). The cDNA library prepared from this relativelypure population of cells was used for subtractive hybridiza-tion to identify new markers unique to DZ and FZ cells.Expression of nephroblastoma overexpressed gene (NovH)and the gene for the ribosomal protein metallopanstimulin-1(Mps-1) was found among DZ cells, whereas the low-densitylipoprotein receptor gene (LDLR) was enriched among FZcells (12). In the present study, we describe CD56 (neural celladhesion molecule) to be a cell-surface marker on DZ cells,which can be used to isolate DZ cells by fluorescence-acti-vated cell sorting (FACS). Expression of CD56 is extensive indeveloping neuronal and endocrine tissues, including the ratadrenal (13–16) and also is expressed by NK cells and someT cells (17, 18). CD56 belongs to the Ig superfamily of ad-hesion molecules and plays a role in the morphogenesis ofseveral organ systems including the liver (19, 20), pancreas(21, 22), and neuromuscular junction (23). Expression ofCD56 has been noted in both developing fetal organ systems(20) and physiologic and regenerative processes in adults (24,25). Its role in these processes has been suggested to rangefrom altering migration to initiating differentiation and stim-ulation of signaling cascades. The use of this and other cellsurface markers has allowed us to greatly enrich DZ cells forin vitro studies of adrenocortical cell differentiation.

Materials and MethodsTissues

Adrenals glands were obtained from human fetuses after therapeutictermination of pregnancy. This study was approved by the Committeeon Human Research, University of California, San Francisco (UCSF). Thegestational ages ranged from 15 to 24 wk. Adrenal glands were obtainedand placed in PBS without Ca2� or Mg2� with 2% fetal bovine serum forcell separation experiments or placed in 4% paraformaldehyde whenintended only for histological examination. Adrenals were kept on icefor transport to the laboratory.

Antibodies

The following fluorescein isothiocyanate (FITC) and phycoerythrin(PE) labeled monoclonal antibodies (mAbs) were purchased from BDBiosciences (San Jose, CA): CD31-FITC (WM59), CD56-PE (MY31),mouse IgG1-FITC, mouse IgG2a-FITC, and mouse IgM-FITC. CD56-FITCand CD56-PE (C5.9) were purchased from Exalpha Corp. (Boston, MA).CD45-PE (HI30), mouse IgG1-FITC, mouse IgG1-PE, mouse IgG2a-PE,mouse IgG2b-FITC, and mouse IgM-FITC were purchased from Caltag(Burlingame, CA). The following conjugated mAbs were purchasedfrom Beckman-Coulter (Miami, FL): CD31-PE (1F11), CD34-FITC (581),CD36-FITC (FA6.152), CD45-FITC (KC56), CD235a-FITC (11E4B-7–6),and CD235a-PE (11E4B-7–6). Polyclonal rabbit anticytochrome P450c17

antibody was kindly provided by Dr. Walter Miller (UCSF) (26). Poly-clonal sheep antihuman CD31 was obtained from Research and Diag-nostic Systems Inc. (Minneapolis, MN) and used for the confocal mi-croscopy experiments.

Immunohistochemistry

Fetal adrenal glands were placed in 4% paraformaldehyde overnight,followed by incubation in 30% sucrose in PBS at 4 C overnight. Tissueswere then embedded in Tissue-Tek OCT compound from Sakura Finetek(Torrance, CA) and frozen in a dry ice ethanol bath. Samples were storedat �80 C until use. Ten-micrometer sections were prepared and used forimmunofluorescence.

Tissue sections were processed according to the method of Basora etal. (27). Autofluorescence was prevented by incubating sections in 0.02 mglycine in PBS with or without 0.1–0.3% Triton X-100, 10% nonfat milkfor 30 min, and 10% goat or donkey serum for 30 min. Primary antibodyincubation was performed with 1:10 dilutions of FITC-conjugated CD31,CD34, CD36, or CD56 mAbs or a 1:300 dilution of anti-P450c17 for 1 hat room temperature. After washing, incubation with a fluorochrome-labeled secondary antibody was performed at room temperature for30 min with either Cy3-conjugated goat antimouse or goat antirabbitantibody (Jackson Laboratories, West Grove, PA) for single color im-munohistochemical studies. For the confocal microscopy experiments,FITC-conjugated goat antimouse antibody (Jackson Laboratories) wasused to detect CD56 staining, and Cy3-conjugated goat antirabbit wasused to detect anti-P450c17 antibody. Cy5-conjugated donkey antisheepantibody (Jackson Laboratories) was used in the confocal microscopyexperiments to detect antibody bound to CD31. After washing, slideswere mounted with Vectorshield (Vector Laboratories, Inc., Burlingame,CA) and examined with a DMR fluorescent microscope (Leica, Quebec,Canada) and a HQ-TRITC filter (Chroma, Brattleboro, VT). Pictureswere taken with a DCS 430 digital camera (Eastman Kodak Co., Roch-ester, NY) and transferred to Adobe Photoshop software (Adobe Sys-tems Inc., San Jose, CA). Confocal microscopy was performed using a510 META laser-scanning microscope (Carl Zeiss Inc., Jena, Germany).Control sections were stained with unconjugated mouse or rabbit IgGor with FITC-conjugated mouse IgG. Background staining using thesecontrols under the conditions described was minimal.

Laser capture microdissection

Laser capture microdissection was performed as described previ-ously (12), using the Pixcell II (Arcturus Engineering, MountainView, CA).

Preparation of RNA, RT-PCR, and real-time quantitativeRT-PCR

RNA was prepared by one of two methods chosen by estimating theamount of RNA to be extracted. For quantities of approximately 5 �g,Mini RNeasy (QIAGEN, Valencia, CA) was used according to the man-ufacturer’s instructions, followed by use of DNA-Free from Ambion(Woodward, TX). For quantities less than 5 �g, a mini-RNA isolation kit(Zymo Research, Orange, CA) was used as instructed, followed by theirDNA-Free RNA kit. Before RT of the RNA, samples were shown to befree of DNA by a RT-PCR reaction using glyceraldehyde-3-phosphatedehydrogenase (GAPDH) oligonucleotides. RT with random primerswas carried out with M-MLV reverse transcriptase (Invitrogen Corp.,Carlsbad, CA).

