Stimulation of Mouse and Human Primitive Hematopoiesis by Murine Embryonic Aorta-Gonad-Mesonephros–Derived Stromal Cell Lines

MUCH ATTENTION has been directed to regulatory mechanisms governing the proliferation, self-renewal, and differentiation of primitive hematopoietic stem and progenitor cells. Knowledge of such mechanisms will facilitate ex vivo expansion of human hematopoietic stem cells for transplantation and for gene therapy. Although cytokines that act on primitive hematopoietic cells have been identified,1 attempts to expand transplantable hematopoietic stem cells by defined cytokines have had limited success.2-4 It is widely accepted that the microenvironment plays an important role in hematopoiesis in vivo5 and that stromal cells are principal components of the microenvironment. Indeed, several cloned stromal cell lines can promote the survival, proliferation, and differentiation of hematopoietic cells in vitro.6-9 

In the developing mouse embryo, hematopoietic cells first appear in the yolk sac (YS) at 7.5 days postcoitum (dpc),10 but recent studies have shown that definitive hematopoiesis might originate in an intraembryonic site. Dissections of 8 to 9 dpc embryos have localized multipotent progenitor activity to the para-aortic splanchnopleural mesoderm (P-Sp).11 It has been shown that spleen colony-forming units (CFU-S) appear simultaneously in YS and aorta-gonad-mesonephros (AGM) region at late 9 dpc.12 The number and frequency of CFU-S in the AGM region greatly exceed those in the YS, increase dramatically at late 10 dpc, and then decrease at 11 dpc before a concomitant increase in the fetal liver (FL). It has been also demonstrated that, using conditioned newborn or embryonic mice as hematopoietic transplant recipients, long-term repopulating hematopoietic stem cells (LTR-HSC) can be detected as early as 9 dpc in YS and P-Sp.13 However, when conditioned adult mice are used as the recipients, LTR-HSC are first noted in the AGM region at 10 dpc, before such activity being observed in YS and FL, and expand in 11 dpc AGM region,14 15 suggesting that the AGM region at 10 to 11 dpc provides a microenvironment suitable for the development of LTR-HSC. These observations prompted us to establish stromal cell lines from the AGM region at 10 to 11 dpc, which would support ex vivo expansion of HSC.

We obtained a stromal cell line, AGM-S3, derived from the AGM region of 10.5 dpc mouse embryo. When cocultured with the stromal cells, hematopoietic cells, especially primitive hematopoietic progenitor/stem cells in adult mouse bone marrow and human cord blood, significantly proliferated without additional cytokines. This cell line can now be used to elucidate the molecular mechanisms regulating early hematopoiesis and provide strategies for the manipulation of primitive hematopoietic progenitor/stem cells.

C3H/HeN and C57BL/6 mice, 8 to 10 weeks old, were purchased from Shizuoka Animal Farm (Shizuoka, Japan) and kept under specific pathogen-free conditions. One or two female C3H/HeN mice were caged with a male for 2 hours late in the afternoon and then examined for vaginal plugs. The appearance of the vaginal plug was designated as day 0 of gestation. On day 10.5 of gestation, pregnant mice were anesthetized by ether and then killed by cervical dislocation; from the embryos we dissected the region surrounding aorta, genital ridge, and mesonephros using a dissecting microscope and phosphate-buffered saline (PBS; Nissui, Tokyo, Japan). Eight-week-old male C57BL/6 mice were used for the assay of hematopoietic progenitors and CFU-S.

Mouse bone marrow cells (BMC) were flushed from femurs and tibiae into a-medium (Flow Laboratories, Rockville, MD). Human umbilical cord blood cells (CBC), collected according to institutional guidance, were obtained during normal full-term vaginal deliveries. Mononuclear cells (MNC) were separated by Ficoll-Hypaque density gradient centrifugation after depletion of phagocytes with Silica (IBL, Fujioka, Japan). CD34⁺ cells were purified from MNC by using Dynabeads M-450 CD34 and DETACHaBEAD CD34 (Dynal, Oslo, Norway).16 Eighty-five percent to 95% of the cells separated were CD34 positive by flow cytometric analysis. MS-5 cells were kindly provided by Dr Kazuhiro J. Mori (Niigata University, Niigata, Japan).

