Modulation of erythropoiesis in rat bone marrow erythroblastic islands by cyclooxygenase inhibition

General Pharmacology 32 (1999) 423–428

Modulation of erythropoiesis in rat bone marrow erythroblastic islands by cyclooxygenase inhibition Vania Kalaidjieva * Department of Physiology, Faculty of Medicine, Thracian University, Armeiska Str. 11, Stara Zagora 6000, Bulgaria Manuscript received April 25, 1998; accepted manuscript July 7, 1998

Abstract We designed our study to explore how the inhibition of prostaglandins (PGs) could affect erythropoiesis in bone marrow erythroblastic islands (EIs). To this end, we used hypoxic-stimulated rats—hypobaric hypoxia (42.55 kPa/6 h)—pretreated or not with indomethacin (4 mg/kg/3 days). Blood sampling was done at 0 h, 24 h, and 72 h after hypoxia. The study included estimations of the plasma erythropoietin (EPO) level (by radioimmunoassay), peripheral blood, number of EI from classes I to V per femur, rate of immature cell’s differentiation into erythroblasts, and rate of repeated participation of macrophages in new El reconstruction. Plasma EPO rose significantly (p , 0.01) in all hypoxic rats: 40.5 6 10.15 mU/ml and 46.75 6 16.28 mU/ml and at 0 h versus 13.83 6 6.82 mU/ml in controls. An increased rate of cell differentiation into erythroblasts in EIs (p , 0.01), an enhanced reconstruction in involuted EIs, and a reduced number of maturing EIs (p , 0.01) were observed in all hypoxic animals. However, in indomethacin-pretreated rats, the stimulation of bone marrow erythropoiesis was better expressed. Our results favor the concept that PG inhibition does not attenuate the erythropoietic response to hypoxia and support the hypothesis about the important role of EI macrophages as a local regulator of bone marrow erythropoiesis.  1999 Elsevier Science Inc. All rights reserved. Keywords: Erythropoietin; Erythroblastic islands; Erythropoiesis; Bone marrow; Indomethacin; Hypoxia

The terminal stage of bone marrow erythropoiesis occurs in particular associations named erythroblastic islands (EIs). They consist of a centrally situated macrophage surrounded by developing red blood cells. The EIs are regarded as morphofunctional units of erythropoiesis where differentiation of colony-forming unitserythroid (CFU-E) into erythroblasts takes place (Bernard, 1991; Sadahira et al., 1995; Sasaki et al., 1993; Wilson, 1997). Studies by many investigators are sources of new insight into the mechanisms involving prostaglandins (PGs) both in erythropoietin (EPO) production, as a major regulator of erythropoiesis, and in the generation of red cell mass in bone marrow. PGE and PGF2a were reported to be inhibitors of hematopoietic progenitor cells (Lewis et al., 1981; Miller, 1992), whereas PGs of the A and I2 series are stimulators of those cells (Nelson et al., 1983). The beneficial effect of PG suppression by indomethacin on hematopoietic recovery in g-irradiated rats and in tumor therapy in laboratory animals * Tel.: 359-42-600-841; Fax: 359-42-600-705; E-mail: kalaidjieva@ hotmail.com.

and humans also has been widely discussed (DellaPuca and Gallicchio, 1996; Fedorocko and Mackova, 1996; Miller, 1992; Misurova et al., 1989). Considering the important role of PGs, we examined the contribution of cyclooxygenase inhibition by indomethacin to the evolution of erythropoiesis in bone marrow erythroblastic islands (EIs) and to the plasma EPO level in rats. To this end, we exposed animals to hypobaric hypoxia, which is known to be the main stimulus for EPO formation (Jelkmann, 1982; Madan et al., 1997; Nijhof et al., 1995) as well as a potent promoter of macrophage activity (Albina et al., 1995; Melillo et al., 1996). 1. Materials and methods The experiment was carried out on 55 male Wistar rats, weighing 160–250 g, with free access to food and water. The animals were divided into three main groups as follows: group I (n 5 15), controls; group II (n 5 15), rats exposed to hypobaric hypoxia (42.55 kPa/6 h) (Jelkmann, 1982); group III (n 5 15), rats pretreated with indomethacin, 4 mg/kg SC (Sigma, USA), for 3 days in a vehicle of polyethyleneglycol:water:ethanol (4.5:5.0:0.5)