The following sense (antisense) oligonucleotides were used: GAPDH,GATGACATCAAGAAGGTGGTG (CTCCTTGGAGGCCATGTGGGC-CAT); NovH, CTAAGTGGACTGGTGTCATAAC (ATTTGAAACAGC-TATCAGAGGG); Mps-1, GATCTCCTTCATCCCTCTCC (GTTTC-CCACTCATCTTGACTC); P450c17, CTTCAAGCTGCAGAAAAAA-TATGG (CAATGTACTGATTTCCTGACAAAT); 3-�HSD, CCTCAC-CAAAGCTATGATAACC (TCCTAACAATACCCACATGCAC); LDLR,TCTAAGCCAAACCCCTAAACTC (CAACACACACGACAGAAAA-CAG); CD31, TCACCATCCAGAAGGAAGATAC (ACCCTCAGAAC-CTCACTTAAC); CD56, TATTTGCCTATCCCAGTGCC (CATACT-TCTTCACCAACTGCTC); CD90, GCTGCTTCTGTCTGGTTTATTTAG(CCTCATCCTTTACCTCCTTCTC); and Chromogranin A (Chr-A)CATCTCCGACACACTTTCCAAG (TCCTCTCTTTTCTCCATAACA-

3922 J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 Muench et al. • CD56 Expression by Definitive Zone Cells

TCC). All oligonucleotides were purchased from Sigma-Genosys (TheWoodlands, TX), and each oligonucleotide was used at 10 pm/�l inconjunction with Ready-To-Go PCR beads (Amersham BiosciencesCorp. Piscataway, NJ) for 35 cycles as recommended. RT-PCR was per-formed on a Gene Amp PCR System 9600 (Perkin-Elmer, Boston, MA).

Expression of 3-�HSD CD56�lin�, CD56�lin� cells was further ana-lyzed using the 5� nuclease assay (real-time TaqMan RT-PCR) as we havedescribed previously (28). Relative expression levels were calculated as2-(Ct 3-�HSD - Ct GAPDH) using GAPDH as an endogenous control gene. Se-quences for the PCR primers and TaqMan probes were: 3-�HSD forward,TCACAGAGAGTCCATCATGAATGTC; reverse, CGGCTACCTCTAT-GCTACTGGTGTA; TaqMan probe, FAM(6-carboxy-fluorescein)-TGAAAGGTACCCAGCTACTGTTGGAGGC-TAMRA(6-carboxy-tetramethyl-rhodamine) (Integrated DNA Technologies, Coralville, IA);GAPDH forward, ATTCCACCCATGGCAAATTC; reverse, TGG-GATTTCCATTGATGACAAG; TaqMan probe, FAM-ATGGCACCGT-CAAGGCTGAGAACG-TAMRA (Integrated DNA Technologies).

Isolation of human fetal adrenal cells

Fetal adrenals were mechanically dispersed and then treated with 3�g/ml Liberase Blendzyme 2 (Roche Molecular Biochemicals, India-napolis, IN) at 37 C for 20 min. Cells were filtered into media containing10% fetal bovine serum to remove aggregates and then concentrated bycentrifugation. The cells were then suspended in 3 ml PBS supplementedwith 0.3% BSA and 0.01% NaN3 and overlaid on 7 ml NycoPrep 1.077(Greiner-Bio-One, Inc., Longwood, FL), centrifuged for 25 min at 600 �g at room temperature. The light-density fraction on top of the NycoPrepsolution was collected and washed twice in PBS with 0.3% BSA and0.01% NaN3 (PBS/BSA) and then held overnight in PBS with 10% goatserum and 0.01% NaN3 at 4 C.

Flow cytometric analysis of cell surface markers

Approximately 2 � 105 cells suspended in up to 200 �l blocking bufferwere incubated in 96-well Costar V-bottom plates (Corning Inc., Corn-ing, NY) with saturating amounts of mAbs on ice for at least 30 min. Cellswere washed twice with 250 �l PBS/BSA. The washed cells were sus-pended in PBS/BSA containing 2 �g/ml propidium iodide (PI), pur-chased from Molecular Probes (Eugene, OR). PI was used to stain dead

cells so that they could be excluded from the FACS analysis. Flowcytometric analyses were performed using either a FACScan or FAC-SCalibur flow cytometer (BD Biosciences). Analyses of results wereperformed using CellQuest software (BD Biosciences).

ResultsImmunohistochemical analysis of CD56 expression in thehuman fetal adrenal gland

CD56 was expressed on cells from human fetal adrenalglands ranging in gestational age from 10 to 24 wk. In allcases, the pattern of CD56 expression was as a band of cellsimmediately below the capsule of the adrenal (Fig. 1, A andB) corresponding to the DZ (Fig. 2A). Human P450c17, whichis highly expressed in the FZ (10, 29), was analyzed as apositive marker of FZ cells using an anti-P450c17 antibodykindly provided by Dr. Walter Miller (Fig. 1C) (26). DZ cellsreacted minimally with anti-P450c17 antibody, althoughsome isolated cells in the DZ demonstrated staining. Addi-tionally, isolated pockets of bright CD56 staining were ob-served within the FZ (Fig. 1A).