Recombinant human (h) and mouse (m) stem cell factor (SCF) and mouse interleukin-3 (IL-3) were kindly provided by Amgen (Thousand Oaks, CA). Recombinant human IL-3, IL-6, granulocyte colony-stimulating factor (G-CSF), erythropoietin (EPO), and thrombopoietin (TPO) were a generous gift from Kirin Brewery Co Ltd (Tokyo, Japan). All the cytokines were pure recombinant molecules and were used at concentrations that induced optimal response in methylcellulose culture of human and mouse hematopoietic cells.

AGM tissues were removed from 10.5 dpc embryos of C3H/HeN mice, dissected to some pieces whose length was approximately 0.3 mm, and cultured in 24-well plates (#143982; Nunc, Naperville, IL) overnight with a drop of α-medium containing 10% fetal bovine serum (FBS; Hyclone Laboratories, Logan, UT) at 37°C in a humidified atmosphere flushed with 5% CO₂ in air, and 1 mL of culture medium was added to the well the next day. The adherent cells appeared around the tissues 1 week later, and the AGM tissues were then removed. After 1 additional week of incubation, the adherent cells were harvested from the well using 0.05% trypsin containing 0.53 mmol EDTA (GIBCO BPL, Grand Island, NY) and were plated in 6-well plates (#152795; Nunc). After 2 weeks of incubation, the cultured cells were irradiated with 900-rad γ-ray to deplete hematopoietic cells. The medium was replaced with fresh culture medium once weekly. Adherent cells were harvested 2 weeks later using trypsin and 50 to 100 cells were seeded into 24-well plates. After 3 weeks of incubation, cells expanded in a well were harvested and used for cell cloning by the limiting dilution technique (0.3 cells/well).17 When proliferating cells were present in each well, they were individually transferred to 25-cm²flasks (#163371; Nunc) containing the culture medium described above. Seventeen cloned cell lines were consequently established. Fourteen of them showed fibroblastoid morphology and had no effect on hematopoiesis, and the remaining 3 cell lines could support hematopoiesis, as described in Results. The three lines were used for the following experiments.

Stromal cells were prepared in 24- or 6-well plates. After 3 to 5 days, mouse and human hematopoietic cells were cocultured with the stromal cells. In stroma-noncontact cultures, human CB CD34⁺ cells were incubated in the upper compartment of transwell inserts placed on top of stromal cell layers prepared in the lower compartment of a 24-well plate (#3421; Costar, Cambridge, MA). The transwell microporous membrane of the insert cultures was a 0.4-μm microporous filter (Costar). Stroma-free cultures were established by seeding cells in the upper compartment of the transwell insert placed in empty wells. The same volume of culture medium was added to wells on day 7, and half the amount of growth medium was exchanged every week from week 2. All the plates were incubated at 37°C in a humidified atmosphere flushed with 5% CO₂ in air. The cells containing stromal cells and expanded cells were harvested using 0.05% trypsin plus 0.53 mmol EDTA after a 1 to 6 weeks of incubation.

