Stimulation of in vitro hematopoiesis by a murine fetal hepatocyte clone through cell—cell contact

JOURNAL OF CELLULAR PHYSIOLOGY 160:445-154 (1994)

Stimulation of In Vitro Hematopoiesis by a Murine Fetal Hepatocyte Clone Through

Cell-CeII Contact MASANOBU NANNO,* MASAHIRO HATA, HlDEYUKl DOI, SUSUMU SATOMI,

HlDEKl YAGI, TSUNEAKI SAKATA, RYUJI SUZUKI, AND TSUNETOSHI ITOH Yakult Central Institute for Microbiological Research (M.N.), Kunitachi, Tokyo 186,

Departments of Immunology and Embryology (M.N., M.H., H.Y., T.I.) and Surgery (M.H., H.D., S.S.), Tohoku University Schooi of Medicine, Sendai, Miyagi 980, and Shionogi

Research Laboratories (T.S., R.S.), Shionogi & Co., ltd., Fukushima-Ku, Osaka 553, japan

We have previously shown that a fetal liver-derived epithelial cell clone, FHC- 4D2, could support hematopoiesis in vitro through i ts colony-stimulating factor (CSF) activities in a short-term culture. In this study, since FHC-4D2 cells were found capable of maintaining hematopoietic progenitors in the coculture for a long time, we examined how FHC-4D2 could exert hematopoietic supporting activity in a long-term culture by coculturing adult bone marrow (BM) cells or fetal liver (FL) cells on a monolayer of FHC-4D2 cells. This clone could maintain the colony-forming unit of granulocytes and macrophages (CFU-GM) of BM for a 1 2 weeks under the coculture condition, but the fibroblastic cell clone from the fetal liver, FHC-4A3, could not support the survival of CFU-GM, even for 1 week. In addition to BM CFU-GM, the FHC-4D2 clone also supported the survival of FL CFU-GM, burst-forming unit of erythroid cells (BFUe), and colony-forming unit of mixed progenitors (CFU-Mix) for longer than 4 weeks. When BM cells were separated by a membrane filter from the FHC-4D2 cells in the coculture, the comparable number of CFU-GM was maintained at day 3, but virtually no he- matopoietic progenitors were detected at the end of the first week. CFU-GM were present in both nonadherent and adherent cells to the FHC-4D2 cells at day 3 of the coculture, but at day 7, the adherent population contained greater number of CFU-GM. CFU-GM derived from the adherent cells formed larger colonies and contained more bipotential CFU-GM than the nonadherent population. When BM cells from mice given 5-fluorouracil were cocultured with FHC-4D2 cells under the limiting dilution condition, interleukin-3 (IL-3)-responsive CFU-GM were in- duced from immature hematopoietic progenitor cells that were otherwise unre- sponsive to IL-3. From these data we conclude that the FHC-4D2 clone could generate and maintain IL-3-responsive hematopoietic progenitors via close con- tact and that, in the fetal liver, the contact between hepatocytes and hematopoi- etic cells may be critically important in inducing the differentiation of resting, IL-3-unresponsive immature hematopoietic cells into CFU-GM (progenitors re- sponsive to IL-3) and in triggering the self-renewal of CFU-GM. 0 1994 Wiley-Liss, Inc

INTRODUCTION The hematopoietic system has a hierarchical struc-

ture consisting of a pluripotent stem cell, lineage-com- mitted progenitor cells, and mature blood cells. Their growth and differentiation are regulated by various hu- moral factors. Besides these factors, the direct interac- tion of hematopoietic cells and stromal cells in the he- matopoietic microenvironment, such as the fetal liver or the adult bone marrow, has been considered indis- pensable for hematopoiesis.

To analyze the function of hematopoietic stromal cells in hematopoiesis, long-term culture of adult bone marrow cells was established (Dexter et al., 1977; Whit- lock and Witte, 1982). These in vitro systems enable us to manipulate myelopoiesis and B lymphopoiesis and to 0 1994 WILEY-LISS, INC.

show clearly that stromal cells play an important role for the growth, differentiation, and survival of he- matopoietic cells. Furthermore, a number of stromal cell lines established from hematopoietic tissues during the last 10 years were found to support the survival in vitro of hematopoietic progenitor cells, including col- ony-forming unit in spleen (CFU-S), CFU-Mix, CFU- GM, and BFUe (Harigaya et al., 1981; Kodama et al., 1982; Zipori et al., 1984; Song et al., 1985; Tsai et al.,

Received October 1,1993; accepted February 11,1994. *To whom reprint requestdcorrespondence should be addressed at Department of Immunology and Embryology, Tohoku University School of Medicine, 2-1 Seiryo-machi, Sendai, Miyagi 980, Japan.

NANNO ET AL. 446

1986; Abkowitz et al., 1987; Campbell et al., 1987; Hunt et al., 1987; Rennick et al., 1987; Whitlock et al., 1987; Nishikawa et al., 1988; Rios and Williams, 1990). Al- though most cell lines were reported to secrete he- matopoietic growth factors, such as macrophage colony- stimulating factor (M-CSF), granulocyte-macrophage CSF (GM-CSF), multi-CSF (interleukin-3; IL-31, IL-6, and IL-7, the precise hematopoiesis-supporting mecha- nism exerted by these cell lines has not been suffi- ciently clarified.

By contrast, the close contact between hematopoietic cells and stromal cells has been observed, suggesting the possibility that cell-cell contact may transduce sig- nals in hematopoietic progenitor cells for proliferation andlor differentiation. Recently, various adhesion mol- ecules have been reported to be expressed on he- matopoietic cells, although their physiological func- tions in hematopoiesis have not yet been delineated (Lewinson et al., 1990; Miyake et al., 1990; Kina et al., 1991; Verfaillie et al., 1991; Williams et al., 1991).