Analysis of gene expression on definitive and fetal zone cellsisolated by laser capture microdissection

To confirm the expression of CD56 by DZ cells, CD56mRNA expression was assessed in DZ and FZ cells. Lasermicrodissection was used to isolate cells from the FZ andDZ (Fig. 2A), and RT-PCR was then used to assess CD56mRNA expression. In two experiments, CD56 mRNA wasfound expressed in DZ cells but not in FZ cells (Fig. 2B).A similar pattern of expression was observed for mRNAencoding the genes Mps-1, the growth regulatory proteinNovH, and CD90 (Thy-1). Mps-1 and NovH are expressedprimarily by DZ cells (30). CD90 expression is found

FIG. 1. CD56 and P450c17 protein expression in the human fetal adrenal. Immunofluorescence of a 17-wk gestation human fetal adrenalshowing CD56 staining of a complete section of fetal adrenal (A). Note the band of staining corresponding to the DZ and islands of expressionwithin the FZ (B). The FZ is demarked using P450c17 immunofluorescence in a 19-wk gestation fetal adrenal (C).

FIG. 2. Gene expression in the DZ and FZ. A schematicillustration of the morphology of the human fetal adrenalis shown in A. Cells from the DZ and FZ of a 22-wk gestationfetal adrenal were captured using laser microscopy, andRT-PCR was used to determine the presence of mRNA forCD56 and a panel of genes expressed in fetal adrenaltissue (B).

Muench et al. • CD56 Expression by Definitive Zone Cells J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 3923

on a variety of cell types including neurons, chromaffincells, hematopoietic stem cells, and connective tissue (16,31–34). Interestingly, P450c17 mRNA was detected amongDZ cells.

Analysis of CD56 and P450c17 expression byconfocal microscopy

The expression of P450c17 by some DZ cells was con-firmed by dual staining of fetal adrenal sections for CD56and P450c17, followed by confocal microscopy. Consistentwith the results in Fig. 1, low-power examination of thefetal adrenal reveals that most staining for CD56 andP450c17 is mutually exclusive and serves to demarcate theDZ and FZ, respectively (Fig. 3A). However, examinationat higher magnification revealed that some cells coexpressCD56 and P450c17 (Fig. 3, B–E). These cells typically wereobserved bordering the FZ and not the capsule. The in-tensity of the cell-surface CD56 staining and the cytoplas-mic P450c17 staining also appeared to be reduced in someof these cells, suggesting that they have an intermediatephenotype between the CD56�P450c17� DZ cells and theCD56�P450c17� FZ cells.

Flow cytometric analysis of CD56 expression on fetaladrenal cells

CD56 expression on the surface of adrenal cells wasanalyzed in fetal specimens ranging in age from 15 to 24wk of gestation. Background fluorescence indicated thepresence of diverse cell populations in the adrenal cellpreparations (Fig. 4A). This also was apparent in the anal-ysis of forward and side light scatter, which indicated the

presence of a large number of small- to medium-sized cellsof moderate complexity (Fig. 4B). Larger and more com-plex cells, with higher forward and side light scatters, alsowere observed. These cells were also responsible for thepopulation of cells with the high background fluorescence(�101 fluorescence) seen in Fig. 4A. Because CD56 is ex-pressed on NK cells found in peripheral blood (35), theadrenal samples also were stained for CD45 and CD235a(Fig. 4C), markers found exclusively on leukocytes anderythrocytes, respectively (36, 37). Together, CD45 andCD235a stained 66% of the isolated adrenal cells and ac-counted for the majority of small- to medium-sized cellswith low side light scatter (Fig. 4D).

CD56 expression was detected on 14% of the nonhema-topoietic cells as seen in Fig. 4E. These CD56� cells variedin size and complexity, as indicated in Fig. 4, F and G.Accordingly, the gated CD56� population in Fig. 2E ap-pears to represent a mixture of small cells with low back-ground fluorescence (Fig. 4G) and larger cells with highbackground fluorescence (Fig. 4F). The smaller CD56�

cells represent 0.9% of the total cell population, whereasthe larger cells comprised the remaining 13% of nonhe-matopoietic CD56� cells. The larger gated population doesnot appear to be of hematopoietic origin because its levelsof CD45/CD235a expression are below those of hemato-poietic cells, as seen in Fig. 4C, and the presence of theselarge CD56� cells corresponds to a decrease in theCD56�CD45�CD235a� cells, indicated by the elliptical re-gion seen in Fig. 4E. Because DZ cells are smaller than FZcells, we focused on isolating DZ cells on smallCD56�CD45�CD235a� cells.

FIG. 3. Expression of both CD56 and P450c17 pro-teins by DZ cells bordering the FZ. Confocal mi-croscopy was used to view sections of fetal adrenalsstained for CD56 (green) and P450c17 (red) protein(all panels). A low-power view spanning the DZ andFZ is shown (A). CD31 expression was also stainedin this one sample indicating the presence of en-dothelial cells lacking CD56 or P450c17 protein ex-pression (white, arrowhead). Higher-magnificationpictures reveal costaining for CD56 on the mem-brane (green) and P450c17 protein in the cytoplasm(red) of cells in three separate experiments (B–E,arrowheads). CD31 staining was not performedin B–E.

3924 J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 Muench et al. • CD56 Expression by Definitive Zone Cells

Flow cytometric and immunohistochemical analysis ofendothelial cell markers in the fetal adrenal gland

We speculated that our adrenal cell preparation also waslikely to contain endothelial cells in addition to the other celltypes present. Three markers associated with endothelialcells were analyzed to determine their pattern of expressionon hematopoietic and CD56� cells in the fetal adrenal. The

following markers were analyzed: CD31 (platelet endothelialcell adhesion molecule-1), CD34, and CD36 (throm-bospondin receptor). Flow cytometric analyses identifiedCD31�, CD34�, and CD36� populations in the adrenal cellpreparations that were not of hematopoietic origin and, thus,likely represented endothelial cells (Fig. 5, A, D, and G). Notethat each of these markers is expressed by some hematopoi-

FIG. 4. Flow cytometric analysis of CD56 expression byhuman fetal adrenal cells. Representative results fromthe analysis of a 19-wk gestation fetus are shown. Back-ground fluorescence obtained with staining using thenonspecific mAb indicated is shown (A). The forward andside light scatter profile of the entire adrenal cell prep-aration is shown (B). Hematopoietic cells are stainedwith CD45 and CD235a (C), and the forward and sidelight scatter profile of these gated cells is shown (D).CD56 expression by nonhematopoietic cells is shown (E),and the forward and side light scatter profiles of thegated regions are shown (F–H).