Methylcellulose clonal culture was performed using a modification of the technique described previously.18,19 Briefly, 1 mL of culture mixture containing cells, α-medium, 0.9% methylcellulose (Shinetsu Chemical Co, Tokyo, Japan), 30% FBS, 1% deionized fraction V bovine serum albumin (BSA; Sigma, St Louis, MO), 5 × 10⁻⁵ mol/L mercaptoethanol (Eastman Organic Chemicals, Rochester, NY), and combinations of cytokines was plated in 35-mm nontissue culture dishes (#1008; Falcon, Lincoln Park, NJ) and incubated at 37°C in a humidified atmosphere flushed with 5% CO₂ in air. A combination of 100 ng/mL mSCF or hSCF, 20 ng/mL mIL-3 or hIL-3, 100 ng/mL hIL-6, 2 U/mL hEPO, 10 ng/mL hG-CSF, and 4 ng/mL hTPO was used for the determination of the murine and the human clonogenic progenitor cells, respectively. All cultures were performed in triplicate and the number of colony-forming cells (CFU-C) was scored at days 7 to 8 and days 12 to 16 of culture in mouse and human, respectively. Colony types were determined by in situ observation using an inverted microscope and according to the criteria described previously.19-21 The abbreviations used for the clonogenic progenitor cells are as follows: BFU-E, erythroid burst-forming units; CFU-GM, granulocyte-macrophage colony-forming units; CFU- Mk, megakaryocyte colony-forming units; CFU-Mix, mixed colony-forming units.

Cells were injected into C57BL/6 mice exposed to 920 rads (⁶⁰Co) of total irradiation via the tail vein. Eight and 12 days after the injection, the recipients were killed and their spleens were removed and fixed in Bouin’s solution. Macroscopic colonies (day-8- and day-12 CFU-S) were counted under a dissecting microscope.22 

Surface markers of stromal cells were analyzed by flow cytometric analysis using the FACSCalibur (Becton Dickinson, Mountain View, CA), as described.23 The cells were stained with fluorescein isothiocyanate (FITC)-conjugated antimouse CD34 (RAM34; Pharmingen, San Diego, CA), phycoerythrin (PE)-conjugated antimouse Sca-1 (E13-161.7; Pharmingen), and biotin-conjugated antimouse c-Kit (2B8), CD3 (145- 2C11), CD4 (RM4-5), CD8 (53-6.7), B220 (RA3-6B2), Mac-1 (M1/70), Gr-1 (RB6-8C5), TR119 (TER-119) vascular cellular adhesion molecule-1 (VCAM-1) (429), platelet endothelial cellular adhesion molecule-1 (PECAM-1) (MEC13.3), E-selectin (10E9.6), P-selectin (RB40.34), and CD13 (R3-242) purchased from Pharmingen, followed by PE-conjugated streptavidin (Becton Dickinson, San Jose, CA). Positivity or negativity for each antibody was determined based on cells stained with FITC-conjugated rat IgG2a (Cedarlane Laboratories, Ltd, Horndy, Canada), PE-conjugated rat IgG2b (Cedarlane Laboratories, Ltd), or only PE-conjugated streptavidin, as negative controls.

Sorting of murine and human hematopoietic progenitors was performed using described methods.23 Briefly, mouse BM MNC were incubated with PE-conjugated anti-Sca-1, allophycocyanin (APC)-conjugated anti-c-Kit (ACK-2; kindly provided by Dr Shin-Ichi Nishikawa, Kyoto University, Kyoto, Japan), and biotin-conjugated antimouse B220, Mac-1, Gr-1, CD4, CD8, and TR119 monoclonal antibodies, followed by incubation with Texas red (TR)-conjugated streptavidin (Life Technologies, Inc, Rockville, MD). The negative controls were cells stained with PE-conjugated rat IgG2a (Cedarlane Laboratories, Ltd), APC-conjugated rat IgG2b (Pharmingen), or only TR-conjugated streptavidin. Based on these controls, Lin⁻c-Kit⁺Sca-1^+/− cells were sorted with a FACSVantage (Becton Dickinson). Human CD34⁺CD38^+/− cells were sorted from CB MNC stained with FITC-conjugated antihuman CD34 (My10; Becton Dickinson) and PE-conjugated antihuman CD38 (HB-7; Becton Dickinson), based on cells stained with FITC- and PE-conjugated mouse IgG1 (Becton Dickinson) as the negative control.

Total RNA was prepared from AGM-S3 and MS-5 cells using QIAGEN RNeasy kit (Amersham, Uppsala, Sweden), incubated with deoxyribonuclease I, and reverse-transcribed using the first-strand cDNA synthesis kit (Pharmacia).