0306-3623/99/$–see front matter  1999 Elsevier Science Inc. All rights reserved. PII: S0306-3623(98)00206-7

424

V. Kalaidjieva/General Pharmacology 32 (1999) 423–428

Table 1 Absolute reticulocyte count (Ret) 3 109/l, classes of maturity 1–4, and EPO concentration (mU/ml) in rats after hypobaric hypoxia (42.55 kPa/6 h) only or with indomethacin pretreatment (4 mg/kg/3 days SC) Reticulocyte classes of maturity Ret total Untreated controls Hypoxia, 0 h Hypoxia, 24 h Hypoxia, 72 h Indo 1 hyp, 0 h Indo 1 hyp, 24 h Indo 1 hyp, 72 h

237.13 314.6 164.06 292.05 149.59 228.75 508.97

6 6 6 6 6 6 6

63.19 91.95 86.69 63.70 52.13* 79.38 55.39**

Ret 1

Ret 2

0 0 0 22.15 6 14.71 0 0 0*

82.19 221.92 72.24 79.12 34.60 121.50 381.23

Ret 3 6 6 6 6 6 6 6

46.29 100.17 33.29 26.95 20.51* 43.93 59.42***

82.12 64.37 33.33 85.22 53.76 60.38 94.45

Ret 4 6 6 6 6 6 6 6

20.38 12.95 12.68 26.36 26.81 33.35 70.52

72.82 28.31 58.55 106.93 62.03 46.86 33.30

EPO 6 6 6 6 6 6 6

22.77 22.69 18.59 49.21 43.38 25.36 28.62*

13.83 40.5 12.07 12.92 46.75 12.88 7.50

6 6 6 6 6 6 6

6.82 10.15 5.51 6.03 16.28 1.03 4.02

Abbreviations: hyp, hypoxia; indo, indomethacin. Mean values 6 SD; statistical significance versus corresponding only hypoxic groups, * p , 0.05, ** p , 0.01, *** p , 0.001.

before hypoxia (Ganchev et al., 1989). To make the effects of indomethacin more clear, two additional groups were observed: with indomethacin 5 days prior to hypobaric subjection (n 5 5) and with indomethacin 3 days without a subsequent hypoxia (n 5 5). Blood samples were taken under ether anesthesia from the inferior vena cava immediately after hypoxia (at 0 h) and 24 and 72 h later. We evaluated the following parameters: Peripheral blood examination: plasma hemoglobin concentration, hematocrit, and erythrocyte count— by standard methods; absolute reticulocyte count and stages of maturity from first to fourth classes— by New Methylene Blue staining of blood smears, and plasma erythropoietin level—by radioimmunoassay (INCSTAR, Epo-trac, USA). Bone marrow examination: Absolute count of EIs per femur; classification of EIs by morphological study—classes of maturity I, II, III according to the predominant cells (young or mature), IV (involutive), and V (reconstructive) (Zakharov et al., 1984). In the hematopoietic tissue, there are EIs of classes of maturity I, II, and III, having, respectively, as many as 8, from 9 to 16, and more than 16 nuclei-containing erythroid cells in their composition. In addition, there are involutive and reconstructing EIs. The involutive EIs (class IV) are nuclei-containing erythroid cells, which are at late stages of differentiation and are not able to divide. Those cells are polychromatophilic and oxyphilic normoblasts and reticulocytes. The reconstructing EIs (class V) consist of both nondividing erythroid cells and cells at early stages of differentiation (pro-, erythroblasts, and basophilic normoblasts). For morphological identification, the femur of each rat was excised and marrow cells were flushed into a medium, composed of bovine serum albumin, heparin, and medium 199. After a slight mechanical dissociation, the cell suspension was filtered, incubated at 378C, and centrifuged in humid chambers on slides; the preparations were

