Mechanism of erythropoietin action on the erythroid progenitor cells induced from murine erythroleukemia cells (TSA8)

Growth or differentiation factors have recently been identified in several cell systems and the mechanism of their receptor-mediated signal transductions has been characterized (see review; Nishizuka, 1987).

Erythropoietin is a factor in the differentiation and proliferation of the erythroid progenitor cells such as BFU-E (blast-forming unit-erythroid) and CFU-E (colony-forming unit-erythroid) (Iscove et al. 1974). In contrast to the well-known function in erythropoietin on erythropoiesis, its mode of action has not been well documented. Although several approaches to determining the mechanism of action of erythropoietin on these erythroid progenitor cells have been described, no conclusive evidence has yet been obtained because in most of the work a mixed population of bone marrow cells was used, thus limiting the analysis.

We have demonstrated that a murine erythroid cell line TSA8 (Shibuya & Mak, 1983) can be induced to become responsive to erythropoietin; in other words, after the addition of inducers such as dimethyl sulfoxide (DMSO), hexamethylene-bis acetamide (HMBA) and butyric acid the cells can form differentiated colonies in the semisolid medium only in the presence of erythropoietin. These colonies resemble the CFU-E from mouse fetal liver cells or bone marrow cells in several aspects (Mishina & Obinata, 1985; Noguchi et al. 1987, 1988). This in vitro system is applicable to the analysis of action of erythropoietin. We have shown that erythropoietin receptors were present before induction and that their number and affinity did not change following induction (Noguchi et al. 1987). Thus, the change in response to erythropoietin may be due to alterations in the second messenger system rather than the receptor level.

Since the receptor-mediated signal transduction system of erythropoietin has not been well documented, the intracellular mediator for the induction of differentiation response by erythropoietin must be determined. Many glycoprotein hormones elicit a biological response by activation of the adenyl cyclase system (Pierce & Parsons, 1981). Addition of erythropoietin was reported to cause biphasic cAMP levels in rabbit bone marrow cells (Chiuini et al. 1979), but to have no effect on calf or rat fetal liver cells (Graber et al. 1974). It has been reported that erythropoietin activated adenylate cyclase in membrane preparations of progenitor cells (Bonanou-Tzedaki et al. 1986). These contradictory reports on the involvement of the adenylate cyclase system in erythropoiesis may be due to a mixed population of erythroid cells being examined during their maturation.

We surveyed several compounds affecting the transduction system and obtained results suggesting that the cAMP pathway is involved in the erythropoietin action. Differing from the usual cAMP pathway systems, however, cAMP cannot act as a direct second messenger but only in combination with signals through the receptor system. The contribution of cAMP pathway to the effect of the erythropoietin receptor pathway changes depending on the induction. This newly recognized system may be important in the differentiation response of erythroid progenitor cells to erythropoietin action.

Forskolin and 3-isobutyl-l-methylxanthine (IBMX, phosphodiesterase inhibitor) were purchased from Calbiochem Co., Ltd, and dibutylyl cAMP, 8-bromo-cAMP and 8-bromo cGMP, from Sigma Chemical. Cholera toxin was obtained from the Chemical and Immunological Research Institute. Cyclic AMP assay kit was purchased from Yamasa Co. Purified recombinant erythropoietin was supplied by Kirin-Amgen Co. through Professor F. Takaku of University of Tokyo. IAP (Islet-activating protein) was a gift from Professor M. Ui of University of Tokyo.

TSA8 cells were grown in Iscove’s modified Dulbecco medium supplemented with 15 % fetal calf serum as described (Mishina & Obinata, 1985). For induction, DMSO (1 %) was added to a suspension of the cells at a density of 2× 10⁵ cells ml⁻¹. In this work, the clone shows better inducibility (70–80%) than that in previous work (30–40 %) (Mishina & Obinata, 1985; Noguchi et al. 1987). This clone was obtained through several recloning steps after long maintenance (3 years) in our laboratory.