Cytokine-specific cDNAs were amplified with Taq DNA polymerase, using pairs of oligonucleotide primers as follows^24-26: IL-3-5′, TCAGACTTTAGGTGCTCTGC-3′; IL-3-3′, TCGTGGAAAGCCAAGGAGAA-3′; IL-6-5′, GATAGTCAATTCCAGAAACCGCTA-3′; IL-6-3′, TACTCCAGGTAGCTATGGTACTCC-3′; oncostatin M (OSM)-5′, TCCGCCTCCAAAACCTGAACAC-3′; OSM-3′, TCTGTGTGGGCTCAGGTATCT-3′; leukemia inhibitory factor (LIF)-5′, GGAGTCCAGCCCATAATGAAGGTC-3′; LIF-3′, GGCCTGGACCACCACACTTATGAC-3′; G-CSF-5′, GACGGCTCGCCTTGCTCTGCACCA-3′; G-CSF-3′, ACCTGGCTGCCACTGTTTCTT-TAGG-3′; granulocyte-macrophage colony-stimulating factor (GM-CSF)-5′, AGAAGCTAACATGTGTGCAGACCCG-3′; GM-CSF-3′, ATTCCAAGTTCCTGGCTCATTACGC-3′; macrophage colony-stimulating factor (M-CSF)-5′, GACAGGCCGTTGACAGAGGTGAACCC-3′; M-CSF-3′, ATAGAATCCTTTCT-ATACTGGCAGTTC-3′; SCF-5′, AAAGTAAAACTCGAGATGAAGAAGACACAAACTTGG-3′; SCF-3′, TTTGACTTTTTAATTAATTAGGCTGCAACAGGGGGTAACAT-3′; Flk2 ligand (FL)-5′, AAAGAAAAACTCGAGATGACAGTGCAGGCGCCAGCC-3′; FL-3′, TTTGACTTTTTAATTAATTACTGCCT GGGCCGAGGCTCTGG-3′; TPO-5′, CAGGACCACAGCTCACAAGGAC-3′; TPO-3′, CCATTCACAGGTCCGTGTGTCC-3′; IL-11-5′, CTCTGGCCAGATAGAGTCGTTG-3′; IL-11-3′, CACGGCGCAGCCATTGTACATG-3′; HPRT-5′, CCTGCTGGATTACATTAAAG CACTG-3′; HPRT-3′, GTCAAGGGCATATCCAACAACAAAC-3′. Samples were denatured at 94°C for 3 minutes, followed by amplification rounds consisting of 94°C for 1 minute (denaturing); 55°C to 65°C for 2 minutes (annealing), and 72°C for 3 minutes (extension) for 40 cycles. Products were separated on a 1.0% agarose gel, stained with ethidium bromide, and photographed.

CB MNC (1 × 10⁶) and those cultured with stromal cells for 4 weeks were injected into 8- to 10-week-old NOD/Shi-scid (NOD/SCID) mice irradiated with 300 rads (⁶⁰Co) of total irradiation through the tail vein. Because natural killer cell activity of NOD/Shi-scid mice is not so low,27 the recipient mice were injected intraperitoneally with 300 mL of PBS containing 20 mL of anti-asialo GM1 antibody (Wako, Osaka, Japan) immediately before the cell transplantation and on day 11 of transplantation. Mice were killed 5 weeks after the transplantation, and BMC were collected. The presence of human hematopoietic cells was determined by detection of cells positively stained with FITC-conjugated antihuman CD45 in flow cytometric analysis. Specific subsets of human hematopoietic cells were quantified by gating on human CD45-PE-cyanine 5-succinimidylester (PECy5)-positive cells and then assessing staining with antihuman CD13-PE, CD33-PE, CD14-PE, CD10-FITC, CD19-PE, CD3-FITC, and CD34-FITC. All antibodies were from Becton Dickinson except for antihuman CD34-FITC (Immunoteck, Marseille, France). Human hematopoietic cells were also determined by detection human ALU sequence in DNA of the BMC by PCR analysis. For PCR analysis, DNA extracted from the BMC was subjected to PCR amplification using as primers the following human ALU²⁸ and mouse β-actin29 sequences: ALU-5′, CACCTGTAATCCCAGCAGTTT-3′; ALU-3′, CGCGATCTCGGCTCACTGCA-3′; β-actin-5′, GTGGGCCGCTCTAGGCACCAA-3′, β-actin-3′, CTCTTTGATGTCACGCACGATTTC-3′.