Muench et al. • CD56 Expression by Definitive Zone Cells J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 3925

etic cells and that such expression was evident in our anal-yses. The indicated CD31�, CD34�, and CD36� populations(arrows in Fig. 5, A, D, and G) represented 6%, 8%, and 19%of all cells, respectively. Analyses of CD31, CD34, and CD36expression on CD56� cells indicated that most CD56� cellswere negative for these markers although someCD34�CD56� and CD36�CD56� cells may exist (Fig. 5, B, E,and H). There was no overlap, however, between the CD31�

and the CD56� cell populations. The patterns of expressionby immunohistochemistry of each of these markers weresimilar to those seen in Fig. 5, C, F, and I. These markersstained cells and vessels in the capsule as well as cells dis-persed in the DZ. However, the pattern of expression wasconsistent with staining of the vasculature, in particular thecircular patterns associated with vessel cross-sections. Therewas no pattern of expression suggesting that DZ cells boundCD31, CD34, or CD36. This was further confirmed by athree-color analysis of adrenals stained for CD56, P450c17,

and CD31 and analyzed by confocal microscopy (Fig. 3A).CD31 expression (white color) did not overlap with cells thatexpressed CD56 or P450c17. These data, therefore, indicatedthat CD31 can be used to deplete adrenal cell preparations ofendothelial cells without depletion of the CD56� DZ cells.

Isolation of an enriched population of DZ cells byflow cytometry

Discontinuous-gradient centrifugation was tested to de-termine whether CD56� adrenal cells could be enriched bythis technique. Dissociated adrenal cells were centrifugedover a 1.077 g/ml layer of NycoPrep and the light-densitycells recovered. From 4.2 � 106 to 7.8 � 106 cells were re-covered per gland by the dissociation procedure in threeexperiments using 23- and 24-wk gestation tissues. The light-density fraction recovered represented 0.6% to 3.6% of start-ing cell population. The discarded high-density fraction con-

FIG. 5. Expression of endothelial cell markers by human fetal adrenal cells. Flow cytometric analysis of a 19-wk gestation adrenal indicatedthe presence of CD31�, CD34�, and CD36� cells that were not of hematopoietic origin (arrows, A, D, and G). Staining with CD56 and CD31,CD34, or CD36 indicated the existence of distinct populations of CD56� cells and the three endothelial cell populations (B, E, and H). However,possible CD34�CD56� and CD36�CD56� cells were also observed (arrows, E and H). Immunohistochemical staining with the three endothelialcell markers, CD31, CD34, and CD36, reveals similar patterns of expression on a 21-wk gestation adrenal (C, F, and I). The adrenal capsuleis at the top of these three panels, with an intensely stained vessel in cross-section (C and F but less so for I). A section of DZ with intermittentstaining is shown below the capsule in the lower portion of the pictures.

3926 J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 Muench et al. • CD56 Expression by Definitive Zone Cells

tained the majority of erythrocytes as well as other cell typesand debris. The light-density fraction contained 48% CD56�

cells, which was an approximate 5-fold enrichment over thetotal cell population (data not shown). Because density sep-aration was an effective means of enriching CD56� adrenalcells, this technique was used to enrich for DZ cells beforestaining with mAbs and FACS.

To isolate an enriched population of DZ cells, light-densityadrenal cells were stained with CD56-PE, CD31-FITC, CD45-FITC, and CD235a-FITC. Live cells, based on the lack of PIstaining, were isolated as shown in Fig. 6. These live cellswere further depleted of CD31�CD45�CD235a� cells, re-sulting in a population collectively called lineage� (lin�)cells. Two cell populations were isolated based on the pres-ence of absence of CD56 expression. The recovery ofCD56�lin� cells from 23- and 24-wk gestation tissues rangedfrom 1.3 � 104 to 4.1 � 104 cells/adrenal gland (n � 3). Therecovery of CD56�lin� cells was similar. There was no ap-parent contamination of the CD56�lin� cell population by

CD56� cells based on CD56 mRNA expression (Fig. 7). How-ever, a low level of CD31 gene expression was detected inboth sorted populations likely caused by either modest con-tamination of the sorted cell populations or CD31 gene ex-pression in the sorted cell populations without cell-surfaceexpression of the CD31 protein.

The profile of gene expression by the sorted cell popula-tions was consistent with the contention that the CD56�lin�

population is enriched for DZ cells. The CD56�lin� cell pop-ulation contained mRNA for the DZ markers NovH, Mps-1,and 3-�HSD (Fig. 7). The levels of 3-�HSD mRNA were alsoquantified using real-time RT-PCR (data not shown). BeforeFACS, 3-�HSD expression was at 2.9% of the levels ofGAPDH expression. Relative 3-�HSD expression was 3.4-fold higher (9.9% of GAPDH expression) in the isolatedCD56�lin� cells. Relative 3-�HSD expression in theCD56�lin� fraction was reduced to 0.2% of GAPDH expres-sion. In addition, a lower level of NovH and Mps-1 was alsoobserved in the CD56�lin� fraction, which is consistent with

FIG. 6. Isolation of subpopulations of hu-man fetal adrenal cells by FACS. Repre-sentative examples of the regions used toisolate CD56�lin� and CD56�lin� cellsfrom a 24-wk gestation adrenal are shown.First, live cells were selected by their lackof staining with PI as shown (A).CD56�lin� cells were isolated using a re-gion 1 (R1), which selected cells expressinghigh levels of CD56 and a complete lack oflin antigens (CD31, CD45, and CD235a). Asecond subpopulation of adrenal cells wasalso isolated based on the lack of staining ofall four markers (region 2, R2) (B). Thelight-scatter profiles of cells in regions 1and 2 are shown (C and D, respectively).