The fetal liver is an important hematopoietic organ during the embryonic life, where hematopoietic cells of multiple lineages are generated. Morphological analy- ses revealed that hepatocytes are almost exclusively in close contact with hematopoietic cells in situ (Medlock and Harr, 1983; Asano et al., 1987). The precise func- tions of hepatocytes in fetal hematopoiesis, however, have not been elucidated. To investigate the physiolog- ical role of fetal hepatocytes in hematopoiesis, we previ- ously established a hepatocyte clone capable of promot- ing the proliferation of hematopoietic cells from the murine fetal liver (FHC-4D2) (Hata et al., 1993). This clone has been found to support the survival of he- matopoietic progenitor cells in short-term coculture (for <3 days) through M-CSF and GM-CSF secreted by FHC-4D2 cells (Hata et al., 1993).

In this study, the capacity of FHC-4D2 clone to sup- port the generation of hematopoietic progenitor cells via cell-cell contact in addition to its CSFs was exam- ined by coculturing hematopoietic cells in a long-term culture. We show that a close contact between he- matopoietic cells and FHC-4D2 clone is essential to maintain in vitro the hematopoietic progenitor cells for a longer period (for ~ 1 2 weeks) and that this interac- tion also induces the differentiation of immature he- matopoietic progenitor cells (CSF-unresponsive cells) into CFU-GM (CSF-responsive cells). These findings suggest that FHC-4D2 exerts its hematopoietic sup- porting function (to maintain hematopoietic progeni- tors in a long-term culture) through cell-cell contact. Accordingly, it strongly supports the involvement of hepatocytes in hematopoiesis in the fetal liver through the close association with hematopoietic cells, previ- ously identified only morphologically until now.

MATERIALS AND METHODS Mice

Male BALB/c and BCF, mice raised in a specific pathogen-free condition were purchased from Shizuoka Laboratory Animal Center (SLC) (Hamamatsu, Japan) and used at the age of 6 1 2 weeks. For preparation of fetal liver cells, fetuses were obtained at the gestational age of 15 days. Mice were given food and water ad libitum.

Culture of cell lines FHC-4D2 clone, established from the liver of day 15

BALB/c fetuses, was used throughout experiments (Hata et al., 1993). This clone was cultured in Dulbec- CO’S modified Eagle’s medium (DME; GIBCO Laborato- ries Life Technologies, Inc., Grand Island, NY) supple- mented with 10% fetal calf serum (FCS; Biocell Lab., Carson, CA), insulin (Sigma Chemical Co., St. Louis, MO, USA) at 5 mg/ml, and epidermal growth factor (Boehringer Mannheim Biochemica, Mannheim, Ger- many), at 10 ng/ml. Fetal liver-derived fibroblastic cell clone, FHC-4A3, and adult bone marrow-derived fibro- blastic cell line, BM1, were cultured in DME supple- mented with 10% FCS (FCS/DME).

Coculture of hematopoietic cells and stromal cells

Stromal cell lines were subcultured 2 days before the onset of coculture, when they became subconfluent. They were plated in flasks or multiwell plates (2.5 x lo6 cells in a 10-ml/T-25 flask; 4 x lo5 cells in a 2-muwell of a 6-well plate; 2 x lo5 cells in a 1-ml/well of a 12-well plate; 2.5 x lo4 cells in a 0.2-ml/well of a 96-well plate). ARer removing the medium, monolayers of the stromal cell lines were washed with prewarmed FCS/DME, and hematopoietic cells were overlaid.

Bone marrow cells were prepared from femurs of adult mice. After the erythrocytes were lysed with a hemolytic buffer, adherent cells were removed by pan- ning on FCS-coated dishes. To enrich hematopoietic progenitor cells, nonadherent bone marrow cells were centrifuged by discontinuous-density gradient of Percoll and the cells with a buoyant density of 1.070- 1.075 g/ml (fraction 3) were collected (Hata et al., 1993). The fetal liver cells were prepared by mincing the liver from day 15 fetus by forceps. To obtain immature he- matopoietic progenitor cells from the bone marrow, 5-fluorouracil (5-FU) (Kyowa Hakko Kogyo Co., Ltd., Tokyo, Japan) was injected at the dose of 75 mg/kg IV twice at a 6-day interval by slightly modifying the method by Hodgson and Bradley (1979). This injection protocol induced more drastic depletion of CFU-GM than single shot of 150 mg/kg 5-FU. Bone marrow cells were prepared 2 days after the second injection and used after erythrocytes were lysed (post-5-FU BM cells).

Hematopoietic cells prepared as described were placed on the monolayers of stromal cell lines in FCS/ DME (1 x lo7 bone marrow cells or 4 x lo6 fetal liver cells in a T-25 flask; 1 x lo6 fraction 3 cells/well or 6 x lo5 fetal liver cells/well of a 6-well plate; 5 x lo5 bone marrow cells/well of a 12-well plate; 6 x lo5 post- 5-FU bone marrow cells/well of a 6-well plate). To in- hibit the contact between hematopoietic cells and stro- ma1 cells, Millicell (Millipore Products Division, Bedford, MA) was set in a culture well already contain- ing FHC-4D2 cells; hematopoietic cells were poured in- side the Millicell. For limiting dilution anal sis usin

or 2.5 x lo3 cells on the monolayer of FHC-4D2 cells. From each culture, nonadherent cells were collected by gentle pipetting. To harvest adherent cells, trypsinl ethylendiamine tetraacetic acid (EDTA) (GIBCO) was added to the culture well after removal of the nonadher-

’

post-5-FU BM cells, we plated 2 x 1 0 , l x 10 x , 5 x 10 8 ,

CONTACT-DEPENDENT HEMATOPOIESIS BY A FETAL HEPATOCYTE CLONE 447

0 2 4 6 8 1 0 1 2 0 2 4

0 0 2 4 6 8 1 0 1 2 0 2 4

Weeks in culture

Fig. 1. Effect of stromal cell clones and line on the survival of CFU-GM in long-term coculture. Bone marrow cells prepared from adult BALB/c (A,B) or BCF, (C,D) mice were cultured at 10' cells/ flask in medium alone (o), or on the monolayers of FHC-4D2 cells (a), FHC-4A3 cells (A), or BMl cells (W). At intervals, nonadherent cells were collected and the numbers of total cells (A,C) and CFU-GM (B,D) were counted.

ent cells, and the adherent population and the FHC- 4D2 cells were incubated at 37°C for 20 min. Thereaf- ter, detached cells were obtained by gentle pipetting. The numbers of total cells and hematopoietic progeni- tor cells were measured at intervals.