TSA8 cells were induced as described previously (Mishina & Obinata, 1985). Usually, 1·5–2 days after the addition of DMSO (1·0%) in a liquid culture, the cells were transferred to a semisolid medium and erythropoietin added (0·5 i.u. ml¹) (Iscove et al. 1974). The single cells formed colonies of 16–32 cells after 2 days of culture. The differentiating colonies producing hemoglobin varied with the amount of erythropoietin added; these were counted after staining with benzidine. In usual induction, on day 1 after the addition of DMSO, approximately 30 % of cells produced hemoglobinpositive colonies and, on day 2, 80% become hemoglobinpositive. To test the stimulatory effect of the compounds, the cells on day 1 after the DMSO addition were transferred to the semisolid medium and the test compounds were added simultaneously with erythropoietin. After 2 days incubation in the medium, the differentiated colonies were scored.

To determine the intracellular cAMP level of TSA8 cells, the cells were collected and washed twice with Hanks’-Hepes balanced solution. They were then resuspended in a buffer containing IBMX (250–500 pui) and were boiled for 2 min to extract cAMP. After centrifugation, the supernatant fluid was assayed for cAMP. Radioimmunoassay of cAMP was done with antisera to cAMP using a Yamasa assay kit as described (Pauline & Philip, 1985).

TSA8 cells were induced with 1 % DMSO for 1–2 days and then were transferred to a semisolid medium containing 0·5 i.u. ml⁻¹ of erythropoietin. Erythroid colony formation was assayed after two days of incubation at 37 °C and the differentiated colonies were counted after benzidine staining (Noguchi et al. 1987). On day 1 after induction with DMSO, approximately 20% of the colonies were benzidine-positive and, on day 2, approximately 80% were benzidine-positive. To examine the stimulatory effect of the compounds, cells 1 day after induction were used. As reported earlier for the effect on erythroid progenitor cells of herbimycin, the specific inhibitor for tyrosine phosphorylation of src-related gene products (Noguchi et al. 1988), both the proliferative and the differentiation responses of the cells induced by erythropoietin can be assayed simply by measuring the colony-forming ability and the differentiated colony formation.

Critical to the characterization of the G proteins that regulate adenylate cyclase was the observation that these proteins serve as substrate for certain bacterial toxins. Cholera toxin, for example, activates the stimulatory G protein of adenylate cyclase (Gs) by catalysing the transfer of ADP-ribose to its 43×10³Af_r α-subunit (Cassel & Pfeuffer, 1978; Gilman, 1984). Pertussis toxin or IAP (islet-activating protein), on the other hand, ADP-ribosylates the 41 × 10³M_r α-subunit of the inhibitory G protein Gi (Katada & Ui, 1982; Bokoch et al. 1983). As a consequence of the action of pertussis toxin, Gi is inactivated and hormonal inhibition of adenylate cyclase is blocked. Both toxins can act upon G proteins that regulate processes other than adenylate cyclase.

Thus, we first examined the effect of these toxins on the CFU-E formation. When the drugs were added to the CFU-E assay of the TSA8 cells at 2 days after induction with DMSO, cholera toxin stimulated slightly the formation of the differentiated colonies. No inhibitory or stimulatory effects were detected with the addition of pertussis toxin (data not shown). It is therefore possible that neither the inactivation of Gi nor the inhibition of adenylate cyclase is involved in the erythropoietin action. Neither toxin shows any effect on the colony formation. Thus, neither toxin affects the proliferative response induced by erythropoietin.

The stimulatory effect of cholera toxin was further examined in the TSA8 cells 1 day after induction with DMSO. With the addition of cholera toxin to the induced TSA8 cells in the semisolid medium, an increase in the benzidine-positive colony formation was observed (Fig. 1). The effect of cholera toxin on differentiation and proliferation is shown in Fig. 2. In this experiment, with increase of toxin concentration, the proportion of differentiated colonies increased from approximately 20 to 70%, while the colony-forming ability did not change. This stimulatory effect was dependent solely on the presence of erythropoietin. Thus, activation of the stimulatory G protein of adenylate cyclase (Gs) may affect the erythropoietin action on the differentiation response.

Since cholera toxin is known to increase the intracellular level of cAMP, we examined the effect of the compounds increasing the cAMP intracellular level.