then routinely stained, and the EIs were examined with a microscope (Zakharov et al., 1984). We analyzed the following indices, based on the study of EIs: rate of CFU-E differentiation into erythroblasts in EIs (i.e., class I 1 reconstructive class V 5 A2); and rate of repeated participation of macrophages in new EI reconstruction (i.e., EI reconstructive/EI involutive 5 A5) (Zakharov et al., 1990). Statistical analysis was carried out by using one-way analysis of variance; p-value differences of less than 0.05 were considered significant. Data are shown as mean values 6 standard deviation. 2. Results The acute hypoxia provoked considerable stimulation of erythropoiesis. The observed changes in peripheral blood, regarding erythrocyte count, hemoglobin, and hematocrit, did not differ considerably in the two experimental groups (II and III). However, a slight reduction in hematocrit was found in the indomethacin group at 24 h—0.39 6 0.03 versus 0.43 6 0.02 in controls (diminution with 9.56%) (p , 0.05). The absolute reticulocyte count was raised as early as 0 h after hypoxia: 314.6 6 91.95 3 109/l versus 237.13 6 63.19 3 109/l in untreated controls (p , 0.05), accompanied by an increase in the young cells: 221.92 6 100.17 3 109/l versus 82.19 6 46.29 3 109/l (p , 0.001). In contrast, the count of mature reticulocytes was reduced—28.31 6 22.69 3 109/l versus 72.82 6 22.77 3 109/l (p , 0.01) (Table 1). In indomethacin-treated rats, the release of reticulocytes into circulation was apparent later—at 72 h: 508.97 6 55.39 3 109/l versus 292.05 6 63.70 3 109/l in only hypoxic animals (p , 0.01) together with an increase in young cells 381.23 6 59.42 3 109/l (versus only hypoxic rats) (p , 0.001) (Table 1). An elevated plasma EPO level immediately after hypoxic exposure was observed—40.5 6 10.15 mU/ml (p , 0.001) and 46.75 6 16.28 mU/ml (p , 0.01), respectively, for groups II and III versus 13.83 6 6.82 mU/ ml in untreated controls (Table 1, Fig. 1). There was no

V. Kalaidjieva/General Pharmacology 32 (1999) 423–428

425

Fig. 1. Changes in plasma EPO (mU/ml) concentration in rats after hypobaric hypoxia (42.55 kPa/6 h) alone or with indomethacin pretreatment, 4 mg/kg/3 days SC; mean values 6 SD; controls (c), hypoxia (hyp), indomethacin (indo).

difference in EPO level between the two hypoxic groups throughout the study period. It was noted that the more indomethacin applications done, the higher EPO concentration. In the additional hypoxic group with 5-day treatment with indomethacin, EPO concentration was measured to be 52.45 6 8.63 mU/ml at 0 h versus 46.75 6 16.28 mU/ml in hypoxic rats that received this drug for 3 days only. In indomethacintreated animals without hypoxia (second additional group), the level of EPO was 35.75 6 9.52 mU/ml (p , 0.05) versus intact controls. The changes in the absolute EI count were apparent in indomethacin-treated animals at 72 h—371.32 6 70.2 3 103/femur versus 606.68 6 68.1 3 103/femur in only hypoxic rats (p , 0.01), and this reduction was consistent with the simultaneous accelerated release of reticulocytes into circulation (Table 2, Fig. 2). The analysis of EI classes of maturity in those animals showed a decrease in class IV (involutive) even at 0 h (p , 0.05) and an enhanced reconstruction (class V) at 0 h (p , 0.01) versus the corresponding islands in only hypoxic rats. The increase in the rate of CFU-E differentiation into erythroblasts in EIs (A2) and in the rate of repeated participation of macrophages in new EI reconstruction (A5) were remarkably expressed at 0 h and 24 h in the two experimental groups (Table 2). In rats with cyclooxygenase inhibition, those changes were expressed earlier. 3. Discussion The effect of prostaglandin inhibition or erythropoiesis is a subject of controversy (Fried, 1989; Lewis et al., 1981; Nelson et al., 1983; Suk et al., 1997; Taniguchi et al., 1989). Our study confirms the concept of the sustained proliferation in bone marrow after cyclooxygenase inhibition. Such an effect has been seen in g-irradiated rats pretreated with indomethacin (Fedorocko and

