Hematopoietic stem cells (HSCs) are defined by their abil-ity to give rise to both new stem cells and all blood cells.During mouse ontogenesis, definitive hematopoiesis initiallyoccurs in the aorta-gonad-mesonephros region as early asgestational day 10 (E10).1) HSCs migrate to the fetal liver(FL) around E12, and exist there during the late gestationalperiod.2,3) After birth, HSCs migrate to the bone marrow(BM) and are present there for life. It is speculated that thelate gestational period is important for expanding the numberof hematopoietic cells. It has been reported that HSCs de-rived from FL have superior repopulation and developmentalcapacities than other hematopoietic cells such as bone mar-row cells,4,5) indicating that the FL microenvironment in-duces active hematopoiesis.

The microenvironment is essential for the generation ofhematopoietic cells and consists of stromal cells. Studies ofthe hematopoietic microenvironment using BM stromal cellshave been reported by several investigators. Dexter et al. es-tablished long term BM cultures as a model system.6) Usingthe system, several molecules that regulate hematopoieticcell differentiation were identified.7,8) For instance, Miyake etal. reported that cell–cell interactions through VLA-4 andVCAM-1 are important for lymphohematopoiesis in BM.9)

The role of FL stromal cells in the fetal hematopoiesishave also been investigated.10—12) Kinoshita et al. reportedthat the FL-derived stromal layer comprises mostly fetal hepatic cells, and these cells support the expansion ofhematopoietic cells. Hepatic differentiation and fetalhematopoiesis are inversely correlated during onto-genesis.13,14) Although many studies have examined thehematopoietic microenvironment, the molecular mechanismsof the FL microenvironment remain unclear.

In the present paper, we describe the establishment of anovel monoclonal antibody (mAb), Ndk-10, that detects asurface molecule on FL stromal cells. Ndk-10 inhibits thesurvival of c-kit� hematopoietic cells on FL stromal cells.The Ndk-10 reactive molecule is identified as mouseaminopeptidase N/CD13 (APN/CD13). These results suggestthat the APN/CD13 is an important element of the

hematopoietic microenvironment in FL.

MATERIALS AND METHODS

Mice All procedures used on experimental animals wereapproved by the experimental Animal Care and Use Commit-tee at Graduate School of Pharmaceutical Sciences, OsakaUniversity. Specific pathogen-free C57BL/6 mice aged 6 to 8weeks were purchased from Charles River Japan (Yokohama,Japan). The mice were mated at night, and females were ex-amined the next morning. The day on which a vaginal plugwas found was considered day 0 of embryonic development.Sprague-Dawley rats were purchased from CLEA Japan(Tokyo, Japan).

Antibodies and Reagents The rat mAb to mouse c-kit(ACK2) was a gift from Dr. S.-I. Nishikawa (Kyoto Univer-sity, Kyoto, Japan),15) and was labeled with FITC in our labo-ratory. FITC-conjugated goat anti-rat IgG, biotin-conjugatedgoat anti-rat IgG (H�L), and the rat mAb to mouse CD13(R3-242) were purchased from ICN pharmaceuticals (Au-rora, OH), Cedarlane (Ontario, Canada), and BD Pharmingen(Franklin Lakes, NJ, U.S.A.), respectively. Ubenimex(Bestatin, MW: 308.4, a substance produced by Streptomycesolivoreticuli), an inhibitor of APN/CD13 and aminopepti-dase B, was purchased from Wako Pure Chemicals (Osaka,Japan).

Preparation of Fetal Liver and Hematopoietic CellFractions FL stromal cells were prepared from the liver ofgestational day 14 (E14) embryos as follows. FL lobes weregently washed in phosphate-buffered saline (PBS) for15 min, and treated with 0.05% trypsin and 0.02% EDTA inPBS for 30 min at 37 °C. The lobes were gently pipetted, andfiltered through a nylon mesh to remove undigested frag-ments, and then cultured in Dulbecco’s modified Eagle’smedium (Sigma Chemical, St. Louis, MO, U.S.A.) contain-ing 10% horse serum (Invitrogen, Carlsbad, CA, U.S.A.) andkanamycin (100 mg/l, Meiji Seika, Tokyo, Japan). Twelvehours later, non-adherent cells were collected by gentleswirling of the culture dishes with a sufficient volume of cul-