In the long-term coculture shown in Figure 1, nonad- herent hematopoietic cells were collected every week as stated; one-half of them were then passaged onto the freshly prepared monolayer of FHC-4D2 cells, and the newly set coculture was continued fkther; the remain- ing half of the collected cells was subjected to the deter- mination of the number of total cells and CFU-GM. Since the number of hematopoietic cells thus decreased to one-half the number a week ago, the actual number was multiplied by 2,.1 at n weeks of the culture (ex- pressed as accumulated numbers) to permit comparison on the initial cell number basis.

In some cultures, conditioned medium prepared from confluent FHC-4D2 cells (4D2-CM), CSF fractions pu- rified from 4D2-CM (Hata et al., 19931, normal goat serum, or goat antimouse GM-CSF antiserum (Hata et al., 1993) were added to the coculture. Measurement of hematopoietic progenitor cells The number of hematopoietic progenitor cells was

enumerated with the clonal culture in a semisolid me- dium (Bradley and Metcalf, 1966; Metcalf et al., 1979). CFU-GM in the bone marrow cells were assessed in the soft agar culture: an underlayer, 1 ml of DME contain- ing L-asparagine (10 pg/ml), 20% FCS, 0.5% Bacto Agar (Difco Laboratories, Detroit, MI), and 20% WEHI3 con- ditioned medmm (WEHI3-CM) (Ihle et al., 1982); an overlayer, 1 ml of DME containing L-asparagine (10

p,g/ml), 20% FCS, and 0.3% Bacto Agar. They were cultured for 7 days and aggregates containing more than 50 cells were counted as colonies. For the fetal liver population, CFU-GM, BFUe, and CFU-Mix were differentially counted under a microscope after a 7-day- cultivation in a medium of 1:l mixture of Iscove's mod- ified DME and a medium (Kojin Bio Co., Ltd., Sakado, Japan), 20% FCS, 0.88% methylcellulose (Iwai Ka- gaku, Co., Ltd., Tokyo, Japan), 10% pokeweed mitogen- stimulated mouse spleen cell CM (PWM-SCM) (Cutler et al., 1985), and 2 U/ml erythropoietin (EPO) (Con- naught Laboratories Ltd., Willowdale, Ontario, Can- ada).

To identify colony types, CFU-GM were cultured in 1 ml of DME containing L-asparagine (10 pg/ml), 20% FCS, 0.88% methylcellulose, and 10% WEHI3-CM for 7 days; the colonies were picked up with an Eppendorf pipette, fixed on a glass slide, and observed under a microscope after May-Griinwald-Giemsa staining.

Measurement of proliferative response of bone marrow cells

Growth-promoting activities of CSFs purified from serum-free FHC-4D2 CM by gel chromatography (Hata et al., 1993) were assessed by culturing fraction 3 cells for 3 days at 5 x lo4 cells/well in the presence of se- quentially diluted samples in a microtiter plate. Cells were pulsed with 0.5 pCi of 3H-thymidine for the last 16 hr, and the radioactivity was counted.

RESULTS Survival of hematopoietic progenitor cells in

coculture with fetal liver-derived stromal cell clones

FHC-4D2 clone, established from the murine fetal liver and identified as fetal hepatocytes by Hata et al. (1993), was used throughout the experiments. Fetal liver-derived fibroblastic FHC-4A3 clone having the spindle-like shape (Hata et al., 1993) was also used. The BMl cell line, consisting of cells morphologically iden- tifiable as fibroblasts, was established from the bone marrow of adult BALB/c mice in FCSDME (Nanno, M., Hata, M., and Itoh, T., unpublished observations).

To examine whether these stromal cell clones and line could support in vitro hematopoiesis under the co- culture condition (via cell-cell contact), bone marrow cells from syngenic adult mice (BALB/c) were overlaid on and cocultured with the monolayer of FHC-4D2, FHC-4A3, or BM1 cells. In the absence of stromal cells, most hematopoietic cells died by the end of the 1st week, but both FHC-4D2 and FHC-4A3 clones appar- ently maintained the bone marrow cells for >5 weeks in terms of the cell number (Fig. 1A). FHC-4D2 cells could also support the survival of CFU-GM for up to 12 weeks, whereas CFU-GM were not detected after 2 weeks of coculture with FHC-4A3 cells. BM1 cells hardly maintained CFU-GM, even for 1 week (Fig. 1B). When bone marrow cells from semi-allogenic BCFl mice were cocultured with these stromal cells, the re- sult was almost similar in terms of the numbers of total cells (Fig. 1C) and CFU-GM (Fig. 1D). The proliferating hemopoietic cells harvested at the end of the 3rd week of the coculture on the FHC-4D2 monolayer included macrophages, granulocytes, and erythroblasts (Fig. 2),

448 NANNO ET AL.

Fig. 2. Morphology of the proliferating cells in coculture. Bone marrow cells prepared from adult BALB/c mice were cultured at 10' cellslflask on the monolayers of FHC-4D2 cells for 3 weeks and the smear of nonadherent cells was stained with May-Griinwald-Giemsa solution. x 1,000.

suggesting that the FHC-4D2 clone supports the sur- vival of hematopoietic cells of various lineages in the adult bone marrow cells.

To quantitate the effect of FHC-4D2 clone on the survival of erythroid progenitor cells, hematopoietic cells in liver cells from day 15 fetuses of BALB/c mice were cocultured. Although neither CFU-GM nor BFUe could survive for 1 week in medium alone, FHC-4D2 clone supported both for a t least 4 weeks (Fig. 3). This finding suggests that FHC-4D2 clone support the sur- vival of hematopoietic progenitors of erythroid, as well as myeloid lineages in the fetal liver, in addition to its supporting effect on BM progenitors.

