Effects of myeloproliferative sarcoma virus on the pluripotential stem cell and granulocyte precursor cell populations of DBA/2 mice.

The hematopoietic stem cell (CFU-S) and granulocyte precursor cell (CFU-C) populations have been assayed in the spleen, blood, and bone marrow of DBA/2 mice at various times after infection with the myeloproliferative sarcoma virus (MPSV). Beginning between 7 and 19 days after virus infection, the number of CFU-S showed a steady, parallel increase in the blood and spleen, reaching a maximum at both sites by days 25-30. At the maximum, in the spleen the concentration of CFU-S was 10 times greater than that in the blood, and the total number of CFU-S was over 100 times greater than that of normal animals. During the same period, in the bone marrow the number of CFU-S decreased to one-half of normal. Nevertheless, the CFU-S from MPSV-infected animals differentiated normally in the spleens of irradiated, normal recipient mice (except for some hyperplasia of the erythroid component of spleen colonies). The CFU-C content of the bone marrow, spleen, and blood paralleled the CFU-S content of these organs: The CFU-S and CFU-C populations changed almost synchronously after MPSV infection. In the terminal stage of the MPSV-induced disease, a variable proportion of the CFU-C population acquired the ability to differentiate in the absence of added colony-stimulating factor.

Hematological toxicity of repeated injections of (chloro-2-ethyl)-ribofuranosyl-3-nitrosourea.

A previous study using a single injection of (chloro-2-ethyl)ribofuranosyl-3-nitrosourea has indicated the low acute hematotoxicity of this nitrosourea. However, because the hematotoxicity of nitrosourea is usually cumulative, we have studied the effect of injecting (chloro-2-ethyl)ribofuranosyl-3-nitrosourea (15 mg/kg) dissolved in 0.2 ml sterile oil C57BL X i.p. into DBA/2F1 mice for 5 consecutive weeks. The dose per injection represents the minimal dose necessary to show the maximal therapeutic efficacy on L1210 leukemia. Bone marrow cellularity and histology, spleen weight, bone marrow and splenic pluripotent stem cells, and colony-forming units committed to granulocyte-macrophage differentiation were measured 1, 2, 4, 7, and 14 days after the last injection in treated and control mice receiving oil only. No morphological or histological changes were found in the spleen and bone marrow of treated mice. In both organs, the number of splenic pluripotent hematopoietic stem cells decreased by about 1 log 1 day after the last injection but rapidly returned to normal values on Days 4 to 14. Granulocyte-macrophage-committed precursors were affected in both organs, at 20% of control values for bone marrow and 5% of control values for spleen on Day 1, with a transient and partial recovery on Day 4 followed by a second drop to 20 and 25% on Day 14. This effect on granulocyte-macrophage precursors contrasts with the absence of significant effect when the same treatment is used on peripheral white blood cell counts. Our results demonstrate that (chloro-2-ethyl)ribofuranosyl-3-nitrosourea belongs to the class of new nitrosoureas with low cumulative hematotoxicity.

Cascade regulation of cytokine production in granulopoiesis.

Injection of IL-3 producing T cells (STIL-3) resulted in a granulocytosis in the syngeneic mice. A high IL-3 activity was detected in the culture supernatant of the spleen cells of these mice, but only a low activity was detectable in the bone marrow cell-conditioned medium. There was no significant difference in the distribution of the STIL-3 cells between the spleen and the bone marrow of the mice injected with the STIL-3 cells. Two possibilities have been envisaged from these observations; i) IL-3 induces production of granulocyte stimulating cytokines (CSFs) from hemopoietic cells hence resulting in the granulocytosis in the STIL-3 injected mice, ii) an inhibitor of IL-3 is produced in response to an excess stimuli with IL-3 in the bone marrow. We found that IL-3 induced a production of IL-6, G- and GM-CSF from bone marrow cells. In contrast, stimulation of the bone marrow cells with an excess level of IL-3 resulted in a production of a heat-labile activity (NIL-3) antagonistic to IL-3. Furthermore, stimulation of bone marrow cells with IL-6, G- or GM-CSF did not induce the production of IL-3, indicating a hierarchical regulation of the cytokine production. These observations have provided positive evidences to the above mentioned 2 possibilities, and indicate the existence of a positive feedback mechanism in the IL-3-induced granulocytosis, as well as the presence of a negative feedback mechanism for the homeostatic regulation of the granulopoiesis.