2014 Biol. Pharm. Bull. 27(12) 2014—2020 (2004) Vol. 27, No. 12

∗ To whom correspondence should be addressed. e-mail: hiroshiy@phs.osaka-u.ac.jp © 2004 Pharmaceutical Society of Japan

Aminopeptidase N/CD13 Regulates the Fetal Liver Microenvironment ofHematopoiesis

Naoki SAKANE, Yoko ASANO, Tomoko KAWAMURA, Tomohiro TAKATANI, Yasuhiro KOHAMA, Kazutake TSUJIKAWA, and Hiroshi YAMAMOTO*

Department of Immunology, Graduate School of Pharmaceutical Sciences, Osaka University; Suita, Osaka 565–0871,Japan. Received September 4, 2004; accepted October 1, 2004; published online October 12, 2004

Fetal liver (FL) hematopoiesis is thought to be important for expanding the cell number during ontogeny. Inorder to investigate the cellular interaction molecules among FL stromal and hematopoietic cells, we establisheda monoclonal antibody, Ndk-10, that reacts with FL stromal cells but not with dish non-adherent cells. WhenNdk-10 was added to an FL stromal and hematopoietic cell-coculture, it inhibited the survival of c-kit� cells. Theinhibitory activity of Ndk-10 was also observed in the fetal liver organ culture. The Ndk-10 recognized a 150 kDmolecule in the adherent cells of FL and kidney, and the N-terminal amino acid sequence was identical to that ofmouse aminopeptidase N/CD13. The peptidase activity of CD13 was inhibited by Ndk-10, and addition of its spe-cific inhibitor resulted in the same inhibitory activity as Ndk-10. We propose that aminopeptidase N/CD13 is acritical molecule that regulates the survival of c-kit� cells in the FL microenvironment.

key words fetal liver; microenvironment; CD13; aminopeptidase; hematopoiesis; stromal cell

ture medium. The non-adherent fraction of FL cells included20—30% c-kit� cells and they died within 24 h of culturewithout any additional cells or growth factors. The adherentand non-adherent fractions of FL cells are called FL stromalcells and hematopoietic cells, respectively, hereafter.

Flow Cytometry About 5�105 cells were incubatedwith hybridoma culture supernatants or about 1 mg/ml puri-fied mAbs in PBS containing 1% fetal calf serum (FCS,Boehringer Mannheim, Acton, MA, U.S.A.) and 0.1%sodium azide for 30 min on ice. The cells were washed oncewith the buffer and incubated with second antibodies for 30min on ice, and rendered to flow cytometry. Flow cytometricprofiles were analyzed with a FACSCalibur analyzer andCELLQuest software (Becton Dickinson ImmunocytometrySystems, Mountain View, CA, U.S.A.).

Cell Coculture System 5�105 FL stromal cells wereprecultured to adhere on a 24-well plate (Nunc, Roskilde,Denmark) for 48 h. Then 1�106 FL hematopoietic cells wereadded to each well and cultured in RPMI 1640 medium(Sigma Chemical) containing 10% FCS and kanamycin (100mg/l). Four days later, non-adherent cells were recovered, andthe number of c-kit� cells was determined. During this co-culture period, various mAbs (10 mg/ml) or Ubenimex (3.2—320 mM) were added. In some experiments, frequencies ofapoptotic cell death were determined by staining the cellswith anti-annexin V and propidium iodide (PI) (using apop-tosis detection kit, product of BioVision, Mountain View,CA, U.S.A.) according to the methods of Vermes et al .16)

Fetal Liver Organ Culture System Fetal liver organculture (FLOC) was performed according to the methods ofOwen et al.17) and Ceredig et al.18) Gestational day 12 (E12)fetal livers were taken and sliced into 4 pieces with a razorblade. The pieces were then transferred onto Nucleporemembrane filters (Nuclepore-Corning, Pleasanton, CA,U.S.A.), and FLOC was performed with or without mAb(10 mg/ml) or Ubenimex (320 mM). Culture medium is thesame as for cell co-culture system. Six days later, culturedfetal livers were recovered, gently homogenized, filteredthrough a nylon mesh, and dish-non-adherent single cell sus-pensions were prepared. Frequencies of c-kit� cells were de-termined by FACS.