Involvement of cell-cell contact in the survival of hematopoietic progenitor cells supported by

FHC-4D2 clone Since FHC-4D2 clone was found to produce he-

matopoietic growth factors, constitutively including M-CSF and GM-CSF (Hata et al., 1993), we investi- gated whether soluble factors play a critical role in the long-term survival of hematopoietic progenitor cells by FHC-4D2 clone. When fraction 3 cells in the bone mar- row with the buoyant density of 1.070-1.075 g/ml were cocultured with FHC-4D2 cells, the membrane filters

(Millicell; pores size, 0.45 pm) were placed to separate hematopoietic cells from the FHC-4D2 clone. At day 3 of the coculture, the cultures showed basically no differ- ence in the number of CFU-GM between those with or without the membrane filter as reported previously (Hata et al., 1993). At day 7 of the coculture, however, the numbers of CFU-GM in the cultures with the mem- brane filter decreased by more than 95%, compared to those in the direct coculture with FHC-4D2 (Fig. 4). Cultures with a WEHI3 conditioned medium (WEHI3- CM; crude IL-3) was used as positive controls and the same number of resultant CFU-GM in the presence or absence of the filter were recovered a t day 7 (data not shown), implicating that a nonspecific inhibitory effect was not detected at all when the filter was used.

We then examined whether the essential role of FHC-4D2 via cell-cell contact in hematopoiesis-sup- porting activity in a long-term culture was also demon- strated when fetal liver cells were used as a source of hematopoietic progenitor cells. It is known that CFU of multiple lineages can readily be inducible from fetal liver hematopoietic progenitor cells. When fetal liver cells were cocultured on the monolayer of FHC-4D2 clone using the same membrane filter, the numbers of CFU-GM, BFUe, and CFU-Mix were obviously reduced

CONTACT-DEPENDENT HEMATOPOIESIS BY A FETAL HEPATOCYTE CLONE 449

71 I

'ij .L 0 1 2 3 4

0 1 2 3 4 Weeks in culture

Fig. 3. Effect of FHC-4D2 clone on the survival of CFU-GM and BFUe in long-term coculture. Fetal liver hematopoietic cells prepared from BALB/c mice were cultured at 4 X 10' celldflask in medium alone (0) or on the monolayers of FHC-4D2 cells (0). At intervals, nonadherent cells were collected, and the numbers of total cells (A), CFU-GM (B), and BFUe (C) were counted.

--C FHC-4D2 ... Q... FHC-dD2+Millicell + Medium

Days in culture

Fig. 4. Effect of Millicell on the survival of CFU-GM in the adult bone marrow supported by FHC-4D2 clone. Fraction 3 cells from the adult bone marrow were cultured in medium alone (A) or on the mono- layer of FHC-4D2 cells at 1 x lo6 cells/well of a 6-well plate without (0) or inside (0) the Millicell for 21 days. Nonadherent cells were collected by gentle pipetting, and the number of CFU-GM was mea- sured.

compared to those of the control culture even at day 3, and virtually no hematopoietic progenitor cells were detected at day 7 (Table 1). These results demonstrate that the hematopoietic progenitor cells of various lin- eages are similarly supported via cell-cell contact with FHC-4D2 clone.

Generation of CFU-GM from the hematopoietic progenitor cells that adhered to FHC-4D2 cells Although we harvested only cells that were easily

obtained by gently pipetting from the coculture for the further colony assay described above, a certain amount of cells present beneath the FHC-4D2 cells were always observed. To analyze qualitatively and quantitatively the hematopoiesis supporting function exerted by FHC- 4D2 cells, especially through cell-cell contact, bone marrow fraction 3 cells were cocultured according to a protocol shown in Figure 5 by examining what influ- ence FHC-4D2 cells have on cells which adhere (or pen- etrate) firmly to (or underneath) FHC-4D2 cells. At day 3 of the coculture, nonadherent cells (harvestable only by gentJe pipetting) were once harvested by gentle pi- petting and transferred with fresh FCS/DME or with a FHC-4D2 conditioned medium (4D2-CM) into new wells containing FHC-4D2 cells. The remaining wells containing the FHC-4D2 cells, and the hematopoietic cells adhering to them (including those present beneath FHC-4D2 cells) were then fed with fresh FCS/DME or with 4D2-CM. All cells were separately cultured for 4 more days. As a control, fraction 3 cells were cocultured on the monolayer of FHC-4D2 clone for 7 days without any disturbance. The number of total CFU-GM was slightly reduced by day 3, and then increased when cocultured for another 4 days (data not shown). Al- though both nonadherent cells and cells adherent to FHC-4D2 cells a t day 3 of the coculture contained CFU- GM, the number of CFU-GM in the former population did not change during the secondary culture for 4 days. By contrast, CFU-GM in the adherent population in- creased by 2-fold during the same period and, further- more, the addition of 4D2-CM into the coculture re- markably stimulated the generation of CFU-GM to 4-fold compared to the number of CFU-GM at day 3 of the coculture (Fig. 6).

In the previous report (Hata et al., 1993)) we identi- fied M-CSF and GM-CSF in the 4D2-CM by gel chroma- tography; therefore, we examined which CSFs have the stimulatory effect on the generation of CFU-GM from the adherent population. When anti-GM-CSF antise- rum was added together with 4D2-CM, the increase in the number of CFU-GM was suppressed by 60% (Fig. 7A). Additionally, both M-CSF and GM-CSF fractions showed the similar degrees of CSF activity, but the GM-CSF fraction had a greater potent effect on the generation of CFU-GM (4.5-fold compared to control) than M-CSF fractions (2-fold to control) (Fig. 7B).