Characterization of murine hemopoietic-supportive (MS-1 and MS-5) and non-supportive (MS-K) cell lines.

Characteristics of hemopoietic-supportive (MS-1 and MS-5) and non-supportive (MS-K) cell lines were compared. Supportive cells adhered to hemopoietic stem cells and produced granulocyte-macrophage colony-stimulating factor (GM-CSF), whereas non-supportive cells did not adhere to hemopoietic cells and only produced macrophage colony-stimulating factor. Both cell lines produced substantial levels of IL-6 and steel factor (SLF) which is reportedly a stem-cell factor. Northern blot analysis revealed that SLF but neither c-kit nor interleukin 3 (IL-3) mRNA was detectable in these cell lines, although IL-3-like activity was found in the supernatant of MS-5 cell culture. These observations suggest that the hemopoietic-supportive function of stromal cells may reside in adherence of stem cells, and production of GM-CSF probably in combination with SLF. SLF may be transferred from stromal cells directly to stem cells through adhesion of stem cells to supportive stromal cells.

CD34high+ CD38(low/-) cells generated in a xenogenic coculture system are capable of both long-term hematopoiesis and multiple differentiation.

CD34+ cells isolated from human umbilical cord blood (HUCB) are thought to have potential in clinical applications such as transplantation and gene therapy. Recently, we developed a xenogenic coculture system involving HUCB-CD34+ cells and murine bone marrow stromal cells, HESS-5 cells, in combination with human interleukin-3 and stem cell factor. Under these xenogenic coculture conditions, the numbers of CD34high+ cells and primitive progenitor cells, such as CD34high+ CD38(low/-) cells and high proliferative potential colony-forming cells (HPP-CFCs), increased dramatically by a factor of 102.1, 66.5 and 104.9, respectively. In the present study, we used a secondary culture of B progenitor cells and long-term culture (LTC)-initiating cells to characterize and compare the progenitor capability of re-isolated CD34high+ CD38(low/-) cells, which have been identified as one of the most primitive progenitor cells, with that of freshly isolated CD34high+ CD38(low/-) cells. Compared with freshly isolated CD34high+ CD38(low/-) cells, the re-isolated CD34high+ CD38(low/-) cells were equally as capable of proliferating and differentiating into myeloid and B progenitor cells. No significant differences were observed in the frequency of LTC-initiating cells in the re-isolated CD34high+ CD38(low/-) cells compared with that in freshly isolated CD34high+ CD38(low/-) cells. Furthermore, the re-isolated CD34high+ CD38(low/-) cells were capable of long-term reconstitution and multiple differentiation in non-obese diabetic mice with severe combined immunodeficiency disease (NOD/SCID mice). The results demonstrate that this xenogenic coculture system can be used for successful in vitro expansion of HUCB-progenitor cells that possess the capability for both long-term hematopoiesis as well as multipotent differentiation into myeloid and lymphoid cells both in vivo and in vitro.

The role of soluble growth factors in inducing transient growth and clonal extinction of stroma cell dependent erythroblastic leukemia cells.

A coculture system of a murine erythroblastic leukemia cell line (ELM-D) with its supportive stromal cell line (MS-5) was established. Long-term growth of ELM-D cells is strictly stroma cell dependent. Interaction between stem cell factor (SCF) and its receptor, c-kit, was demonstrated to be important for stroma cell-dependent growth by anti c-kit neutralizing monoclonal antibody (mAb) inhibition experiments. Significantly, soluble growth factors such as granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin-3 (IL-3) or SCF of MS-5 stromal cells (MS-5 CM) could replace the requirement of stroma cells for a considerable period. However, ELM-D cells maintained in these growth factors underwent clonal extinction after 3-6 weeks unless contact with stroma was re-established. Furthermore, IL-3 or GM-CSF acted in a dominant manner in inducing cell death in the presence of stroma cells. Cells showing clonal extinction undergo programmed cell death and do not differentiate. These altered growth properties of ELM-D cells exposed to soluble growth factors or to stroma cells appear to be analogous to those described for T or B cells primed by antigen presenting cells and then grown in growth factors.

In vitro long-term culture of human primitive hematopoietic cells supported by murine stromal cell line MS-5.