Generation of Monoclonal Antibodies MAbs were es-tablished according to the usual method. Briefly, intact FLstromal cells were repeatedly injected intraperitoneally intorats. Three days after the last injection, the rats were sacri-ficed under anesthesia, and splenocytes were fused withmyeloma cells, P3X63Ag8U.1 (P3U1). Strategies for mAbselection are described in the results section.

Immunohistochemical Staining Cryostat sections ofvarious tissues were fixed in acetone for 10 min and then in-cubated with 5% bovine serum albumin (Sigma Chemical) inPBS. The sections were incubated for 1 h at 37 °C with mon-oclonal antibodies (30 mg/ml), followed by further incubationfor 30 min with biotinylated goat anti-rat IgG (H�L), andthen with 0.3% H2O2 (with 0.1% NaN3) in PBS, to removeendogenous peroxidase activity. The avidin–biotin reactionwas achieved by using Vectastain ABC elite kit (Vector Lab-oratories, Burlingame, CA, U.S.A.). Visualization of peroxi-dase activity was performed by incubating the sections withan equivalent mixture of 1 mg/ml DAB (3,3�-diaminobenzi-dine, Dojindo, Kumamoto, Japan) in 0.1 M Tris–HCl and

0.02% H2O2. The sections were counterstained with hema-toxyline.

Immunoprecipitation and Immunoblotting FL stromalcells and various tissues were homogenized in ice-cold lysisbuffer (1% NP40, 150 mM NaCl, 50 mM Tris (pH 7.4), 5 mM

EDTA, and 1 mM benzamide), and the supernatants were col-lected after centrifugation. The cell lysates were precleanedwith protein G-Sepharose (Amersham Pharmacia Biotech,Arlington Heights, IL, U.S.A.) for over 1 h at 4 °C. For im-munoprecipitation, the lysates were incubated with 5 mgmAb (Ndk-10 or R3-242) for over 2 h at 4 °C. After washingtwice with 1 ml lysis buffer, immunoprecipitates were re-solved by 7.5% SDS-PAGE. Proteins in the immunoprecipi-tates were transferred from the SDS-polyacrylamide gel to anitrocellulose membrane (Schleicher & Schuell, Keene, NH,U.S.A.), blocked with 5% skim milk in TBS-T (20 mM

Tris–HCl (pH 8.0), 137 mM NaCl, and 0.1% Tween 20) andincubated with primary antibody (Ndk-10) at room tempera-ture for 1 h, and then washed three times with TBS-T. To de-tect antibody binding, horseradish peroxidase (HRP)-conju-gated anti-rat IgG (Santa Cruz Biotechnology, Santa Cruz,CA) diluted in TBS-T was incubated at room temperature for1 h. After washing three times in TBS-T, the bound HRPconjugates were visualized with enhanced chemiluminescentreagent (Wako Pure Chemicals).

Amino Acid Sequencing A kidney lysate was immuno-precipitated with Ndk-10, and resolved by SDS-PAGE as de-scribed above. Proteins in the immunoprecipitate were trans-ferred from the SDS-polyacrylamide gel to a PVDF mem-brane (BIO-RAD, Hercules, CA, U.S.A.). The membranewas stained with Coomassie R-250, the 150 kD band was ex-cised, and the N-terminal sequence was determined by NippiCo. Ltd. (Tokyo, Japan).

Peptidase Activity The peptidase activity of cell surfaceAPN/CD13 was measured according to the method of Ash-mun et al.19) Intact FL stromal cells were incubated at 37 °Cfor 1 h in PBS containing 6 mM alanine-p-nitroanilide (SigmaChemicals), a substrate for APN/CD13. After centrifugationat 4 °C, the absorbance at 405 nm of the cell free super-natants was measured to detect free p-nitroaniline liberatedby cleavage of the substrate by APN/CD13. All measure-ments were made in triplicate. FL stromal cells were preincu-bated with Ndk-10 or control IgG (20 mg/ml) for 12 h.

RESULTS

Production of mAb MAbs were screened by the follow-ing two-step selection. First, mAbs that stained FL stromalcells were selected. Among them, mAbs that stained FLhematopoietic cells were discarded. Because we intended toexamine molecules participating in the FL microenviron-ment, we chose mAbs that reacted only with the stromal cellfraction of FL. Figure 1 shows the flow cytometric patternsof 3 typical clones among the 14 selected clones. Theseclones reacted only with the adherent FL stromal cell frac-tion, but not with FL hematopoietic cells. The isotype ofNdk-10 was IgG2b.