To elucidate the differentiation status of the nonad- herent and the adherent populations cocultured for 3 days with FHC-4D2 cells, we compared the potentials of CFU-GM in both the nonadherent and the adherent populations in terms of the proliferation and the differ- entiation. As in the histogram shown in Figure 8, CFU-GM in the nonadherent fraction formed colonies with a peak diameter of 0.5-1.0 mm and virtually no

450 NANNO ET AL

TABLE 1. Effect of Millicell on the survival of CFU-GM, BFUe, and CFU-Mix in the fetal liver supported by FHC-4D2 clone'

No. of BFUe No. of CFU-Mix Culture NO OfCFU-GM condition 3 days 7 days 3 days 7 days 3 days 7 days

No stroma <6 <5 <6 <5 <6 <5 FHC-4D2 88 -C 14 223 + 74 82 + 5 9 77 + 14 25 2 15 91 lr 33 FHC-4D2 + Millicell 16 t 0 <2 54 2 3 <2 1 1 t 8 <2

'Fetal liver hematopietic cells were cultured in medium alone or on the monolayer of FHC4D2 cells a t 6 X l o 5 celldwell without or inside the MdliceU for the lndieated period Nonadherent cells were collected by gentle pipetting, and the number of CFE-GM. BFUe, and CFU-Mix was measured. Input numhen of CFU-GM, BFUe, and CFU-Mix were 344.159, and 56iwel1, respectively.

Experimental Desiqn

Fraction 3 cells

FHC-4D2 cells Coculture for 3 days

Assay for CFU-GM::: non-ad. or without

1 1 ...... -.... : - . ?

. . : : j i . . l i

Assay for CFU-GM

Fig. 5. Experimental design to examine the functional difference between nonadherent and adherent populations in the coculturc of bone marrow cells with FHC-4D2 cells. This experiment was carried out with a 12-well plate.

colonies larger than those with a 2.5-mm diameter, whereas apparently larger colonies were formed by CFU-GM derived from adherent population; the peak diameter was 1.0-1.5 mm, and the larger colonies (>2.0 mm) greatly outnumbered (nearly 30% to the total number of the colonies) those of the nonadherent frac- tion (about 2%). Morphological examination revealed that colonies from CFU-GM in the control (not dis- turbed) coculture at day 7 consisted predominantly of both granulocytes and macrophages (bipotential). The majority of CFU-GM in the adherent fraction at day 3 formed colonies consisted of both granulocytes and macrophages. On the contrary, colonies detected in the nonadherent CFU-GM population were predominantly of macrophage type (monopotential) (Table 2).

Induction of CFU-GM from the immature hematopoietic progenitor cells by

FHC-4D2 clone Since cell-cell contact with FHC-4D2 seemed to have

a greater effect on the proliferation and differentiation

of more immature hematopoietic progenitors (adherent cell population) than CFU-GM detected in the nonad- herent fraction, we next tried to enrich immature pro- genitors from the bone marrow by intravenous injec- tion of 5-FU to elucidate whether FHC-4D2 clone has a stimulatory effect on this resting hematopoietic pro- genitor cell presumed to be one of the early stem cells. The number of CFU-GM increased severalfold in the coculture for 7 days, but 4D2-CM had no stimulatory effect on the induction of CFU-GM. Furthermore, FHC- 4D2 clone-induced CFU-GM generation was com- pletely inhibited by separating postd-FU BM cells from FHC-4D2 cells with the membrane filter (Fig. 9). We then examined whether the increase in CFU-GM was due to the self-renewal of pre-existing CFU-GM or to the differentiation from more immature hematopoi- etic progenitor cells into CFU-GM. Frequencies of CFU-GM in fresh post-5-FU-BM cells and in cocultured post-5-FU-BM cells for 7 days with FHC-4D2 cells were compared by culturing them under the limiting dilu- tion condition. Frequency of CFU-GM-containing wells reproducibly increased (1.5-2.5-fold) in three indepen- dent experiments (Table 31, indicating strongly that immature (CSF-irresponsive) hematopoietic progenitor cells in the post-5-FU BM cells differentiated into CFU-GM (CSF-responsive) during the coculture with FHC-4D2 clone.

DISCUSSION The fetal liver-derived epithelial cell clone, FHC-

4D2, supported the survival and generation of CFU-GM in the adult bone marrow for 12 weeks and of CFU-GM, BFUe, and CFU-Mix in the fetal liver for 4 weeks in the coculture condition, i.e., through cell-cell contact, even in the absence of any exogenously added cytokines or CSFs. Semi-allogenic cells as well as syn- genic cells were maintained. Since the fetal liver-de- rived fibroblastic cell clone, FHC-4A3, or the adult bone marrow-derived fibroblastic cell line, BM1, were found to exert actually no hematopoiesis-supporting ac- tivities in this study, we came to the conclusion that, in addition to its CSFs, FHC-4D2 clone has the specific function of stimulating, maintaining and differentiat- ing hematopoietic progenitors through cell-cell con- tact.

In the previous study we reported that FHC-4D2 clone constitutively produces M-CSF and GM-CSF and, in the short-term coculture (<3 days) supports the sur- vival of CFU-GM via soluble factors. It was clearly demonstrated in the present study that, if the coculture was continued for more than 1 week, the maintenance of hematopoietic progenitor cells was achieved only

CONTACT-DEPENDENT HEMATOPOIESIS BY A FETAL HEPATOCYTE CLONE 451

Fig. 6. Generation of CFU-GM from nonadherent and adherent populations in the coculture of bone marrow cells with FHC-4D2 clone. Fraction 3 cells were cultured as described in Fig. 5, and the number of CFU-GM in each culture was measured at day 3 and at day 7. Data were expressed as mean 2SE of results obtained from 3 independent experiments.