Three stromal cell lines, AGM-S1, 2, and 3, were obtained from the 10.5 dpc mouse embryo AGM region. They were maintained in α-medium containing 10% FBS in 25-cm² flasks at 37°C in a humidified atmosphere flushed with 5% CO₂ in air. These conditions led to doubling times of 5 to 7 days and 100% viability. The photomicograph (Fig 1A) shows the appearance of one of these, AGM-S3. AGM-S3 cells had large, flat cell morphology with widely spread cytoplasm, and some of them converted to fat-containing cells in confluent culture. AGM-S1 and AGM-S2 cells manifested a similar morphology to AGM-S3 cells.

We first examined the effect of AGM-S1, 2, and 3 stromal cells on adult hematopoietic cells in the coculture with BMC of C57BL/6 mice. When 1 × 10⁵ BMC were plated on the stromal cell layers, cobblestone-like colonies appeared under the stromal layers at day 3 of incubation and then gradually increased. The colonies in the cocultures with AGM-S1 and 3 cells were almostly twice the number of those with AGM-S2 cells at day 7 (Table 1). Clonal assay of the cells in the coculture at day 7 of incubation showed that the three cell lines supported the expansion of CFU-C. However, the numbers of CFU-C in the cocultures with AGM-S1 and 3 cells were approximately 5 to 7 times larger than that with AGM-S2 cells. Day-8 and day-12 CFU-S also expanded at day 7 of the coculture with AGM-S1 and 3 cells. This result indicates that AGM-S1, 2, and 3 cells are able to promote the proliferation of hematopoietic progenitors without any exogenous cytokines, but AGM-S2 cells seem to have less ability on murine hematopoietic progenitors than do AGM-S1 and 3 cells.

To investigate the effect of AGM-S3 cells on murine hematopoietic progenitors in more detail, sorted Lin⁻c-Kit⁺Sca-1⁺ and Lin⁻c-Kit⁺Sca-1⁻ cells shown to contain murine primitive hematopoietic cells and more mature ones, respectively,30 were cocultured with AGM-S3 cells, which were considered to be the most potent stimulator on murine hematopoietic progenitors among the three cell lines (Table 2). When 100 sorted Lin⁻c-Kit⁺Sca-1⁻ cells were cocultured with AGM-S3 cells, no hematopoietic progenitors were detected in the coculture at day 10 of culture. By contrast, in the culture of 100 Lin⁻c-Kit⁺Sca-1⁺cells, all types of progenitors, including CFU-GM, BFU-E, CFU-Mk, CFU-Mix, and day-8 and day-12 CFU-S, showed a remarked expansion. In particular, CFU-Mix and day-12 CFU-S, immature hematopoietic progenitors, increased 80-fold and 10-fold, respectively, and remained detectable after 6 weeks of coculture of AGM-S3 cells with Lin⁻c-Kit⁺Sca-1⁺ cells (data not shown). Thus, most of hematopoietic progenitors expanded in the coculture with AGM-S3 cells were derived from Lin⁻c-Kit⁺Sca-1⁺ cells, indicating that AGM-S3 cells act on primitive hematopoietic cells of the adult mouse.