Inhibition of c-kit� Cell Survival by mAb To study therole of the antigen recognized by these mAbs in the FL mi-croenvironment, we established an in vitro culture method todetect the interaction between FL stromal cells and

December 2004 2015

hematopoietic cells. Hematopoietic cells were cultured on0—5�105 FL stromal cells. Four days later, the number of c-kit� cells was determined. c-kit is widely known as a markermolecule of hematopoietic progenitor cells.20) As shown inFig. 2, c-kit� hematopoietic cells survived on FL stromalcells in proportion to the number of FL stromal cells. Theseresults suggest that FL stromal cells can support hematopoi-etic cell survival.

To examine the inhibitory activity of the mAbs onhematopoietic cell survival, 14 mAbs were added to the co-culture. As shown in Fig. 3A, several mAbs inhibited the sur-vival of c-kit� cells. Among them, Ndk-10 showed thestrongest inhibition of c-kit� cell survival on FL stromalcells. The inhibitory activity of Ndk-10 was also observed inthe FLOC system (Fig. 3B).

Immunohistochemical Analysis of Ndk-10 Antigen Toexamine the expression pattern of the Ndk-10 antigen, weperformed an immunohistochemical analysis of various fetaland adult tissues. Cryostat sections of E14 embryonic organswere stained with Ndk-10 (Figs. 4A—E). In fetal tissues,Ndk-10 antigen was expressed strongly in the liver andslightly in the thymus. The expression of the Ndk-10 antigenin various tissues from 6-week-old C57BL/6 mice is shownin Figs. 4F—K. The Ndk-10 antigen was detected in lym-phoid tissues such as the spleen (Fig. 4J) and thymus (Fig.4K). Because we checked that lymphocytes did not expressthe Ndk-10 antigen by flow cytometry (data not shown), thedistribution was thought to be due to expression on stromalcells. On the other hand, in non-hematopoietic tissues, strongexpression was observed in the kidney, especially in renaltubular epithelial cells. Other tissues displayed weak expres-sion. Because the Ndk-10 antigen was detected in various

2016 Vol. 27, No. 12

Fig. 1. Establishment of Monoclonal Antibodies Recognizing FL Adher-ent Cells

FL-derived adherent (stromal) and non-adherent (hematopoietic) cells were stainedwith mAbs together with FITC-conjugated goat anti-rat IgG. The viable cell fractionwas gated and single-color flow cytometric patterns are shown. Thin lines indicatebackground staining (second Ab alone) (X-axis: log fluorescence intensity, Y-axis: cellcounts/arbitrary units).

Fig. 2. Survival of Hematopoietic Cells is Dependent on Fetal Liver Stro-mal Cells

FL-derived stromal cells (0—50�104 cells/well: X-axis) and FL-derived non-adher-ent (hematopoietic) cells (� 100�104 cells/well, � 20�104 cells/well, � 104 cells/well)were cocultured. c-kit� cells in the FL-derived non-adherent fraction were 18% in thisexperiment. Four days later, non-adherent cells were collected and counted. Recoveredcells were stained with FITC-conjugated anti-c-kit mAb (ACK2) and the percentage ofc-kit� cells was measured by flow cytometry. The results indicate the number of c-kit�

cells (�104 cells) (Y-axis).

Fig. 3. Survival of c-kit� Cells Is Inhibited by mAb

(A) Hematopoietic cells derived from FL were cultured on FL stromal cells in thepresence of mAbs (10 mg/ml). Four days later, floating cells were collected andcounted. The cells were stained with FITC-conjugated anti-c-kit mAb (ACK2) and thepercentage of c-kit� cells was measured by flow cytometry. The results are given aspercent recovery compared with medium only. The number of c-kit� cells when cul-tured in medium only was 5.4—19.6�104 in each experiment. (B) Effect of Ndk-10was examined by FLOC system. E12 FLs were cultured for 6 d on Nuclepore mem-brane with or without Ndk-10 (10 mg/ml) or Ubenimex (320 mM). Number of c-kit�

cells were determined by FACS. Statistical analysis was performed by Student’s t-test(∗1: p�0.002, ∗2: p�0.001 compared with control IgG).

hematopoietic tissues, such as FL, spleen and thymus, theantigen is thought to be an important molecule for maintain-ing the hematopoietic microenvironment. Since it is also detected in the kidney, the Ndk-10 antigen may have func-tion(s) other than the maintenance of the hematopoietic mi-croenvironment.