FIG. 7. Analysis of mRNA expressionby CD56�lin� and CD56�lin� humanfetal adrenal cells. RT-PCR products ofnycoll-treated presort and two sortedfractions, CD56�lin� and CD56�lin�,are shown.

Muench et al. • CD56 Expression by Definitive Zone Cells J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 3927

this cell population likely being enriched for FZ cells (Fig. 7).Chr-A mRNA, associated with fetal chromaffin cells (38, 39),also was detected in the CD56�lin� fraction. P450c17 andLDLR gene expression was detected in both sorted fractions.The expression of P450c17 was studied further by analyzingprotein expression by immunohistochemistry. CD56�lin�

cells were tested for P450c17 protein expression in threeexperiments. In situ staining of CD56�lin� cells indicated11–13% of these cells bound anti-P450c17 antibody (n � 3;data not shown).

Discussion

We have hypothesized that DZ cells represent a progenitorpopulation of the human fetal adrenal gland. Our eventualgoal is to provide direct evidence for the centripetal migra-tion of DZ adrenal cells to populate the remainder of thecortex, i.e. to demonstrate that cells from the outer DZ dif-ferentiate into cells of the underlying zones as they migrateinward. Studies support this hypothesis indirectly. Chimericgene expression analysis in the rat and mouse revealed clonalbands of cells extending from the innermost cell layers to thecapsule (40, 41). Furthermore, in the adult rat, a zone of cellswas identified that lacked steroid enzyme expression andpresumably is a progenitor cell population (42). The stron-gest evidence for the DZ cell population functioning as pro-genitors would be the demonstration that these cells cangrow and differentiate, either in vivo or in vitro, into matureFZ cells. This requires that the DZ cells can be isolated, witha high degree of purity, for such studies to be undertaken.

To date, DZ cells in the human fetal adrenal gland havebeen identified only by their location within the gland or bytheir lack of expression of steroidogenic enzymes. Our on-going aim has been to identify unique markers for these cellsto use as tools for the purification and characterization of theDZ cells. Recently, we used subtractive hybridization ofRNA from cells prepared by laser capture microdissection todefine novel markers unique to DZ cells in the fetal adrenalgland (12). The present study extends this effort by describ-ing CD56 as a cell surface marker for DZ cells. We identifiedthis marker through an extensive series of phenotyping ex-periments using FACS. We have shown that CD56 is ex-pressed on a cell population of small, dense, low side scattercells that lack markers of hematopoietic origin (CD31, CD34and CD235a). This cell population is enriched for cells ex-pressing known markers of DZ cells, NovH and Mps-1. Thus,we believe we have isolated a highly enriched DZ cellpopulation.

The CD56�lin� cell pool also contains mRNA for P450c17,which is a steroid enzyme abundantly expressed in TZ andFZ cells (10, 29). However, RT-PCR, as was employed in ourstudy, is a very sensitive but minimally quantitative tech-nique and can detect very low levels of mRNA. Therefore, wequantified the percentage of P450c17� cells in the purifiedcell pool by immunofluorescence. P450c17 protein was de-tected in 11–13% of the sorted cell population, which ac-counts for the strong signal in the RT-PCR experiments.Moreover, using confocal microscopy some cells expressingP450c17 were visible among the CD56� DZ cells. The patternof expression observed suggests that these cells may repre-

sent a differentiating cell population that has not yet mi-grated centripetally. The presence of these cells lends cre-dence to the concept of the DZ cells serving as a progenitorpopulation. If one cell truly is the progenitor of another, anintermediate cell likely exists that expresses some markersfrom both the progenitor and mature cell types. Furtherinvestigation of human fetal adrenocortical cells likely willlead to the description of other markers that can be used tofurther purify and subdivide the DZ population. TheseCD56�P450c17� cells can then be better characterized re-garding their status as maturing FZ cells.

Some CD56�lin� cells also express Chr-A, a marker forchromaffin cells. Previous studies have indicated CD56 ex-pression on chromaffin cells in the developing rat adrenal(43, 44). Chromaffin cells are a major component of the ad-renal medulla and produce catecholamines. Traditionally,the adrenal medulla and cortex are thought of as two func-tionally and structurally distinct entities coexisting withinthe adrenal capsule. However, data have accumulated indi-cating that glucocorticoids produced by adrenal cortical cellscan influence medullary function (45). Cortisol is known tostimulate phenylethanolamine-N-methyltransferase, the en-zyme that catalyzes the conversion of norepinephrine toepinephrine in the adrenal medulla. Teleologically, it maynot be coincidence that the adrenal cortex, the site of cortisolsynthesis, envelops the adrenal medulla, as suggested byWurtman and Axelrod (46). Neuronal processes have beennoted that extend from medullary neurons to cells within allthree zones of the cortex (47). During human intrauterine life,we found that a well-formed medulla is not present beforebirth (48). Rather, there are scattered chromaffin cellsthroughout the fetal adrenal gland that express phenyleth-anolamine-N-methyltransferase, the enzyme that convertsnorepinephrine to epinephrine. Furthermore, close apposi-tion of adrenocortical and medullary cells was noted byothers. Nests of chromaffin cells have been found throughoutthe human adrenal cortex, particularly in subcapsular loca-tions, in both the adult (49) and fetus (50). In fact, chromaffincells are scattered in clusters, or nests, in the fetal adrenalgland because a more structurally distinct medulla does notform until after birth (51). These findings are consistent withthe pattern of CD56 expression we observed in this study,which indicated islands of CD56� cells found within the FZ.Thus, it is not surprising to find some evidence of similargene expression in adrenocortical and medullary cells. Ex-periments are underway to identify cell surface moleculesthat can be used to separate chromaffin cells from the DZcells. Alternatively, chromaffin cells are not adherent to tis-sue culture plates and, thus, may be eliminated from culturesby washing (52).