Mackova, 1996). Because the primary site of EPO formation in adult mammals is established to be the kidney (Jelkmann, 1982; Kramer et al., 1997; Madan et al., 1997), the observed changes in plasma EPO level after hypoxic exposure are probably indicative of the rate of renal EPO production. However, extrarenal sites, such as liver and bone marrow macrophages, are capable of producing small amounts of EPO (Fried, 1989). In addition, the relation between humoral and short cell-to-cell interactions in the regulation of red cell mass formation may be judged by the development of erythropoiesis in bone marrow EI (Wilson, 1997; Zakharov and Medianik, 1994; Zakharov et al., 1990). In confirmation, with these assumptions, the results obtained can be explained as follows: The increase in young reticulocyte count, as well as in the absolute number of reticulocytes in rats exposed to hypoxia only, was probably due to their facilitated release into the circulation by EPO. The observed decrease in the mature reticulocyte count might be related to the action of EPO to accelerate reticulocyte maturation toward a final erythrocyte stage. In indomethacin-treated rats, the release of reticulocytes into the circulation was apparent later— at 72 h (Table 1). In these cases, plasma EPO in animals exposed to hypoxia only and in those treated with cyclooxygenase inhibitor did not differ. This means that the effect of EPO on bone marrow barrier in both kinds of rats is similar. This observation suggests an indomethacin-mediated influence on reticulocyte production rate and on reticulocyte membrane adhesion to the other erythroblastic cells of the islands (Weiss and Geduldig, 1991; Zakharov et al., 1990). The EPO level rose significantly in all experimental groups after hypoxic exposure with or without prostaglandins suppression. The lack of EPO attenuation

Abbreviations: hyp, hypoxia; indo, indomethacin. Absolute count per femur, classes of maturity (I–V), and indices of erythropoiesis: rate of CFU–E differentiation into erythroblasts in EIs (i.e., class I 1 reconstructive class V 5 A2) (3 103/femur); rate of repeated participation of macrophages in new EI reconstruction (i.e., EI reconstructive/EI involutive 5 A5) in rats after hypobaric hypoxia (42.55 kPa/6 h) alone or with indomethacin pretreatment, 4 mg/kg SC; mean values 6 SD; statistical significance versus corresponding only hypoxic groups, * p , 0.05, ** p , 0.01, *** p , 0.001.

0.12 1.03 0.44 5.22 3.75 2.78*** 2.31* 6 6 6 6 6 6 6 0.87 2.53 2.37 9.74 6.77 11.08 2.55 59.3 78.1 142.8 42.8 99.0 77.9 67.4**

EI I 1 EI V

6 6 6 6 6 6 6 133.63 187.72 266.96 291.74 336.1 202.84 176.15 160.22 78.25 114.93 40.33 56.17 19.81 92.81 149.79 46.42 85.51 52.9 95.77 27.71 39.43 43.77 34.83 25.53 109.84 43.72 86.23 23.49

EI II

36.5 33.1 41.3 49.3 32.2 52.7* 18.2 6 6 6 6 6 6 6 78.6 102.78 70.42 111.88 94.24 148.41 45.02