In liver, a supportive function of hematopoietic cells existsonly during the fetal period. We then examined the expres-sion patterns during liver ontogenesis. The Ndk-10 antigenwas observed in the liver from E12 to adult (Figs. 4L—Q).The Ndk-10 antigen was strongly expressed in the liver in thefetal period, and the expression was lower in neonatal andadult livers. It is interesting to note that the intensity of Ndk-10 antigen expression correlates well with the supportivefunction of hematopoietic cells in liver.

Identification of Ndk-10 Antigen In order to character-ize Ndk-10 antigen, cell lysates from FL stromal cells wereimmunoprecipitated and immunoblotted with Ndk-10. Asshown in Fig. 5A, Ndk-10 precipitated an approximately150 kD protein from FL stromal cells. Next, we examined thetissue distribution of the Ndk-10 antigen by immunoblot

analysis. Fig. 5B indicates the expression of Ndk-10 antigenin various tissues. Except for FL stromal cells (positive con-trol), the 150 kD protein was detected only in FL and kidney.Especially, the kidney showed the strongest expressionamong the tissues examined. Therefore, we performed aminoacid sequence analysis of the Ndk-10 antigen immunoprecip-itated from kidney. It was found that a 14-amino acid se-quence at the N-terminal of the Ndk-10 antigen was identicalto that of mouse APN/CD13. Then, we immunoprecipitatedthe 150 kD protein from FL adherent cell lysates with a com-mercially available anti-mouse CD13 mAb (R3-242), andimmunoblotted with Ndk-10. As shown in Fig. 5C, Ndk-10as well as the anti-CD13 mAb immunoprecipitated the150 kD protein, indicating that the Ndk-10 antigen is mouseAPN/CD13.

Analysis of CD13 Peptidase Activity on FL StromalCells The peptidase activity of cell surface CD13 on FLstromal cells was measured by a spectrophotometric assay.When alanine-p-nitroanilide was incubated with FL stromalcells, p-nitroaniline liberated by cleavage of the substrate wasdetected (O.D. 405 nm�1.03�0.22 was obtained as com-

December 2004 2017

Fig. 4. Expression of Ndk-10� Cells in Various Tissues

Ndk-10 (with peroxidase) (A—Q) and H–E staining (A�—Q�) are shown. A—E (and A�—E�) are tissues of E14 embryo; brain (A, A�), intestine (B, B�), thymus (C, C�), heart(D, D�), and liver (E, E�). F—K (and F�—K�) are tissues of adult C57BL/6 mouse ; brain (F, F�), heart (G, G�), kidney (H, H�), lung (I, I�), spleen (J, J�), and thymus (K, K�). L—Q(L�—Q�) are various developmental stages of embryonic liver; E12 (L, L�), E14 (M, M�), E16 (N, N�), E18 (O, O�), neonatal (P, P�), and adult (Q, Q�). Bars: 100 mm in A (applica-ble to A—K), 50 mm in L (applicable to L—Q).

pared with the cell-free negative control). Because this assaydetects the CD13 peptidase activity specifically, the resultsuggests that CD13 on FL stromal cells has peptidase activ-ity. As shown in Fig. 6, when Ndk-10 was added to FL stro-mal cells, the peptidase activity was significantly inhibited,indicating that Ndk-10 downregulates the peptidase activityof CD13 on FL stromal cells.

Effect of Ubenimex on the Survival of Fetal LiverHematopoietic Cell Fraction The effect of Ubenimex, aninhibitor of APN/CD13,21) on hematopoietic cell survivalwas examined. As shown in Fig. 7A, the recovery of c-kit�

cells decreased when Ubenimex was added to the culture.Similar to the experiment with Ndk-10, the percentage of c-kit� cells also decreased (data not shown). The inhibitory ef-

fect of Ubenimex was also observed in the FLOC system(Fig. 3B). As shown in Fig. 7B, the addition of Ubenimex in-creased the cell death. Apoptotic cell death was higher in thegroup of Ubenimex (41—43%) than control group (21—27%). Since the APN/CD13 is not expressed on the surfaceof non-adherent cell fraction of FL cells, some unknownmechanisms are operating in the induction of cell death byanti-APN/CD13 or Ubenimex.