Some murine stromal cell lines have been shown to support human hematopoiesis.31 For example, MS-5 cells,17 a stromal cell line esablished from adult mouse BMC, can stimulate proliferation of human hematopoietic progenitors.31 We then cocultured human CB CD34⁺ cells with AGM-S1, 2, and 3 and MS-5 cells to examine the effect of these stromal cells on human hematopoietic progenitors. Five hundred human CB CD34⁺cells were cocultured with AGM-S1, 2, and 3 and MS-5 cells prepared in 24-well plates. At days 7 to 10 of incubation, cobblestone-like colonies appeared under these stromal cells (Fig 1B), and nonadherent hematopoietic cells continued to be released into the culture medium after day 14. Clonal assay of the cells harvested from the coculture at week 3 of incubation showed that all the cell lines stimulated the expansion of human CB hematopoietic progenitors, but AGM-S1 and 3 cells again showed more significant stimulatory activity than AGM-S2, and MS-5 cells showed similar effects to AGM-S2 cells (Table 3). The number of total progenitors on AGM-S1, 2, and 3 and MS-5 cells increased 5.3-fold, 3-fold, 4.8-fold, and 2.3-fold, respectively, at week 3 of culture. Although clonogenic progenitors were detectable after 6 weeks of incubation with these stromal cells, the numbers of progenitors in coculture with AGM-S1 and 3 cells were approximately 5 times that with AGM-S2 and MS-5 cells. Interestingly, all of the CFU-GM, BFU-E, and CFU-Mix were present in cocultures with AGM-S1 and 3 cells, whereas only CFU-GM were detected in coculture with AGM-S2 and MS-5 cells at 6 weeks of culture. This result indicates that AGM-S1 and 3 cells are more potent stimulators than AGM-S2 and MS-5 cells on human CB hematopoietic cells.

To examine whether AGM-S3 cells act on human primitive hematopoietic progenitor cells, we compared the effects of AGM-S3 cells on CD34⁺CD38⁻ and CD34⁺CD38⁺ cells, which have been shown to reflect human primitive hematopoietic cells and more mature populations, respectively32(Table 4). When both fractions sorted from CB MNC were cocultured with AGM-S3 cells, hematopoietic progenitors generated from CD34⁺CD38⁺ cells exceeded those from CD34⁺CD38⁻ cells at week 4 of incubation. However, at week 6, numbers of progenitors from CD34⁺CD38⁻ cells surpassed those from CD34⁺CD38⁺ cells, and neither BFU-E nor CFU-Mix were detectable in the culture of CD34⁺CD38⁺cells, whereas progenitors from CD34⁺CD38⁻cells contained 7% of BFU-E and 20% of CFU-Mix (Fig 1C). The result indicates that AGM-S3 cells act on both CD34⁺CD38⁺ and CD34⁺CD38⁻ cells and that the generation of CFU-Mix, immature multipotential progenitors, from CD34⁺CD38⁻ primitive hematopoietic cells was supported by AGM-S3 cells at least for 6 weeks.

To investigate the effect of AGM-S3 cells on human reconstituting hematopoietic stem cells, we performed transplantation experiments using NOD/SCID mice and examined the reconstituting abilities of CB MNC before and after the coculture with AGM-S3 cells. When 1 × 10⁶ CB MNC cocultured with AGM-S3 cells for 4 weeks were transplanted into NOD/SCID mice, human CD45⁺ cells were found in BMC of the recipient mice, as determined by flow cytometry (Fig 2A) after 5 weeks of transplantation. Chimerism percentages of human CD45⁺ cells in BMC were 6.2% and 5.2% in the mice transplanted with CB MNC before and after the coculture, respectively. Human CD45⁺ cells in BMC of the mouse transplanted with cocultured CB MNC were further analyzed to determine multilineage reconstitution. In human CD45⁺cells, 29% of CD13⁺ cells, 38% of CD33⁺cells, 12% of CD14⁺ cells, 58% of CD19⁺cells, and 8% of CD34⁺ cells, but no CD3⁺cells, were detected (Fig 2B and data not shown). The presence of human hematopoietic cells in the recipient BMC was also confirmed by the detection of human ALU sequences in PCR analysis of the BMC DNA (data not shown). Hence, AGM-S3 cells can also support survival or self-renewal of human reconstituting stem cells at least for 4 weeks.