EI I

109.6 93.5 230.1 68.1 123.5 122.8 70.2** 6 6 6 6 6 6 6

EI total

566.0 450.0 563.3 606.7 626.0 485.0 371.3 Controls Hypoxia, 0 h Hypoxia, 24 h Hypoxia, 72 h Indo 1 hyp, 0 h Indo 1 hyp, 24 h Indo 1 hyp, 72 h

Table 2 Bone marrow EIs (3 103/femur)

EI classes of maturity

6 6 6 6 6 6 6

25.0 19.2 10.9 57.6 19.7 43.1* 17.2*

EI III

6 6 6 6 6 6 6

32.9 10.1 55.1 20.8 16.5*** 10.3 24.1

EI IV

6 6 6 6 6 6 6

52.8 18.7 46.78 26.4 16.3* 10.1*** 39.1**

EI V

6 6 6 6 6 6 6

19.6 57.4 102.3 64.6 70.4** 62.6 54.6**

Indices

EI V/EI IV

V. Kalaidjieva/General Pharmacology 32 (1999) 423–428

229.43 290.5 337.38 403.62 430.14 351.25 223.37

426

Fig. 2. Bone marrow EIs (3103/femur)—absolute count per femur— in rats after hypobaric hypoxia (42.55 kPa/6 h) alone or with indomethacin pretreatment, 4 mg/kg/3 days SC. **Statistical significance versus only hypoxic rats (p , 0.01); mean values 6 SD; controls (c), hypoxia (hyp), indomethacin (indo).

after indomethacin treatment is consistent with the findings of other authors (Fried, 1989; Ganchev et al., 1989). This fact might be explained in several ways: (1) severe renal hypoxia leading to a significant EPO secretion despite the block of PG production; (2) modulation of the adrenergic transmission—attenuation by indomethacin of the PG suppressive action on the vasoconstriction due to angiotensin II, vasopressin, catecholamines, sympathetic nerve activity—factors that contribute to the decreased tissue oxygen delivery (respectively in the kidney); (3) absence of the direct vasodilator effect of PG; (4) strong stimulation of extrarenal EPO production by liver cells and bone marrow macrophages (through indomethacin action on tissue PG level) (Brandan et al., 1997; Fried, 1989). The state of the terminal stage of erythropoiesis may be estimated by observing the interactions between the central macrophage and the erythroblasts within bone marrow EI (Bernard, 1991; Sadahira et al., 1995; Zakharov and Medianik, 1994). The enhanced development of EIs is demonstrated by the reduction in absolute EI count in indomethacin-treated animals and is consistent with the simultaneous accelerated release of reticulocytes into the circulation. Furthermore, in animals with cyclooxygenase inhibition, an observable stimulation of erythropoiesis was expressed by diminution of the involutive class IV of EIs, as well as by an enhanced generation of new EIs (EI reconstructive). This finding is supported by the increase in both rate of CFU-E differentiation into erythroblasts (A2) and rate of repeated participation of macrophages in new EI reconstruction (A5). The latter is in accordance with the concept of the important role of the centrally situated macrophage through potentiation of its adhesive capacity to erythroblasts after hypoxic exposure (Albina et al., 1995; Brandan et al., 1997). An increased proliferation potential of the red cell line in bone marrow as early as 1 day after indometha-

V. Kalaidjieva/General Pharmacology 32 (1999) 423–428

cin administration has also been found by other authors (Fedorocko and Mackova, 1996; Ganchev et al., 1989; Taniguchi et al., 1989). PG inhibitors are assumed to be able to shift the regulatory balance toward the predominance of a positive control. The relations between interleukin-1 and PGs and between granulocyte–macrophage colony-stimulating activity and PGs could be the reason for the greatly amplified proliferation in indomethacin-pretreated mice (Pospisil et al., 1992). Inhibition of PGE2 production by indomethacin releases erythroid cells from PGE2-mediated suppression (Miller, 1992; Suk et al., 1997). PGE2 appears to mediate their effects by raising cAMP levels in macrophages and therefore reducing macrophage activity (Rutherford and Schook, 1992). In conclusion, the cyclooxygenase inhibition by indomethacin does not attenuate adequate erythropoietin production after hypoxic exposure but enhances the development of bone marrow erythroblastic islands. The activation of centrally situated macrophage cells within erythroblastic islands plays a significant role in the local regulation of bone marrow erythropoiesis.