Fig. 7. Effect of CSFs secreted by FHC-4D2 clone on the generation of CFU-GM from the adherent population. Fraction 3 cells were cocul- tured on the monolayer of FHC-4D2 cells in wells of a 12-well plate for 3 days, and the adherent cells were cultured with or without 4D2-CM or partially purified CSF fractions of 4D2-CM (Hata et al., 1993). A Effect of anti-GM-CSF antiserum on the generation of CFU-GM in-

duced by 4D2-CM. B: Effects of CSF fractions on the generation of CFU-GM. Peak I, M-CSF fraction with MW. 400kD; peak 11, M-CSF fraction with MW 80 kD; peak ITI, GM-CSF fraction with MW 40 kD. Significance in difference of mean values was evaluated by nonpaired Student’s t-test. NS, not significant.

452 NANNO ET AL.

No. of CFU-GM/well @lo3)

Fig. 8. Proliferative potentials of CFU-GM derived from nonadher- ent cells and adherent cells at day 3 of the coculture with FHC-4D2 clone. Agar layers containing colonies were fixed with 3% glutaralde- hyde for 10 min and dried overnight. The size of the colonies was measured on a microscope (expressed as mean +SD of results obtained from three independent experiments).

TABLE 2. Type of CFU-GM maintained by FHC-4D2 clone'

Types of colonies (%)'

Cells cultured GM G M

Total cells 64 Nonadherent cells 25 Adherent cells 67

3 33 13 63 8 25

'Fraction 3 cells from the adult bone marmw were cocultured on the monolayer of FHC-4D2 clone for 7 days. At day 3, nonadherent cells were mllected and transferred to the fresh monolayer of FHC-4D2 cells; the fresh medium was then added to the remain- ing adherent layer. CFU-GM in each culture were allowed to form colonies at day 7. Colonies were picked up at random and their smears were stained with Mays*- wald-Giemsa solution. Numbers of colonies tested were 36 for total cells and 24 for nonadherent and adherent cells. The type of colonies formed from nonadherent cells is significant1 different from tho% formed from total cells or adherent cells when evalu- ated with &est (P < 0.001). 'GM. granulocyte-macrophage type; G, granulocyte type: M, macrophage type.

when hematopoietic cells werexultured in contact with FHC-4D2 cells. Hematopoietic growth factors, includ- ing M-CSF and GM-CSF produced by FHC-4D2 clone may have played an important role in stimulating the hematopoietic progenitor cells in the short-term cocul- ture, but they were insufficient for long-term survival of hematopoietic progenitor cells in vitro.

In the present analysis, we found that there were two distinct populations in the hematopoietic cells cocul- tured with FHC-4D2: one is a population easily ob- tained by gently pipetting from the coculture (the non- adhering fraction) and the other adhering rather firmly t o (or penetrating underneath) FHC-4D2 cells (the ad- herent fraction). We attempted to characterize these two populations and found the following. First, when the hematopoietic cells were cocultured with FHC-4D2 cells for 3 days, transferred, and kept separately for another 4 days with FHC-4D2 cells, the number of CFU-GM in the adherent fraction increased consider- ably (2- to 5-fold compared to the nonadherent popula- tion), whereas essentially no increment in the number of CFU-GM in the nonadherent fraction was observed. Second, the adherent fraction contained CFU-GM that

Fig. 9. Effect of FHC-4D2 clone on the generation of CFU-GM from immature hematopoietic progenitor cells. Postd-FU BM cells were cultured in medium alone, in the presence of 4D2-CM (loo%), or on the monolayer of FHC-4D2 clone at 6 X lo5 cells/well of a 6-well plate for 7 days. To separate post-5-FU BM cells from FHC-4D2 cells, Millicell was set inside the culture well. After the 7 days' culture, nonadherent cells were harvested, and the number of CFU-GM was assayed. Non- paired Student's t-test was performed.

TABLE 3. Limiting dilution analysis of CFU-GM generation induced by FHC-4D2 clone'

Frequency of CFU-GM-containing well Experiment Before coculture After coculture ARerEiefore

1 111,040 11700 1.5 2 111,210 1/480 2.5 3 11680 11400 1.7

'Posb5-FU BM cells were cultured in the presence of 10% WEM3-CM under the limit ing dilution condition to measure the frequency of CFU-GM before coculture. Those cells were cocultured on the monolayer of FHC-4D2 clone for 7 days under the same limiting dilution condition, and then the frequencyof CFU-GM containing well was measured. Difference in the frequency of CFU-GM-mntaining well before and aRer coculture is statistically significant with paired Student's t-test (P < 0.05).

could form larger colonies than the nonadherent frac- tion. These results clearly demonstrated that he- matopoietic progenitors adhering to FHC-4D2 cells have much stronger proliferating activity than that of those not in direct contact with FHC-4D2 cells. Third, the adherent population generated more bipotential CFU-GM capable of forming granulocyte/macrophage colonies than the nonadherent fraction; the majority of colonies formed by the nonadherent CFU-GM were mo- nopotential, suggesting that the adherent population contained more immature CFU than that of the nonad- herent fraction. Altogether, hematopoietic cells held by FHC-4D2 cells via cell-cell contact contained cells more immature than were found in the nonadherent population; immature hematopoietic cells are more ad- hesive to stromal cells like FHC-4D2 clone than more committed cells; as the progenitor cells lose the adhe- siveness to stromal cells, they become less active in cell proliferation and much more committed to a single lin- eage (less potential for differentiation). Interestingly, humoral factors produced by FHC-4D2 clone stimu- lated the generation of CFU-GM from the adherent population; this activity was at least in part attributed to both M-CSF and GM-CSF, although the latter seems more stimulatory.