We investigated the characteristics of AGM-S3 cells using flow cytometric analysis. Our result (Fig 3 and data not shown) showed that hematopoietic lineage markers such as CD3, CD4, CD8, B220, Mac-1, Gr-1, and TR119 were undetectable on the surface of AGM-S3 cells. Adhesion molecules, VCAM-1 but no PECAM-1, E-selectin, and P-selectin were expressed on AGM-S3 cells. AGM-S3 cells also expressed CD13 and Sca-1, but not c-Kit and CD34.

We then compared cytokine expression between AGM-S3 cells and MS-5 cells by RT-PCR (Fig 4). Both AGM-S3 and MS-5 cells showed detectable levels of SCF, IL-6, and OSM, but undetectable levels of IL-3, LIF, G-CSF, GM-CSF, IL-11, TPO, and FL. MS-5 cells, but not AGM-S3 cells, showed a detectable level of M-CSF.

To examine the effect of conditioned medium of AGM-S3 cells, human CB CD34⁺ cells were plated either in stroma-contact cultures with AGM-S3 cells or in transwell inserts placed above the stromal cell layers (stroma-noncontact cultures; Table5). CD34⁺ cells were also cultured in transwell inserts placed in empty wells as controls (stroma-free cultures), where only few or no clonogenic cells were detected at weeks 2 and 4 of incubation. The number of total clonogenic progenitors decreased at both weeks 2 and 4 in stroma-noncontact cultures, whereas it increased 4-fold and 3.2-fold in stroma-contact cultures, respectively. The number of progenitors in stroma- contact cultures were more than 7 times than that in stroma-noncontact cultures at week 4. Moreover, all of CFU-GM, BFU-E, and CFU-Mix were present in stroma-contact cultures, whereas only CFU-GM were detected in stroma-noncontact cultures. These results indicates that the direct interaction between AGM-S3 and human hematopoietic cells is involved in the stimulation of human hematopoietic cells, especially primitive hematopoietic cells.

There is abundant evidence to indicate that microenvironmental stroma within the hematopoietic tissues play important roles in the support and regulation of hematopoiesis.5,33 Stroma is composed of heterogeneous cell populations and an extracellular matrix (ECM) that constitute a suitable microenvironment for the proliferation and differentiation of hematopoietic stem/progenotor cells.34Not only the cytokines locally produced by stroml cells,35but also cell-to-cell36 and cell-to-extracellular matrix contacts37,38 are known to be involved in the maintenance of hematopoiesis. To make the mechanisms clear, several stromal cell lines have been established from fetal and adult mouse hematopoietic tissues, such as MS-5 from adult BM,17 PA-6 from newborn calvaria,6 SPY3-2 from adult spleen,7CD34⁺ endothelial cells from yolk sac,8 and AFT024 cells from fetal liver,9 some of which can support the long-term survival of murine LTR-HSC.9 

Recent evidence shows that the microenvironment of AGM region plays important roles in the self-renewal and expansion of hematopoietic stem cells.14 Hence, we established three stromal cell lines, AGM-S1, 2, and 3, from 10.5 dpc mouse embryo AGM region. Among these three cell lines, which were similar in apperance, AGM-S1 and 3 cells showed more significant stimulatory activity to proliferate murine and human primitive hematopoietic progenitors than AGM-S2 cells. The expression of CD13 and VCAM-1 on AGM-S3 cells may indicate the stromal cells possess the character of endothelial cells to some extent. AGM-S3 also expressed Sca-1, which was reported to be detected in 10 dpc mouse embryo AGM region39 and endothelial cells.40Recently, Satoh et al41 reported that the expression level of Sca-1 in PA-6 cells correlate with their hematopoietic supporting activity.