4. Summary The terminal stage of bone marrow erythropoiesis occurs in particular formations named erythroblastic islands (EIs) and depends on short-range cell-to-cell interactions within EIs as well as on a number of hormonal factors. Having in view the influence of prostaglandins on the erythropoietin production rate, we designed our study to explore how inhibition of those modulators could affect erythropoiesis in EIs. To this end, we used hypoxia-stimulated rats—hypobaric hypoxia (42.55 kPa, 6 h) with or without pretreatment with indomethacin (4 mg/kg/3 days). Blood sampling was done at 0 h, 24 h, and 72 h after hypoxia. The study included estimations of plasma EPO level (by radioimmunoassay), peripheral blood and erythroblastic island examination— number of EIs from classes I to V per femur, rate of immature cell differentiation into erythroblasts, and rate of repeated participation of macrophages in new EI reconstruction. As expected, plasma EPO rose significantly (p , 0.01) in all hypoxic rats—40.5 6 10.15 mU/ ml and 46.75 6 16.28 mU/ml at 0 h versus 13.83 6 6.82 mU/ml in controls. Afterward, EPO levels fell to normal values. Bone marrow recovery was markedly expressed throughout the observed period—an increased rate of cell differentiation into erythroblasts in EIs (p , 0.01), an enhanced reconstruction in involuted EI, and a reduced number of maturing EI (p , 0.01) were observed in all hypoxic animals. However, in indomethacin-pretreated rats, the stimulation of bone marrow erythropoiesis appeared to be earlier and better expressed compared with only hypoxic rats. Our results favor the concept that cyclooxygenase in-

427

hibition does not attenuate the erythropoietic response to hypoxia. Inhibition of PGE2 production by indomethacin releases erythroid cells from PGE2-mediated suppression. The activation of centrally situated macrophage cells within erythroblastic islands plays a significant role in the local regulation of bone marrow erythropoiesis.