The effect of phosphodiesterase inhibitor (IBMX) was first examined (Fig. 3). In this experiment, with the increasing concentration of IBMX, the proportion of the differentiated colonies increased from approximately 35 % to approximately 70 % without affecting the colony formation. Again, this stimulatory effect was only observed when erythropoietin was present in the assay. Forskolin, another nonhormonal stimulator of adenylate cyclase, stimulated in a similar fashion (Fig. 3).

The synergistic or additive effect of the two compounds was examined (Fig. 4); the only difference observed was a notable decrease in the colony formation in the presence of both drugs.

The above results indicate that the increasing level of intracellular cAMP has a stimulatory effect on erythropoietin action on the erythroid progenitor cells. However, these stimulatory effects are only observed in the presence of erythropoietin. Cholera toxin, forskolin and IBMX themselves do not induce the differentiation or proliferation of erythroid progenitor cells in the absence of erythropoietin, nor are they potent inducers for TSA8 cells to commit to CFU-E as shown in Table 1. This result indicates that the cAMP pathway induction is not involved in the commitment event of the cells (Housman et al. 1978). The differentiation response of the progenitor cells to erythropoietin thus cannot be mimicked by simply increasing the intracellular level of cAMP with either forskolin or IBMX. This is confirmed by the fact that the addition of 8-bromo cAMP to the cells showed no differentiation effect on the induced TSA8 cells as shown in Table 1.

If erythropoietin activates adenylate cyclase and cAMP is used as the second messenger for the intracellular signalling as demonstrated in several hormone systems, the level of adenylate cyclase and therefore the intracellular level of cAMP may increase in the erythroid progenitor cells depending on the addition of erythropoietin. The intracellular cAMP levels were then determined in the TSA8 cells by radioimmunoassay at 2 days after induction with DMSO in the presence of IBMX. As shown in Fig. 5, in the presence of phosphodiesterase inhibitor, the intracellular cAMP accumulated for 30 min but there was no increase in its accumulation with the addition of erythropoietin. Thus, erythropoietin receptor does not mimic adenylate cyclase system directly. When the cAMP level during induction was measured, its level decreased with induction as shown in Table 2.

The responsiveness of TSA8 cells to erythropoietin was induced with DMSO and this change is not caused by an increase in the number or affinity of the receptors (Noguchi et al. 1987). Since adenylate cyclase systems did not show a notable change by 2 days after induction with DMSO when the responsiveness to erythropoietin was maximum, the activation or de novo synthesis of adenylate cyclase systems is not involved in acquisition of the responsiveness to erythropoietin. The stimulatory effect of cholera toxin or compounds affecting the adenylate cyclase system was only observed in the TSA8 cells on day 1 after induction and not on day 2. On day 2, the formation of benzidine-positive colonies is dependent on the concentration of erythropoietin and no stimulation of differentiation was detected by these compounds even in the reduced concentration of erythropoietin (data not shown). Therefore, this drastic change in the stimulatory response of the TSA8 cells after induction may be related to the acquisition of responsiveness to erythropoietin. The effect of the compounds stimulating adenylate cyclase was then examined in the TSA8 cells during induction (Fig. 6). Without induction, the addition of IBMX did not show any stimulation on the CFU-E formation. However, the stimulatory effect increased with induction with DMSO, but near 2 days it decreased. The stimulatory step seems to coincide with the responsiveness to erythropoietin. Essentially similar results were obtained with cholera toxin (data not shown). These indicate that, not only is the increase in the cAMP level itself required, but some factor may also be needed in combination with the receptor-mediated pathways.