Taken together, the results suggest that the peptidase activ-ity of cell surface APN/CD13 is important for the mainte-nance of c-kit� hematopoietic cells.

DISCUSSION

Addition of a newly established mAb directed to FL stro-mal cells, Ndk-10, to the coculture of FL adherent (stromal)cells and non-adherent (hematopoietic) cells decreased therecovery of c-kit� cells (Fig. 3A). In the coculture system,the FL-derived, non-adherent cell fraction was used as ahematopoietic cell source. Although the fraction containscell types other than hematopoietic cells, the cells in the frac-tion could not survive alone and did not express the Ndk-10antigen. Thus, it is clear that FL stromal cells and the cellsurface Ndk-10 antigen are important for the maintenance ofc-kit� cells. The effect of Ndk-10 was confirmed by theFLOC system (Fig. 3B). We then examined the expression ofthe Ndk-10 antigen on various tissues by immunohistochemi-cal (Fig. 4) and immunoblotting (Fig. 5) analyses, and finally

2018 Vol. 27, No. 12

Fig. 5. Immunoblot Analyses of the Ndk-10 Antigen

Various cell lysates were immunoprecipitated (IP) with Ndk-10 or anti-CD13 mAb(R3-242). Proteins in the immunoprecipitates were resolved by 7.5% SDS-PAGE, fol-lowed by immunoblotting (IB) with Ndk-10. Lysates of fetal liver stromal cells (A), andcell lysates of various tissues (B) were immunoprecipitated with Ndk-10. The lysate offetal liver stromal cells (C) was immunoprecipitated with Ndk-10 or anti-CD13 mAb(R3-242).

Fig. 6. The Peptidase Activity of APN/CD13 Is Inhibited by Ndk-10

FL stromal cells were incubated at 37 °C with alanine-p-nitroanilide, a substrate forCD13. After 1 h, the absorbance of the cell-free supernatants was measured at O.D.405 nm to detect the amount of free p-nitroaniline. Before peptidase analysis, FL stro-mal cells were pre-cultured with Ndk-10 (10 mg/ml) or isotype-matched control IgG for12 h. Results are given as percents compared with medium only. ∗ p�0.01 comparedwith control IgG. Statistical analysis was performed by Student’s t-test.

Fig. 7. Recovery of c-kit� Cells on FL Stromal Cells Is Inhibited byUbenimex

(A) FL non-adherent (hematopoietic) cells were cultured on FL stromal cells in thepresence of Ubenimex (1 mg/ml: 3.2 mM, 10 mg/ml: 32 mM, 100 mg/ml: 320 mM), an in-hibitor of APN/CD13. Four days later, the floating cells were recovered and counted.The cells were then stained with FITC-conjugated anti-c-kit mAb (ACK2), and percent-ages of c-kit� cells were measured by flow cytometry. Results are given as percent re-covery compared with medium only. Recovered c-kit� cells were ranged from 6.3�106

to 11.8�104 in each experiment. Statistical analysis was performed by Student’s t-test(∗1: p�0.006, ∗2: p�0.09, ∗3: p�0.002, compared with medium control). (B) FL non-adherent cells were cultured on FL stromal cells in the presence or absence of Uben-imex (100 mg/ml: 320 mM). Recovered cells were stained with anti-annexin V and PI.Closed column: apoptotic cell death (annexin V-positive, PI-negative), open column:necrotic cell death (annexin V-positive, PI-positive). Data show means of two indepen-dent experiments.

the amino acid sequence data definitively revealed that theNdk-10 detects CD13 molecule. Because E14 FL adherentcells expressed albumin when they cultured (4 d, data notshown), CD13 expressing FL adherent cells are thought to befetal hepatocytes or its precursors.

APN/CD13 is one of the membrane-bound metallopepti-dases that preferentially degrade proteins with an N-terminalneutral amino acid. In humans, CD13 is expressed in variousmyeloid lineage cells, including monocytes, granulocytes,and immature myeloid cells, and is used as a myeloid lineagemarker.22) But in mouse hematopoietic cells, CD13 is ex-pressed only in mature macrophages and dendritic cells.23,24)

This fact enabled us to investigate the role of CD13 on stro-mal cells. In non-hematopoietic cells, CD13 is expressed onepithelial cells of the intestine and kidney, synaptic mem-branes in the central nervous system, fibroblasts, endothelialcells, and some human tumor cells.22) It has also been re-ported that CD13 is expressed in human BM stromal cells.22)

However, little is known about the role of CD13 molecule onstromal cells.