CD56 (neural cell adhesion molecule) is an adhesion mol-ecule that has been extensively studied in diverse tissues. Itsrole in axonal growth, migration, and guidance has beenexamined in detail (15, 53). In vitro antibody perturbation andmutational studies have established the role of CD56 as aregulator of neuronal growth, and it is essential for properneuronal patterning. CD56 expression was originallythought to be confined to neuronal and neuroendocrine tis-sues. However, when multiple alternatively spliced isoformsof CD56 were recognized and new antibodies generated,

3928 J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 Muench et al. • CD56 Expression by Definitive Zone Cells

expression was noted in other tissues, most notably endo-crine organs (54, 55). Rat pituitary cells (54), human thyro-cytes (56), and human adrenocortical cells (57) all expressCD56. Interestingly, the adult adrenal gland expresses CD56primarily in the zona glomerulosa (57). Late in gestation, theDZ, which also is immediately beneath the capsule, is anal-ogous to the zona glomerulosa in that it begins to expresssteroidogenic enzymes and produce aldosterone (10).

CD56 is expressed by cells in as many as 20–30 differentforms. Alternate splicing from a single gene accounts formost of this variability, but posttranslational modificationplays a lesser role (58). Several groups have noted the 140-kDa isoform in endocrine tissues, but the 180-kDa isoform ismore common in neuronal tissue (59–61). The 140-kDa iso-form, when expressed in glioma cells, decreased cell motilityto a greater extent than the 180-kDa isoform (62). Furtherinvestigation into CD56 isoform expression in the fetal ad-renal and its potential role in cell migration is warranted. Itis possible that cells in the DZ express the 140-kDa isoformof CD56 until differentiation has progressed to the point atwhich migration into other zones is necessary. Then, as CD56expression is altered or decreases, the cells begin to migrateinward and differentiate into cells of the TZ or FZ.

In conclusion, the method developed in this study for theisolation of DZ cells is an important step toward the detailedmolecular analysis of the events involved in the cellulardevelopment of the human fetal adrenal. The developmentof a reproducible, rapid isolation method is critical to obtainhighly pure cells, which can now be studied in vitro. We planto use this method to explore the role of CD56� DZ cells inthe development of the human fetal adrenal gland.

Acknowledgments

We are indebted to Dr. Walter Miller for providing antibody recog-nizing P450c17 and Dr. Yuet Wai Kan for his generous support. We alsothank Paul Dazin (Howard Hughes Medical Institute, UCSF) and JaneGordon (Laboratory for Cell Analysis, UCSF) for assistance with cellsorting and confocal microscopy.

Received January 30, 2003. Accepted April 25, 2003.Address all correspondence and requests for reprints to: Marcus O.

Muench, Ph.D., Department of Laboratory Medicine, University of Cal-ifornia, San Francisco, 533 Parnassus Avenue, Room U-440, San Fran-cisco, California 94143-0793. E-mail: muench@itsa.ucsf.edu.

This work was supported in part by NIH Grants DK59301 (to M.O.M.)and HD08478 (to R.B.J.). J.R. is the recipient of a fellowship from theEndocrine Fellows Foundation.

M.O.M. and J.V.R. contributed equally to this work.

References

1. Moore MAS, Muench MO, Warren DJ, Laver J 1990 Cytokine networksinvolved in the regulation of haemopoietic stem cell proliferation and differ-entiation. In: Molecular control of haemopoiesis. Chichester, UK: John Wileyand Sons; 43–58

2. Risbud MV, Bhonde RR 2002 Models of pancreatic regeneration in diabetes.Diabetes Res Clin Pract 58:155–165

3. Marshman E, Booth C, Potten CS 2002 The intestinal epithelial stem cell.Bioessays 24:91–98

4. Oh SH, Hatch HM, Petersen BE 2002 Hepatic oval ‘stem’ cell in liver regen-eration. Semin Cell Dev Biol 13:405–409

5. Chmielnicki E, Goldman SA 2002 Induced neurogenesis by endogenousprogenitor cells in the adult mammalian brain. Prog Brain Res 138:451–464

6. Laywell ED, Steindler DA 2002 Glial stem-like cells: implications for ontog-eny, phylogeny, and CNS regeneration. Prog Brain Res 138:435–450

7. Verfaillie CM 2002 Adult stem cells: assessing the case for pluripotency.Trends Cell Biol 12:502–508

8. Shafritz DA, Dabeva MD 2002 Liver stem cells and model systems for liverrepopulation. J Hepatol 36:552–564

9. Bonner-Weir S, Sharma A 2002 Pancreatic stem cells. J Pathol 197:519–52610. Mesiano S, Coulter CL, Jaffe RB 1993 Localization of cytochrome P450 cho-

lesterol side-chain cleavage, cytochrome P450 17 alpha-hydroxylase/17, 20-lyase, and 3 beta-hydroxysteroid dehydrogenase isomerase steroidogenic en-zymes in human and rhesus monkey fetal adrenal glands: reappraisal offunctional zonation. J Clin Endocrinol Metab 77:1184–1189

11. Coulter CL, Goldsmith PC, Mesiano S, Voytek CC, Martin MC, Mason JI,Jaffe RB 1996 Functional maturation of the primate fetal adrenal in vivo. II.Ontogeny of corticosteroid synthesis is dependent upon specific zonal expres-sion of 3 beta-hydroxysteroid dehydrogenase/isomerase. Endocrinology 137:4953–4959

12. Ratcliffe JV, Nakanishi M, Jaffe RB 2003 Identification of definitive and fetalzone markers in the human fetal adrenal gland reveals putative developmentalgenes. J Clin Endocrinol Metab, 88:3272–3277

13. Thiery JP, Duband JL, Rutisauser U, Edelman GM 1982 Cell adhesion mol-ecules in early chicken embryogenesis. Proc Natl Acad Sci USA 79:6737–6741