CONTACT-DEPENDENT HEMATOPOIESIS BY A FETAL HEPATOCYTE CLONE 453

What are then the actual effects of FHC-4D2 cells via cell-cell contact on immature hematopoietic cells if we assumed that FHC-4D2 could really maintain imma- ture progenitor cells under the coculture condition? To this end, we used post-5-FU BM cells as representative immature progenitor cells. FHC-4D2 clone substan- tially supported the proliferation of hematopoietic pro- genitor cells in the post-5-FU BM cell population via cell-cell contact. The increase in number of CFU-GM in the coculture of post-5-FU BM cells (5-26-fold; Nanno, M., Hata, M., and Itoh, T., unpublished data) was much greater than that of fraction 3 BM cells (at most 2-fold in the absence of additional growth factors), and this increase was completely inhibited by the membrane filter. Furthermore, it was confirmed by limiting dilu- tion analysis that the frequency of CFU-GM-containing wells increased by 1.5-2.5-fold during coculture with FHC-4D2 cells for 7 days. The result clearly pointed out that CFU-GM (detectable progenitors in the colony as- say using multi-CSF) was generated from the imma- ture hematopoietic progenitor cells (irresponsive pro- genitors to multi-CSF) in contact with FHC-4D2 cells. The increase in frequency (1.5-2.5-fold) was much less than the increase in the total number of CFU-GM (5- 26-fold), suggesting that the self-renewal of pre-exist- ing CFU-GM is also the critical event resulting in the overall increase of CFU-GM in the coculture. FHC-4D2 cells not only stimulated the proliferation of immature hematopoietic progenitor cells but induced their differ- entiation into more mature CSF-responsive CFU-GM through direct contact.

It has recently been reported that the pluripotent hematopoietic stem cell attach to the culture dish (Kiefer et al., 1991). Various adhesion molecules such as Pgp-1 (CD44), VLA-4, or the ligand for fibronectin have been found to be expressed on hematopoietic pro- genitor cells (Lewinson et al., 1990; Miyake et al., 1990; Kina et al., 1991; Verfaillie et al., 1991; Williams et al., 1991). Besides their role in cell-cell adhesion, however, their physiological functions in hematopoiesis have not yet been delineated. Our data show that the putative adhesion molecules on FHC-4D2 clone may anchor im- mature hematopoietic progenitor cells, stimulate pro- liferation, and induce their differentiation into CFU-GM with the functional IL-3 receptor. Recently, stem cell factor (SCF) has been identified as the S1 gene product (Anderson et al., 1990; Zsebo et al., 1990) and found to act on immature hematopoietic progenitor cells synergistically with other hematopoietic growth factors (de Vries et al., 1991; Migliaccio et al., 19911, although the mechanism of SCF in such stimulation also remains to be elucidated. Therefore, we are now exploring the adhesion molecules expressed on the FHC-4D2 clone that are involved in stimulating CFU-GM generation from immature hematopoietic progenitor cells.

FHC-4D2 is an epithelial cell clone having the he- matopoiesis-supporting activity derived from the fetal liver where marked hematopoiesis takes place. In the coculture system, FHC-4D2 cells could be considered to maintain the hematopoietic progenitor cells through two distinct modes: the differentiation of immature he- matopoietic progenitor cells into CFU-GM induced by cell-cell contact and the stimulation of self-renewal of

CFU-GM also induced through cell-cell contact and at least partly via soluble factors. Taken together, FHC- 4D2 clone proved a useful tool with which to study the fetal liver hematopoiesis, and the detailed analysis of FHC-4D2 clone-induced hematopoiesis may lead to the discovery of the fetal liver specific hematopoietic regu- lation mechanism.

ACKNOWLEDGMENTS We gratefully acknowledge the valuable discussions

by Dr. M. Ohwaki, Dr. T. Tokumitsu, and Mrs. W. Ohyama. We also thank Ms. M. Watanabe for her secre- tarial assistance and Mr. D. Arakawa for reviewing English. This work was supported in part by a Grant-in Aid for Scientific Research from the Ministry of Educa- tion, Science and Culture (034541141, and by the Spe- cial Coordination Fund of the Science and Technology Agency of the Japanese Government, and by the Gonryo Medical Research Fund from the Gonryo Foun- dation.

LITERATURE CITED Abkowitz, J., Holly, R.D., Segal, G.M., and Adamson, J.W. (1987)

Multilineage, nonspecific hemopoietic growth factor(s) elaborated by a feline fibroblast cell line: Enhancement by virus infection. J . Cell. Physiol., 127.189-196.

Anderson, D.M., Lyman, S.D., Baird, A,, Wignall, J.M., Eisenman, J., Rauch, C., March, C.J., Boswell, H.S., Gimpel, S.D., Cosman, D., and Williams, D.E. (1990) Molecular cloningof mast cell growth factor, a hematouoietin that is active in both membrane bound and soluble forms. cell, 633235-243.

Asano, H., Kobayashi, M., and Hoshino, T. (1987) The hemopoietic microenviror&ent in the fetal liver of mice: Relationship between developing hepatocytes and erythroblasts. J . Electron Microsc., 36: 15-25.

Bradley, T.R., and Metcalf, D. (1966) The growth of mouse bone mar- row cells in vitro. Aust. J . Exp. Biol. Med. Sci., 44.287300.

Campbell, A.D., Long, M.W., and Wicha, M.S. (1987) A stromal cell line from myeloid long-term bone marrow cultures can support my- elopoiesis and B lymphopoiesis. J. Immunol., 138:1082-1087.

Cutler, R.L., Metcalf, D., Nicola, N.A., and Johnson, G.R. (1985) Puri- fication of a multipotential colony-stimulating factor from pokeweed mitogen-stimulated mouse spleen cell conditioned me- dium. J . Biol. Chem., 260:65794587.

de Vries, P., Brasel, K.A., Eisenman, J.R., Alpert, A.R., and Williams, D.E. (1991) The effect of recombinant mast cell growth factor on purified murine hemopoietic stem cells. J . Exp. Med., 173:1205- 1211.

Dexter, T.M., Allen, T.D., and Lajtha, L.G. (1977) Conditions control- ling the Droliferation of hemouoietic stem cells in vitro. J. Cell. Ph;siol., i1:335344.

Harigaya, K., Cronkite, E.P., Miller, M.E., and Shadduck, R.K. (1981) M&ne bone marrow cell line producing colony-stimulating factor. Proc. Natl. Acad. Sci. USA, 78.6963-6966.