In the hematopoietic microenvironment, it is known thatstromal cells are an important element in the maintenanceand differentiation of hematopoietic cells.7—11) In recentyears, several reports have indicated that some proteases per-form essential roles in hematopoiesis. For instance, Heissiget al. reported that MMP-9 induced in BM cells releases asoluble kit-ligand, permitting the transfer of endothelial cellsand hematopoietic stem cells from the quiescent proliferativeniche.25) Levesque et al. reported that VCAM-1 expressed onBM stromal cells is cleaved by neutrophil proteases.26) Thesereports indicate that the enzymatic digestion of membranebound molecules might be a key step in hematopoiesis onstromal cells. In the present study, we found that APN/CD13regulates the maintenance of hematopoietic cells on FL stro-mal cells. Similar to the reported proteases, CD13 may func-tion as an essential protease for regulating the hematopoieticmicroenvironment.

Known substrates of CD13 are enkephalines, opioid pep-tides,22) interleukin-1b , and extracellular matrix proteinssuch as collagen type-IV.27) Mishima et al. reported that thecell surface CD13 appear to allow prevents human leukemiccells to resist endothelial IL-8 induced cell death.28) Anti-APN/CD13 induced cell death (Fig. 7B) may be a result ofinhibition of APN/CD13-induced degradation of someapopotosis-blocking molecules, such as IL-8 in the case ofhuman leukemias,28) on the surface of FL stromal cells. Be-cause the Ndk-10 was less effective at the dose tested thanUbenimex (see Fig. 3B), apoptotic cell death of the Ndk-10group was not significantly different from the control group(data not shown). The molecular mechanisms of anti-APN/CD13 induced cell death are not yet known and thisshould be clarified in the future experiments.

We report here that the Ndk-10 blocks the peptidase activ-ity of CD13, and also an APN/CD13-specific inhibitor,Ubenimex, decreases cell recovery. Judging from these re-sults, it is suggested that the enzymatic activity ofAPN/CD13 is necessary for the maintenance of hematopoi-etic cells in the FL microenvironment (Figs. 6, 7). However,the natural substrate of CD13 in hematopoietic cell survivalremains unknown. Tani et al. reported that CD13 has achemotactic activity for T lymphocytes, and the activity is

dependent on its peptidase activity.29) The report proposedthe possibility that CD13 has function(s) other than cleavingsome substrates.

CD13 is also expressed in other hematopoietic tissues, in-cluding spleen, thymus (Fig. 4), and BM stromal cells.22) Weconfirmed that CD13 is also important for BM and thymusstromal cell functions, using a coculture method similar toFL coculture or fetal thymus organ culture methods (unpub-lished data). These results suggest that CD13-mediated cellu-lar interactions are common events during various develop-mental stages of hematopoietic cells. It is interesting to notethat the distribution of CD13 was altered during ontogenesis(Fig. 4). Therefore, there is a possibility that CD13 or its sub-strate may play different roles during development than inadult.

Using the newly established Ndk-10, we found that cellsurface APN/CD13 plays an important role in FL hemato-poiesis that is dependent on its peptidase activity. Our find-ings provide a novel model that may explain the molecularmechanisms of cell survival and development of hematopoi-etic cells in the FL microenvironment.

Acknowledgments This work is supported by grants-in-aid from the Ministry of Education, Culture, Sports, Scienceand Technology of Japan. N.S. is a recipient of a fellowshipfrom the Japan Society for the Promotion of Science. The au-thors are grateful to Dr. Margaret Dooley Ohto for criticalreading this manuscript.