Erythropoietin stimulates committed erythroid progenitor cells to differentiate, but the mechanism involved is unknown. An erythropoietin preparation obtained using recombinant technology made possible analysis of the surface receptors on erythroid progenitor cells (Mayeux et al. 1987; Tojo et al. 1987) or erythroleukemia cells responsive to erythropoietin (Noguchi et al. 1987; Todokoro et al. 1987). However, the mechanism of erythropoietin action and the developmental response process of the erythroid progenitor cells to erythropoietin have not been elucidated. Murine erythroleukemia cells (TSA8) are suitable for these analyses, because these cells become responsive to erythropoietin only after induction, and these changes may resemble the differentiation and proliferation processes of the erythroid progenitor cells. (Mishina & Obinata, 1985; Noguchi et al. 1987). The proliferative response and differentiation response of erythroid progenitor cells to erythropoietin are separable (Noguchi et al. 1988). Using TSA8 cells, we were able to measure separately the stimulatory effect of the compounds on the two types of erythropoietin-induced response. In the present studies, we demonstrated that the differentiation response can be mimicked by the compounds affecting the adenylate cyclase system without affecting the proliferation. However, cAMP itself cannot replace the action of erythropoietin. It is likely that the adenylate cyclase systems couple with the ligand-occupied receptor and enhance its function. This is the first report clearly showing the synergism of the adenylate cyclase system in the erythropoietin action on erythroid progenitor cells.

It has been reported that the adenylate cyclase activity of rabbit marrow immature erythroblasts was activated by erythropoietin and haemin (Bonanou-Tzedaki et al. 1986). However, forskolin failed to produce a mitogenic response in cell culture. Thus, it is not clear that activation of adenylate cyclase by erythropoietin is a key event for the erythropoietin action. On the contrary, in the TSA8 cell system, the intracellular levels of cAMP did not increase with the addition of erythropoietin (Fig. 5). A likely explanation for the discrepancy between the two sets of data may be that the adenylate cyclase system is involved in the proliferation response to the erythropoietin action but not in the differentiation response. It is possible that high levels of adenylate cyclase are required for the proliferation of the normal progenitor cells. Since TSA8 cells are transformed cells and, therefore, possess a high level of adenylate cyclase, further increase in adenylate cyclase is not required to achieve proliferation even after erythropoietin addition.

A tentative model of the process of induction of responsiveness of TSA8 cells to erythropoietin is as follows. Adenylate cyclase systems are active enough to be responsive to erythropoietin before induction, and the increase in the intracellular cAMP by compounds like forskolin, IBMX, and cholera toxin is not enough to account for the differentiation response. The addition of erythropoietin to the cells did not increase the intracellular cAMP level. Thus, the erythropoietin receptor and the stimulatory G protein (Gs) are not linked as observed in other systems. However, during induction, the effect of the compounds that affected the intracellular cAMP level increased transiently. The contribution of cAMP pathway to erythropoietin action may be to facilitate the erythropoietin receptor pathway. It is likely that the cAMP pathway is involved in the activation of protein kinase A and therefore in the phosphorylation of an acceptor protein. The acceptor protein may, in turn, modify some components of the receptor pathway and/or act synergistically with the receptor pathway. The transient contribution of the cAMP pathway during the induction of TSA8 cells as shown in Fig. 6 may reflect the stage of development of the erythroid progenitor cells. It is noted that the intracellular cAMP levels decreased depending on the induction of differentiation of TSA8 cells (Table 2). Similarly, the adenylate cyclase levels decreased continuously as the cells developed (Setchenska & Arnstein, 1983). Furthermore, Bonanou-Tzedaki et al. (1986) reported that the magnitude of the response to hormonal stimulation depends on the stage of erythroid cell development and is greater in the more immature cells. The development of the response in erythroid progenitor cells to erythropoietin requires a factor combining both the receptor pathway and the cAMP pathway to produce an intracellular signal. It would be interesting to know whether this type of coupled signalling system is requisite for other blood progenitor cells. In embryonic induction, part of the cells in the embryos are .responsive only to the inducer substances; this phenomenon is called competence (see review; Gurdon, 1987). The process in the TSA8 cells of becoming responsive to erythropoietin may be similar to the acquisition of competence. It is hoped that by using these new murine erythroleukemia cells responsive to erythropoietin, the molecular events of the erythropoietin action on erythroid progenitor cells can be elucidated.

Most of the work was performed in the laboratory of Professor S. Natori of University of Tokyo. We thank him for his kind support and helpful discussions. We also thank Drs M. Ui and H. Kurose for their helpful suggestions on signal transduction studies. This work was partly supported by a Grant-in-aid for scientific research from the Ministry of Education, Science and Culture of Japan.