14. Rutishauser U 1984 Developmental biology of a neural cell adhesion molecule.Nature 310:549–554

15. Ronn LC, Hartz BP, Bock E 1998 The neural cell adhesion molecule (NCAM)in development and plasticity of the nervous system. Exp Gerontol 33:853–864

16. Poltorak M, Shimoda K, Freed WJ 1990 Cell adhesion molecules (CAMs) inadrenal medulla in situ and in vitro: enhancement of chromaffin cell L1/Ng-CAM expression by NGF. Exp Neurol 110:52–72

17. Hercend T, Griffin JD, Bensussan A, Schmidt RE, Edson MA, Brennan A,Murray C, Daley JF, Schlossman SF, Ritz J 1985 Generation of monoclonalantibodies to a human natural killer clone. Characterization of two naturalkiller-associated antigens, NKH1A and NKH2, expressed on subsets of largegranular lymphocytes. J Clin Invest 75:932–943

18. Lanier LL, Testi R, Bindl J, Phillips JH 1989 Identity of Leu-19 (CD56) leu-kocyte differentiation antigen and neural cell adhesion molecule. J Exp Med169:2233–2238

19. Fabris L, Strazzabosco M, Crosby HA, Ballardini G, Hubscher SG, Kelly DA,Neuberger JM, Strain AJ, Joplin R 2000 Characterization and isolation ofductular cells coexpressing neural cell adhesion molecule and Bcl-2 fromprimary cholangiopathies and ductal plate malformations. Am J Pathol 156:1599–1612

20. Libbrecht L, Cassiman D, Desmet V, Roskams T 2001 Expression of neuralcell adhesion molecule in human liver development and in congenital andacquired liver diseases. Histochem Cell Biol 116:233–239

21. Gaidar YA, Lepekhin EA, Sheichetova GA, Witt M 1998 Distribution ofN-cadherin and NCAM in neurons and endocrine cells of the human embry-onic and fetal gastroenteropancreatic system. Acta Histochem 100:83–97

22. Esni F, Taljedal IB, Perl AK, Cremer H, Christofori G, Semb H 1999 Neuralcell adhesion molecule (N-CAM) is required for cell type segregation andnormal ultrastructure in pancreatic islets. J Cell Biol 144:325–337

23. Walsh FS, Hobbs C, Wells DJ, Slater CR, Fazeli S 2000 Ectopic expression ofNCAM in skeletal muscle of transgenic mice results in terminal sprouting atthe neuromuscular junction and altered structure but not function. Mol CellNeurosci 15:244–261

24. Perl AK, Dahl U, Wilgenbus P, Cremer H, Semb H, Christofori G 1999Reduced expression of neural cell adhesion molecule induces metastatic dis-semination of pancreatic beta tumor cells. Nat Med 5:286–291

25. Libbrecht L, De Vos R, Cassiman D, Desmet V, Aerts R, Roskams T 2001Hepatic progenitor cells in hepatocellular adenomas. Am J Surg Pathol 25:1388–1396

26. Lin D, Black SM, Nagahama Y, Miller WL 1993 Steroid 17 alpha-hydroxylaseand 17, 20-lyase activities of P450c17: contributions of serine106 and P450reductase. Endocrinology 132:2498–2506

27. Basora N, Vachon PH, Herring-Gillam FE, Perreault N, Beaulieu JF 1997Relation between integrin alpha7Bbeta1 expression in human intestinal cellsand enterocytic differentiation. Gastroenterology 113:1510–1521

28. Geva E, Ginzinger DG, Zaloudek CJ, Moore DH, Byrne A, Jaffe RB 2002Human placental vascular development: vasculogenic and angiogenic(branching and nonbranching) transformation is regulated by vascular endo-thelial growth factor-A, angiopoietin-1, and angiopoietin-2. J Clin EndocrinolMetab 87:4213–4224

29. Hanley NA, Rainey WE, Wilson DI, Ball SG, Parker KL 2001 Expressionprofiles of SF-1, DAX1, and CYP17 in the human fetal adrenal gland: potentialinteractions in gene regulation. Mol Endocrinol 15:57–68

30. Martinerie C, Gicquel C, Louvel A, Laurent M, Schofield PN, Le Bouc Y 2001Altered expression of novH is associated with human adrenocortical tumor-igenesis. J Clin Endocrinol Metab 86:3929–3940

31. Morris R 1985 Thy-1 in developing nervous tissue. Dev Neurosci 7:133–16032. Craig W, Kay R, Cutler RL, Lansdorp PM 1993 Expression of Thy-1 on human

hematopoietic progenitor cells. J Exp Med 177:1331–134233. Kennedy PG, Lisak RP, Raff MC 1980 Cell type-specific markers for human

glial and neuronal cells in culture. Lab Invest 43:342–35134. Muench MO, Cupp J, Polakoff J, Roncarolo MG 1994 Expression of CD33,

CD38 and HLA-DR on CD34� human fetal liver progenitors with a highproliferative potential. Blood 83:3170–3181

Muench et al. • CD56 Expression by Definitive Zone Cells J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 3929

35. Lanier LL, Chang C, Azuma M, Ruitenberg JJ, Hemperly JJ, Phillips JH 1991Molecular and functional analysis of human natural killer cell-associated neu-ral cell adhesion molecule (N-CAM/CD56). J Immunol 146:4421–4426

36. Woodford-Thomas T, Thomas ML 1993 The leukocyte common antigen, CD45and other protein tyrosine phosphatases in hematopoietic cells. Semin Cell Biol4:409–418

37. Robinson J, Sieff C, Delia D, Edwards PAW, Greaves M 1981 Expression ofcell-surface HLA-DR, HLA-ABC and glycophorin during erythroid differen-tiation. Nature 289:68–71