Hata, M., Nanno, M., Doi, H., Satomi, S., Sakata, T., Suzuki, R., and Itoh, T. (1993) Establishment of a hepatocytic epithelial cell line from the murine fetal liver capable of promoting hemopoietic cell proliferation. J . Cell. Physiol., 154:381-392.

Hodgson, G.S., and Bradley, T.R. (1979) Properties of haematopoietic stem cells surviving 5-fluorouracil treatment Evidence for a pre- CFU-S cell? Nature (Lond.), 281:381-382.

Hunt, P., Robertson, D., Weiss, D., Rennick, D., Lee, F., and Witte, O.N. (1987) A single bone marrow-derived stromal cell type sup- ports the in vitro growth of early lymphoid and myeloid cells. Cell, 48:997-1007.

Ihle, J.N., Keller, J., Henderson, L., Klein, F., and Palaszynski, E. (1982) Procedures for the purification of interleukin 3 to homogene- ity. J. Immunol., 129~2431-2436.

Kiefer, F., Wagner, E.F., and Keller, G. (1991) Fractionation of mouse bone marrow by adherence separate primitive hemopoietic stem cells from in vitro colony-forming cells and spleen colony-forming cells. Blood, 78.2577-2582.

Kina, T., Majumdar, A.S., Heimfeld, S., Kaneshima, H., Holzmann, B., Katsura, Y., and Weissman, I.L. (1991) Identification of a

454 NANNO ET AL.

107-kD glycoprotein that mediates adhesion between stromal cells and hematolymphoid cells. J . Exp. Med., 173:373-381.

Kodama, H., Amagai, Y., Koyama, H., and Kasai, S. (1982) A new preadipose cell line derived from new born mouse calvaria can pro- mote the proliferation of pluripotent hemopoietic stem cells in vitro. J. Cell. Physiol., 112:8%95.

Lewinson, D.M., Nagler, A., Ginzton, N., Greenberg, P., and Butcher, E.C. (1990) Hemopoietic progenitor cell expression of the H-CAM (CD44) homing-associated adhesion molecule. Blood, 75r589-595.

Medlock, E.S., and Ham, J.L. (1983) The liver hemopoietic environ- ment: Developing hepatocytes and their role in fetal hemopoiesis. Anat Res., 207:3141.

Metcalf, D., Johnson, G.R., and Mandel, T.E. (1979) Colony formation in agar by multipotential hemopoietic cells. J . Cell. Physiol., 98: 401-420.

Migliaccio, G., Migliaccio, A.R., Valinsky, J . , Langley, K., Zsebo, K., Visser, J.W.M., and Adamson, J.W. (1991) Stem cell factor induces proliferation and differentiation of highly enriched murine he- mopoietic cells. Proc. Natl. Acad. Sci. USA, 88r7420-7424.

Miyake, K., Medina, K.L., Hayashi, S., Ono, S., Hamaoka, T., and Kincade, P.W. (1990) Monoclonal antibodies to Pgp-l/CD44 block lympho-hemopoiesis in long-term bone marrow cultures. J. Exp. Med., 171 :477-488.

Nishikawa, S.I., Ogawa, M., Nishikawa, S., Kunisada, T., and Kodama, H. (1988) B lymphopoiesis on stromal cell clone: Stromal cell clone acting on different stages of B cell differentiation. Eur. J . Immunol., 18:1767-1771.

Rennick, D., Yang, G., Gemmell, L., and Lee, F. (1987) Control of hemopoiesis by a bone marrow stromal cell clone: Lipopolysaccha- ride- and interleukin-1-inducible production of colony-stimulating factors. Blood, 69:682-691.

Rios, M., and Williams, D.A. (1990) Systematic analysis of the ability

of stromal cell lines derived from different murine adult tissues to support maintenance of hemopoietic stem cells in vitro. J. Cell. Physiol., 145:434-443.

Song, Z.X., Shadduck, R.K., Innes, D.J., Waheed, A., and Quesen- berry, P.J. (1985) Hemopoietic factor production by a cell line (TC-1) derived from adherent murine marrow cells. Blood, 66:273-281.

Tsai, S., Emerson, S.G., Sieff, C.A., and Nathan, D.G. (1986) Isolation of a human stromal cell strain secreting hemopoietic growth factors. J. Cell. Physiol., 127:137-145.

Verfaillie, C.M., McCarthy, J.B., and McGlave, P.B. (1991) Differenti- ation of primitive human multipotent hemopoietic progenitors into single lineage clonogenic progenitor is accompanied by alterations in their interaction with fibronectin. J. Exp. Med., 174t69S-703.

Whitlock, C.A., and Witte, O.N. (1982) Long-term culture of B lym- phocytes and their precursors from murine bone marrow. Proc. Natl. Acad. Sci. USA, 79:36083612.

Whitlock, C.A., Tidmarsh, G.F., Miiller-Sieburg, C., and Weissman, I.L. (1987) Bone marrow stromal cell lines with lymphopoietic activ- ity express high levels of a pre-B neoplasia-associated molecule. Cell, 48:1009-1021.

Williams, D.A., Rios, M., Stephens, C., and Patel, V.P. (1991) Fi- bronectin and VLA-4 in haematopoietic stem cell-microenviron- ment interactions. Nature (Lond.), 352:438441.

Zipori, D., Friedman, A., Tamir, M., Silverberg, D., and Malik, Z. (1984) Cultured mouse marrow cell lines: Interactions between fi- broblastoid cells and monocytes. J . Cell. Physiol., 118:143-152.

Zsebo, K.M., Williams, D.A., Geissler, E.N., Broudy, V.D., Martin, F.H., Atkins, H.L., Hsu, R., Birkett, N.C., Okino, K.H., Murdock, D.C., Jacobson, F.W., Langley, K.E., Smith, K.A., Takeishi, T., Cat- tanach, B.M., Galli, S.J., and Suggs, S.V. (1990) Stem cell factor is encoded at the S1 locus of the mouse and is the ligand for the c-kit tyrosine kinase receptor. Cell, 63:213-224.