When Lin-CD34+CD38- cells from normal human cord blood were cocultured with MS-5, colony forming cells were maintained for over 8 weeks. Prevention of contact between MS-5 and Lin-CD34+CD38- cells by using a membrane filter was negligible for this activity, indicating that the activity of MS-5 on human primitive hematopoietic cells may be due to soluble factor(s) secreted from MS-5. We tried to purify this activity by a [3H]TdR incorporation assay. The activity was found in 150 kD fraction and was neutralized with anti-mSCF (stem cell factor) antibody. Another 20-30 kD fraction synergized with mSCF to stimulate the growth of Lin-CD34+CD38- cells but failed alone. This fraction supported the growth of the G-CSF (granulocyte-colony stimulating factor)-dependent cell line FD/GR3, FDC-P2 transfected with mG-CSF receptor cDNA. This synergy was canceled in the presence of soluble mG-CSF receptor. Addition of anti-mSCF antibody and soluble mG-CSF receptor to the culture completely abrogated the activity of MS-5-culture supernatant. These results indicate the activity of MS-5 on Lin-CD34+CD38- cells is due to synergistic effect of mSCF and mG-CSF.

Culture system for extensive production of CD19+IgM+ cells by human cord blood CD34+ progenitors.

We established a co-culture system with a monolayer of the murine bone marrow (BM) stroma cell line, MS-5, in which human cord blood CD34+ cells differentiated to CD19+ cells. The addition of stem cell factor (SCF) and granulocyte colony-stimulating factor (G-CSF) highly enhanced the production of CD19+ cells. The expansion of the cell numbers was over 10(3)-fold. Furthermore, a significant proportion (<45%) of the cells expressed surface IgM (sIgM) after 5 weeks of co-culture. CD34+CD19- cells also showed a similar development of CD19+ cells and CD19+sigM+ cells. Filter separation of MS-5 cells and CD34+ cells did not inhibit the growth of CD19+ cells. However, when further purified CD34+CD19-CD13- CD33- cells were cultured in the presence of MS-5 cells with or without a separation filter, CD19+ cells did not appear in the non-contact setting. This result suggested that the highly purified CD34+CD19-CD13-CD33- progenitors require the cell-cell contact for the development of CD19+ cells, whereas other CD34+ fractions contain progenitors that do not require the contact. This co-culture system should be useful for the study of early human B-lymphopoiesis.

Human cord blood-derived primitive progenitors are enriched in CD34+c-kit- cells: correlation between long-term culture-initiating cells and telomerase expression.

We studied the functional characteristics of subpopulations of cord blood-derived CD34+ cells expressing different levels of CD38 and c-kit antigens, using clonal cell culture and long-term culture with allogeneic bone marrow stromal cells or the MS-5 murine stromal cell line to assay long-term culture-initiating cells (LTC-IC) in each subpopulation. To investigate the capacity for replication, proliferation, and differentiation of each subpopulation of CD34+ cells, we also studied the correlation between LTC-IC and telomerase activity. After 5 weeks of coculture, LTC-IC accounted for one out of 32 CD34+CD38- cells and one out of 33 CD34+c-kit- cells. In contrast, the frequency of LTC-IC was low in their antigen-positive counterparts (one per 84 CD34+CD38+ cells, one per 90 CD34+c-kit(low) cells, and very low among CD34+c-kit(high) cells). It was noteworthy that some LTC-IC derived from CD34+CD38- as well as CD34+c-kit- cells generated colony-forming cells (CFCs) after up to 9 weeks of coculture. Telomerase activity was consistently low in CD34+CD38- and CD34+c-kit- cells compared to CD38+ or c-kit(high or low) cells, suggesting that CD34+CD38- or c-kit- cells are likely to be more quiescent. These results suggest that the CD34+CD38- and CD34+c-kit- cell populations are primitive stem/progenitor cells, and that the telomerase activity of these cells correlates with their proliferative capacity as well as their stage of differentiation.

Distribution and differentiation of murine mast cell progenitors determined with soft agar culture.

We have examined the distribution and differentiation of mast cell progenitors (mast-CFC) using a sensitive semisolid agar culture stimulated with STIL-3 (leukemic T cell line)-conditioned medium as a source of interleukin 3. The number of mast-CFC in the bone marrow of normal (+/+) mice was much higher than in previously reported data, although the number was almost the same in the spleen and peripheral blood as in previously reported data. Although mast-CFC were detectable in genetically anemic W/Wv mice as well, the concentration in the bone marrow was significantly lower than that of the +/+ mice. In the bone marrow there were more immature mast-CFC forming large-sized colonies (greater than 500 cells), whereas these were very few in the spleen and peripheral blood. Most mast-CFC were in the resting state. We conclude that mast-CFC differentiate to some extent in the bone marrow and then migrate in the peripheral blood to some organs where they proliferate and differentiate.