38. Molenaar WM, Lee VM, Trojanowski JQ 1990 Early fetal acquisition of thechromaffin and neuronal immunophenotype by human adrenal medullarycells. An immunohistological study using monoclonal antibodies to chromo-granin A, synaptophysin, tyrosine hydroxylase, and neuronal cytoskeletalproteins. Exp Neurol 108:1–9

39. Bocian-Sobkowska J, Wozniak W, Malendowicz LK, Ginda W 1996 Stere-ology of human fetal adrenal medulla. Histol Histopathol 11:389–393

40. Iannaccone PM, Weinberg WC 1987 The histogenesis of the rat adrenal cortex:a study based on histologic analysis of mosaic pattern in chimeras. J Exp Zool243:217–223

41. Morley SD, Viard I, Chung BC, Ikeda Y, Parker KL, Mullins JJ 1996 Var-iegated expression of a mouse steroid 21-hydroxylase/beta-galactosidasetransgene suggests centripetal migration of adrenocortical cells. Mol Endo-crinol 10:585–598

42. Mitani F, Suzuki H, Hata J, Ogishima T, Shimada H, Ishimura Y 1994 A novelcell layer without corticosteroid-synthesizing enzymes in rat adrenal cortex:histochemical detection and possible physiological role. Endocrinology 135:431–438

43. Leon C, Grant NJ, Aunis D, Langley K 1992 Expression of cell adhesionmolecules and catecholamine synthesizing enzymes in the developing ratadrenal gland. Brain Res Dev Brain Res 70:109–121

44. Grant NJ, Leon C, Aunis D, Langley K 1992 Cellular localization of the neuralcell adhesion molecule L1 in adult rat neuroendocrine and endocrine tissues:comparisons with NCAM. J Comp Neurol 325:548–558

45. Carballeira A, Fishman LM 1980 The adrenal functional unit: a hypothesis.Perspect Biol Med 23:573–597

46. Wurtman RJ, Axelrod J 1966 Control of enzymatic synthesis of adrenaline inthe adrenal medulla by adrenal cortical steroids. J Biol Chem 241:2301–2305

47. McNicol AM, Richmond J, Charlton BG 1994 A study of general innervationof the human adrenal cortex using PGP 9.5 immunohistochemistry. Acta Anat(Basel) 151:120–123

48. Wilburn LA, Jaffe RB 1988 Quantitative assessment of the ontogeny of met-enkephalin, norepinephrine and epinephrine in the human fetal adrenal me-dulla. Acta Endocrinol (Copenh) 118:453–459

49. Bornstein S, Gonzalez-Hernandez J, Ehrhart-Bornstein M, Adler G,Scherbaum W 1994 Intimate contact of chromaffin and cortical cells within thehuman adrenal gland forms the cellular basis for important intraadrenal in-teractions. J Clin Endocrinol Metab 78:225–232

50. Yon L, Breault L, Contesse V, Bellancourt G, Delarue C, Fournier A, LehouxJ-G, Vaudry H, Gallo-Payet N 1998 Localization, characterization, and secondmessenger coupling of pituitary adenylate cyclase-activating polypeptide re-ceptors in the fetal human adrenal gland during the second trimester ofgestation. J Clin Endocrinol Metab 83:1299–1305

51. Wilburn LA, Goldsmith PC, Chang KJ, Jaffe RB 1986 Ontogeny of enkephalinand catecholamine-synthesizing enzymes in the primate fetal adrenal medulla.J Clin Endocrinol Metab 63:974–980

52. Crickard K, Ill CR, Jaffe RB 1981 Control of proliferation of human fetaladrenal cells in vitro. J Clin Endocrinol Metab 53:790–796

53. Walsh FS, Doherty P 1997 Neural cell adhesion molecules of the immuno-globulin superfamily: role in axon growth and guidance. Annu Rev Cell DevBiol 13:425–456

54. Langley OK, Aletsee-Ufrecht MC, Grant NJ, Gratzl M 1989 Expression of theneural cell adhesion molecule NCAM in endocrine cells. J Histochem Cyto-chem 37:781–791

55. Jin L, Hemperly J, Lloyd R 1991 Expression of neural cell adhesion moleculein normal and neoplastic human neuroendocrine tissues. Am J Pathol 138:961–969

56. Zeromski J, Lawniczak M, Galbas K, Jenek R, Golusinski P 1998 Expressionof CD56/N-CAM antigen and some other adhesion molecules in varioushuman endocrine glands. Folia Histochem Cytobiol 36:119–125

57. Ehrhart-Bornstein M, Hilbers U 1998 Neuroendocrine properties of adreno-cortical cells. Horm Metab Res 30:436–439

58. Cunningham BA, Hemperly JJ, Murray BA, Prediger EA, Brackenbury R,Edelman GM 1987 Neural cell adhesion molecule: structure, immunoglobulin-like domains, cell surface modulation, and alternative RNA splicing. Science236:799–806

59. Langley OK, Aletsee-Ufrecht MC, Grant NJ, Gratzl M 1989 Expression of theneural cell adhesion molecule NCAM in endocrine cells. J Histochem Cyto-chem 37:781–791

60. Lahr G, Mayerhofer A, Bucher S, Barthels D, Wille W, Gratzl M 1993 Neuralcell adhesion molecules in rat endocrine tissues and tumor cells: distributionand molecular analysis. Endocrinology 132:1207–1217

61. Mayerhofer A, Lahr G, Gratzl M 1991 Expression of the neural cell adhesionmolecule in endocrine cells of the ovary. Endocrinology 129:792–800

62. Prag S, Lepekhin EA, Kolkova K, Hartmann-Petersen R, Kawa A, WalmodPS, Belman V, Gallagher HC, Berezin V, Bock E, Pedersen N 2002 NCAMregulates cell motility. J Cell Sci 115:283–292

3930 J Clin Endocrinol Metab, August 2003, 88(8):3921–3930 Muench et al. • CD56 Expression by Definitive Zone Cells