REFERENCES

1) Medvinsky A., Dzierzak E., Cell, 86, 897—906 (1996).2) Ema H., Douagi I., Cumano A., Kourilsky P., Eur. J. Immunol. 28,

1563—1569 (1998).3) Ema H., Nakauchi H., Blood, 95, 2284—2288 (2000).4) Rebel V. I., Miller C. L., Eaves C. J., Lansdorp P. M., Blood, 87,

3500—3507 (1996).5) Watanabe Y., Aiba Y., Katsura Y., Cell. Immunol., 177, 18—25 (1997).6) Dexter T. M., Allen T. D., Lajtha L. G., J. Cell. Physiol., 91, 335—344

(1976).7) Nishikawa M., Ozawa K., Tojo A., Yoshikubo T., Okano A., Tani K.,

Ikebuchi K., Nakauchi H., Asano S., Blood, 81, 1184—1192 (1993).8) Winemann J., Moore K., Lemischka I., Muller-Sieburg C., Blood, 10,

4082—4090 (1996).9) Miyake K., Weissman M. I., Greenberger J. S., J. Exp. Med., 173,

599—607 (1991).10) Friedrich C., Zausch E., Sugrue S. P., Gutierrez-Ramos J. C., Blood,

87, 4596—4606 (1996).11) Palacios R., Golunski E., Samaridis J., Proc. Natl. Acad. Sci. U.S.A.,

92, 7530—7534 (1995).12) Hackney J. A., Charbord P., Brunk B. P., Stoeckert C. J., Lemischla I.

R., Moore K. A., Proc. Natl. Acad. Sci. U.S.A., 99, 13061—13066(2002).

13) Kinoshita T., Sekiguchi T., Xu M. J., Ito Y., Kamiya A., Tsuji K.,Nakahata T., Miyajima A., Proc. Natl. Acad. Sci. U.S.A., 96, 7265—7270 (1999).

14) Kamiya A., Kinoshita T., Ito Y., Matsui T., Morikawa Y., Senba E.,Nakashima K., Taga T., Yoshida K., Kishimoto T., Miyajima A.,EMBO J., 18, 2127—2136 (1999).

15) Nishikawa S., Kusakabe M., Yoshinaga K., Ogawa M., Hayashi S., Ku-nisada T., Era T., Sakakura T., Nishikawa S.-I., EMBO J., 10, 2111—2118 (1991).

16) Vermes I., Haanen C., Steffens-Nakken H., Reutelingsperger C., J. Im-munol. Methods, 184, 39—51 (1995).

17) Owen J. J. T., Cooper M. D., Raff M. C., Nature (London), 249, 361—363 (1974).

18) Ceredig R., ten Boekel E., Rolink A., Melchers F., Andersson J., Int.

December 2004 2019

Immunol., 10, 49—59 (1998).19) Ashmun R. A., Shapiro L. H., Look A. T., Blood, 79, 3344—3349

(1992).20) Ogawa M., Matsuzaki Y., Nishikawa S., Hayashi S., Kunisada T., Sudo

T., Kina T., Nakauchi H., Nishikawa S.-I., J. Exp. Med., 174, 63—71(1991).

21) Umezawa H., Aoyagi T., Suda H., Hamada M., Takeuchi T., J. An-tibiot., 29, 97—99 (1976).

22) Shipp M. A., Look A. T., Blood, 82, 1052—1070 (1993).23) Leenen P. J. M., Melis M., Kraal G., Hoogeveen A. T., Ewijk W., Eur.

J. Immunol., 22, 1567—1572 (1992).24) Hansen A. S., Noren O., Sjostrom H., Werdelin O., Eur. J. Immunol.,

23, 2358—2364 (1993).

25) Heissig B., Hattori K., Dias S., Friedrich M., Ferris B., Hackett N. R.,Crystal R. G., Besmer P., Lynden D., Moore M. A. S., Werb Z., RafiiS., Cell, 109, 625—637 (2002).

26) Levesque J. P., Takamatsu Y., Nilsson S. K., Haylock D. N., SimmonsP. J., Blood, 98, 1289—1297 (2001).

27) Saiki L., Fujii H., Yoneda J., Abe F., Nakajima M., Tsuruo T., AzumaI., Int. J. Cancer, 54, 137—143 (1993).

28) Mishima Y., Matsumoto-Mishima Y., Terui Y., Katsuyama M., Ya-mada M., Mori M., Ishizaka Y., Ikeda K., Watanabe J., Mizunuma N.,Hayasawa H., Hatake K., J. Natl. Cancer Inst., 94, 1020—1028(2002).

29) Tani K., Ogushi F., Huang L., Kawano T., Tada H., Hariguchi N., SoneS., Am. J. Respir. Crit. Care Med., 161, 1636—1642 (2000).

2020 Vol. 27, No. 12