Characterization of possible receptor of colony-promoting activity (CPA) by comparison with that of colony-stimulating factor (CSF).

Supernatants of long-term mouse bone marrow cell cultures contain colony-promoting activity (CPA). CPA itself does not stimulate colony formation of granulocyte-macrophage progenitor cells (GM-CFC) when added to a semisolid agar culture of bone marrow cells, but augments colony formation in the presence of colony-stimulating factor (CSF). The CPA-responsive cells are postulated to be pre-GM-CFC, a cell compartment younger than GM-CFC. In absorption experiments, CSF-coated bone marrow cells failed to absorb CPA, whereas normal bone marrow cells absorbed the activity. Cholera toxin (CT) is known to inhibit GM-CFC colony formation of bone marrow cells, probably by binding to GM-CFC-receptors [8, 9]. In the present studies, preincubation of the cells with CPA, before exposure to CT, reduced the suppressive effect of CT on GM-CFC colony formation. CT also inhibited pre-GM-CFC colony formation. Such suppressive effects, however, were reduced by preincubation of the cells not only with CPA but also with CSF. These results suggest that CSF and CPA might share the same receptors. It may also suggest that CT does not bind to specific sites. A difference in the susceptibility of GM-CFC and pre-GM-CFC to suppression by CT was also observed. Incubation of the cells with a low concentration of CT resulted in the substantial decrease of the number of GM-CFC, whereas the number of pre-GM-CFC remained high. Therefore, it seems that CPA shares some but not all of the CSF receptors.

Age-related changes in various hemopoietic progenitor cells in senescence-accelerated (SAM-P) mice.

The effect of aging process on the hemopoietic system in senescence-accelerated (SAM-P) mice with respect to numbers of hemopoietic progenitor cells was investigated. The numbers of femoral granulocyte-macrophage colony-forming cells (CFU-GM), mast cell progenitors (mast colony-forming units, CFU-Mast), erythroid burst-forming units (BFU-E), and erythroid colony-forming units (CFU-E) in old mice (30-35 weeks old) decreased to 96%, 81%, 83%, and 87% of those of young mice (8-12 weeks old), respectively. The numbers of femoral fibroblast colony-forming cells (CFU-F) in old mice increased to 315% of those of young mice. The numbers of splenic CFU-GM, CFU-Mast, BFU-E, and CFU-E in old mice decreased to 7%, 43%, 25%, and 40% of those of young mice, respectively. In contrast, significant changes in these progenitor cells were not observed in the bone marrow. These findings suggest that the effect of the aging process on hemopoietic tissues in SAM-P mice is predominantly in the spleen.

Biological activities of tetrapeptide AcSDKP on hemopoietic cell binding to the stromal cell in vitro.

The biological activity of the tetrapeptide acetyl-N-Ser-Asp-Lys-Pro (AcSDKP) on hemopoietic cell binding to the stromal cells was studied by using a rosette formation technique that quantitatively represents the specific hemopoietic cell binding to the stroma. Marrow hemopoietic cell binding to the stroma was enhanced by AcSDKP. This enhancement was completely neutralized by addition of anti-AcSDKP polyclonal antibody. Furthermore, preincubation of stromal cells with AcSDKP increased hemopoietic cell binding to the stroma, whereas preincubation of hemopoietic cells with AcSDKP showed no increment of the binding. These findings suggest that AcSDKP enhanced hemopoietic cell binding through the activation of stromal cells.

Effects of androgen on transient endogenous spleen colonies and other hemopoietic stem cells in mice.

To clarify the effects of androgen on hemopoiesis in mice, we investigated changes in hemopoietic stem cells at various stages of differentiation after injection of 19-nandrolone decanoate (19-ND). 19-ND induced (a) a marked increase in the population of CFUe, especially in the spleen, and (b) a less significant increase in the number of CFUs, GM-CFC, and BFUe. The number of endogenous spleen colonies developed transiently on day 4 after irradiation (TE-CFU) increased significantly, and the initiation and induction of erythropoietic maturation was enhanced. Growth of the pluripotent stem cells (CFUs) assessed by endogenous spleen colony formation was influenced more than was the population size. Production of erythropoietin (Ep) in mice was not affected by the treatment with 19-ND. These results indicate that 19-ND affects mature precursors in erythroid cell lineage rather than pluripotent stem cells and that the effect is unlikely to be induced through an increase in the production of Ep.