The activity of the stromal cells on murine hematopoiesis was confirmed by the expansion of CFU-C and day-8- and day-12 CFU-S in the coculture wih murine BMC. The coculture of AGM-S3 cells with sorted hematopoietic progenitor cells showed that these expanded progenitors were derived from Lin⁻c-Kit⁺Sca-1⁺ cells, indicating that AGM-S3 cells act on primitive hematopoietic progenitors.

Some evidence has accumulated showing that human hematopoiesis can develop in a nonhuman environment.31,42,43 For example, MS-5 cells can support human hematopoiesis successfully, as shown by the present and previous studies. Issaad et al31 reported that MS-5 cells supported the generation of various hematopoietic progenitors, including CFU-Mix from human BM CD34⁺CD38⁻ cells in long-term culture. In contract, Nishi et al42 showed that MS-5 cells were able to support the generation of CFU-GM, but not CFU-Mix from CB CD34⁺CD38⁻ cells, in accordance with our studies. Our data presented here demonstrated that AGM-S3 cells also promoted the expansion of human CB hematopoietic progenitor cells without the addition of human growth factors. When CB CD34⁺CD38⁻ cells were cocultured with AGM-S3 and MS-5 cells, only AGM-S3 cells supported the generation of CFU-Mix, at least for 6 weeks. Moreover, we performed experiments using NOD-SCID mice to examine the effect of AGM-S3 cells on human hematopoietic stem cells. The result indicated that AGM-S3 cells had the ability to maintain the survival or self-renewal of human reconstituting hematopoietic stem cells for more than 4 weeks. These results suggest that AGM-S3 cells have more potent stimulatory activity in human primitive hematopoiesis than MS-5 cells.

Little is known on how murine stromal cells stimulate human hematopoietic cells, but the involvement of growth factor(s) with species-cross activity produced by stromal cells is most likely. In RT-PCR analysis, AGM-S3 cells produced detectable levels of SCF, IL-6, and OSM, but no detectable levels of M-CSF, IL-3, LIF, G-CSF, GM-CSF, IL-11, TPO, and FL. Our data and previous reports44 showed that the cytokine expression of AGM-S3 cells was similar to that of MS-5 cells, except for the expression of M-CSF. It is of interest that M-CSF is expressed on MS-5 but not on AGM-S3 cells. OP9 stromal cells lacking M-CSF were found to support murine primitive hematopoietic progenitors45 and embryonic stem cells.46Because the combination of SCF and IL-6 stimulates the proliferation of murine primitive hematopoietic progenitors47 and OSM has been shown to induce hematopoiesis in the 11.5 dpc mouse AGM region,48 these cytokines may be involved in expansion of the murine primitive hematopoietic progenitors observed in the present study. However, mouse IL-6 and OSM have no apparent effects on human cells.49 Although mouse SCF has cross-activity with human SCF, mouse SCF alone or in combination with mouse IL-6 and OSM could not support human clonogenic progenitors for 6 weeks (data not shown). Therefore, AGM-S3 cells seem to express species-cross reactive molecule(s) other than the cytokines we examined and that act on primitive hematopoietic cells.

Because the production of clonogenic progenitors in stroma-noncontact cultures were higher than that in stroma-free cultures, soluble form factor(s) was produced by AGM-S3 cells. However, the direct contact between AGM-S3 and CD34⁺ cells was required for the optimal stimulation of clonogenic progenitors, and the generation of CFU- Mix and BFU-E was detected only in stroma-contact cultures. Although the effect seen in this experiment may have been limited by factor concentration, stability, or dependence on ECM attachment, the result suggests a possibility that membrane-bound molecule(s) plays some roles in the expansion of primitive hematopoietic cells by AGM-S3 cells. Detection and characterization of the molecule(s) from AGM-S3 cells may contribute toward understanding of regulatory mechanisms in the development of primitive hematopoietic stem and progenitor cells and provide a novel strategy for ex vivo expansion of human transplantable HSC.

The authors thank M. Nishihara and A. Kaneko for technical support, Drs K. Kobayashi and Y. Ueyama for collaboration, and M. Ohara for comments on the manuscript.