Quantitative development of adherent cell colonies in bone marrow cell culture in vitro.

Quantification of the formation of adherent cell colonies in bone marrow cell culture was attempted. By secondary transfer of the bone marrow cells as a single cell suspension after 4 days' culture of fine marrow fragments, a linear relationship was obtained between the number of adherent cell colonies developing and the number of cells secondarily inoculated into the culture bottle. This suggests that 4 days' culture of the bone marrow cells with close intercellular interactions is sufficient for the 'conditioning' of the cells to develop adherent cell colonies. Activity of such colonies to support haemopoietic stem cell proliferation was also shown.

Enhancement of the adherence of hematopoietic stem cells to mouse bone marrow-derived stromal cell line MS-1-T by a tetrapeptide acetyl-N-Ser-Asp-Lys-Pro.

The tetrapeptide acetyl-N-Ser-Asp-Lys-Pro (AcSDKP) has been shown to inhibit in vivo the hematopoietic stem cell (spleen colony-forming unit; CFU-S) entry into cell cycle in cytosine arabinoside-treated mice. Our data showed that AcSDKP has no effect on interleukin 3 (IL-3)-dependent DA-1 cell proliferation or on granulocyte-macrophage colony-forming cell and mast cell colony formation, whereas it enhances the adherence of CFU-S to the bone marrow-derived hematopoietic supportive stromal cell line MS-1-T. AcSDKP suppresses MS-1-T proliferation but does not modify granulocyte-macrophage colony-stimulating activity secretion by these cells. This suggests that the peptide does not counteract the activity of the IL-3 receptor on CFU-S but acts on MS-1-T and in particular at the level of CFU-S/MS-1-T interactions.

Granulocyte-colony stimulating factor and stem cell factor are the crucial factors in long-term culture of human primitive hematopoietic cells supported by a murine stromal cell line.

The findings that murine marrow stromal cell line MS-5 supported the proliferation of human lineage-negative (Lin-) CD34+CD38- bone marrow cells in long-term culture have been reported. In this study, we analyzed this proliferating activity of MS-5-conditioned medium (CM) on human primitive hematopoietic cells. When Lin-CD34+CD38- cells of normal human cord blood cells were co-cultured with MS-5, colony forming cells (CFCs) were maintained over 7 weeks in vitro. Prevention of contact between MS-5 and Lin-CD34+CD38- cells by using membrane filter (0.45 micron) was negligible for this activity. This indicated that the activity of MS-5 on human primitive hematopoietic cells is a soluble factor(s) secreted from MS-5, which is not induced by the contact between MS-5 and Lin-CD34+CD38- cells. We tried to purify this soluble activity. An active material with a molecular weight of about 150 kDa, determined by gel filtration chromatography, solely supported the growth of Lin-CD34+CD38- cells and Mo7e, a human megakaryocytic cell line. This activity not only reacted with anti-mouse stem cell factor (mSCF) antibody on Western blots, but it was also neutralized in the presence of anti-mSCF antibody. Another active material with a molecular weight of about 20-30 kDa synergized with mSCF to stimulate the growth of Lin-CD34+CD38- cells but failed to do so alone, although this synergy was inhibited in the presence of soluble mouse granulocyte-colony stimulating factor (mG-CSF) receptor, which is a chimeric protein consisting of the extracellular domain of mG-CSF receptor and the Fe region of human IgG1. In addition, the latter molecule supported the growth of the G-CSF dependent cell line FD/GR3, which is a murine myeloid leukemia cell line, FDC-P2, transfected with mG-CSF receptor cDNA. Adding of anti-mSCF antibody and soluble mG-CSF receptor to the culture completely abrogated the activity of MS-5-CM. Recombinant (r) mSCF and rmG-CSF had synergistic activity on the growth of Lin-CD34+CD38- cells. These results indicated that the activity on Lin-CD34+CD38- cells included in MS-5-CM is based upon the synergistic effects of mSCF and mG-CSF.

Destruction of hematopoietic microenvironment by cytotoxic T cells.

Coculture of cytotoxic T cells (STIL-3 C5) derived from L8313 leukemic mice with hematopoietic supportive stromal cells (MS-5) resulted in the detachment of MS-5 cells from the culture dish, whereas helper T cells (STIL-3 DF) did not induce this detachment. The response of bone marrow (BM) adherent cells to the same treatment was similar to that of MS-5 cells. The detached cells were unable to proliferate further, and genomic DNA of these cells showed fragmentation, suggesting that hematopoietic stromal cells died of apoptosis. Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis revealed that STIL-3 C5 cells, but not STIL-3 DF cells expressed perforin, granzyme A & B, and Fas ligand. Fas was expressed in MS-5, BM adherent cells, MS-K and NIH/3T3 cells, which do not support hematopoiesis. These data suggest that the aforementioned factors mediate induction of apoptosis in MS-5 cells induced by direct cell-to-cell interaction with STIL-3 C5. This may explain the mechanism responsible for the destruction of the hematopoietic microenvironment by cytotoxic T cells in L8313 leukemia, from which STIL-3 cells are derived; it also suggests that destruction of hematopoietic tissue may be caused by leukemic cytotoxic T cells in some cases of leukemia.

alpha-Galactosylceramide (AGL-517) treatment protects mice from lethal irradiation.

AGL-517 (AGL) has an alpha-galactosylceramide structure and is a derivative of agelasphin-9b, which in turn is isolated from Agelas mauritianus and has immunomodulating activity. When administered before irradiation, AGL has been found to increase survival rates in lethally irradiated mice. In this study, we found that a single injection of AGL administered within 2 hours of lethal irradiation resulted in the long-term survival of mice without bone marrow transplantation. Peripheral blood hematology showed that AGL administration accelerated the recovery of hematopoietic parameters, including reticulocytes and red and white blood cells. Recovery of platelets was moderate. In addition, AGL significantly increased the number of endogenous colony forming units-spleen (E-CFU-S). AGL itself displayed no colony-stimulating activity, but AGL-stimulated spleen cell-conditioned medium (AGL-SCM) promoted the proliferation and differentiation of bone marrow mononuclear cells from normal mice and Lin marrow cells from 5-fluorouracil (5-FU)-treated mice. Using suitable assay systems, we analyzed cytokines in AGL-SCM and found significant increases in stem cell factor (SCF), interleukin-3 (IL-3), granulocyte-macrophage colony-stimulating factor (GM-CSF), and IL-6 levels compared with control SCM. Additionally, using immunoenzymetric assays, we assessed serum levels of these factors in AGL-treated mice after lethal irradiation. The serum concentrations of IL-3, GM-CSF, and IL-6 were substantially elevated, the maximum levels being reached within 2 hours of injection. Despite inducing the in vitro increase in SCF, AGL did not elevate serum SCF levels. However, certain levels of SCF (approximately 5 ng/mL) were detected in mouse serum regardless of irradiation or AGL treatment. When irradiated mice were given a cytokine cocktail composed of recombinant murine (rm) IL-3, rmGM-CSF, and recombinant human (rh) IL-6 three times a day for 6 days (1 microg of each factor per mouse per day) starting 2 hours after irradiation, 60% of the mice achieved 50-day survival. The radioprotective effect of AGL can be attributed, in part, to the cooperative effect of the cytokines induced by AGL in vivo. These findings suggest that AGL may be a useful in treating radiation-induced hematopoietic damage.

Effects of anti-CD44 monoclonal antibody on adhesion of erythroid leukemic cells (ELM-I-1) to hematopoietic supportive cells (MS-5): CD44, but not hyaluronate-mediated, cell-cell adhesion.

Cocultivation of erythroid leukemic cells (ELM-I-1) with hemopoietic supportive cells (MS-5) resulted in a specific adhesion of ELM-I-1 cells to MS-5 cells. This phenomenon was designated as rosette formation. After induction of differentiation of ELM-I-1 cells, rosette formation was reduced, and no rosette formation was observed between erythrocytes and MS-5 cells. Studies on anti-adhesion molecule antibody treatment have revealed that CD44 plays a key role in rosette formation. Expression of CD44 on (the membrane of) ELM-I-1 cells was reduced after differentiation, and no CD44 expression was detected on erythrocytes. CD44 was also expressed on MS-5. Hyaluronate is known as the ligand of CD44, but neither hyaluronidase treatment nor addition of excess hyaluronate to the assay system affected rosette formation. These data indicate that hyaluronate is not responsible for rosette formation. Anti-CD44 antibody (KM81), which recognized the hyaluronate binding site of CD44, inhibited rosette formation. But other monoclonal antibodies against different epitopes except for the hyaluronate binding site, even those against CD44's hyaluronate binding site, did not inhibit rosette formation. Thus, rosette formation between MS-5 cells and ELM-I-1 cells is mediated by CD44 but not by the hyaluronate binding site of CD44.