References Albina, J.E., Henry, W.L. Jr., Mastrofrancesco, B., Martin, B.A., Reichner, J.S., 1995. Macrophage activation by culture in an anoxic environment. J Immunol 115, 4391–4396. Bernard, J., 1991. The erythroblastic island: past and future. Blood Cells 17, 5–14. Brandan, N., Arguirre, M., Carmuega, R., Alvarez, M., Juaristi, J., 1997. Proliferative and maturative behaviour patterns on murine bone marrow and spleen erythropoiesis along hypoxia. Acta Physiol Pharmac Ther Latinoam 47, 125–135. DellaPuca, R., Gallicchio, V.S., 1996. The regulation of phospholipase-A2 (PLA-2) by cytokines expressing hematopoietic growthstimulating properties. Proc Soc Exp Biol Med 212, 174–184. Fedorocko, P., Mackova, O., 1996. Radioprotective effects of combination broncho-vaxom, a macrophage activator, and indomethacin, an inhibitor of prostaglandin production: relationship to myelopoiesis. Eur J Haematol 56, 54–61. Fried, W., 1989. Regulation of extrarenal erythropoietin production. Adv Exp Med Biol 271, 39–51. Ganchev, T., Negrev, N., Mileva, V., 1989. Effects of indomethacin on erythropoiesis and plasma iron in rats. Acta Physiol Pharmac Bulg 15, 53–57. Jelkmann, W., 1982. Temporal pattern of erythropoietin titers in kidney tissue during hypoxic hypoxia. Pfluegers Arch 393, 88–91. Kramer, B.K., Bucher, M., Sandner, P., Ittner, K.P., Riegger, G.A., Ritthaler, T., Kurtz, A., 1997. Effects of hypoxia on growth factor expression in the rat kidney in vivo. Kidney Int 51, 444–447. Lewis, J.P., Moores, R.R., Neal, W.A., Garver, F.A., Lutcher, C.L., Zucali, J.R., Mirand, E.A., 1981. Inhibition of Friend virus (FVP)induced murine erythropoiesis with prostaglandin (PFG2a): potentiation and inhibition of erythropoietin and prevention of both with PGD2. Exp Hematol 9, 540–549. Madan, A., Varma, S., Cohen, H.J., 1997. Co-transactivation of the 39 erythropoietin hypoxia inducible enhancer by the HIF-1 protein. Blood Cells Mol Dis 10, 169–176. Melillo, G., Taylor, L.S., Brooks, A., Cox, G.W., Varesio, L., 1996. Regulation of inducible nitric oxide synthase expression in IFNgamma-treated murine macrophages cultured under hypoxic conditions. J Immunol 157, 2638–2644. Miller, S., 1992. Age-related differences in the effect of in vivo administration of indomethacin on hemopoietic cell lineage of the spleen and bone marrow of mice. Experimentia 48, 674–677. Misurova E., Kropacova K., Chlebovsky O., Pado D., 1989. The effects of indomethacin on nucleic acids in blood, hemopoietic and lymphoid tissues in continuously irradiated rats. Neoplasma 36, 541–547. Nelson, P., Brookins, J., Fisher, J., 1983. Erythropoietic effects of prostacyclin (PGI2) and its metabolite 6-keto-prostaglandin (PGE1). J Pharmac Exp Ther 226, 493–499. Nijhof, W., Haan, G., Pietens, J., Dontje, B., 1995. Mechanistic options of erythropoietin-stimulated erythropoiesis. Exp Hematol 23, 362–368. Pospisil, M., Hofer, M., Pipalova, I., Viklicka, S., Netikova, J., Sandula, J., 1992. Enhancement of hematopoietic recovery in gamma-

428

V. Kalaidjieva/General Pharmacology 32 (1999) 423–428

irradiated mice by the joint use of diclofenac, an inhibitor of prostaglandin production, and glucan, a macrophage activator. Exp Hematol 20, 891–895. Rutherford, M., Schook, L., 1992. Macrophage function in response to PGE2, l-arginine deprivation and activation by colony-stimulating factors is dependent on hematopoietic stimulus. J Leukocyte Biol 52, 228–235. Sadahira, Y., Yoshino, T., Monobe, Y., 1995. Very late activation antigen 4-vascular cell adhesion molecule 1 interaction is involved in the formation of erythroblastic islands. J Exp Med 181, 411–415. Sasaki, K., Iwatsuki, H., Suda, M., Itano, C., 1993. Scavenger macrophages and central macrophages of erythroblastic islands in liver hemopoiesis of the fetal and early postnatal mouse: a semithin light- and electron-microscopic study. Acta Anat 147, 75–82. Suk, K., Sipes, D.G., Erickson, K.L., 1997. Enhancement of B-cell translocation gene-1 expression by prostaglandin E2 in macrophages and the relationship to proliferation. Immunology 91, 121–129.

Taniguchi, S., Shibuya, T., Harada, M., Niho, Y., 1989. Prostaglandinmediated suppression of in vitro growth of erythroid progenitor cells. Kidney Int 36, 712–718. Weiss, L., Geduldig, U., 1991. Barrier cells: stromal regulation of hematopoiesis and blood cell release in normal and stressed murine bone marrow. Blood 78, 975–990. Wilson, J.G., 1997. Adhesive interactions in hemopoiesis. Acta Haematol 97, 6–12. Zakharov, Y.M., Medianik, B.V., 1994. Patterns of erythropoietic reconstruction in the erythroblastic islands. Fiziol Zh Im I M Sechenova 80, 76–82. Zakharov, Y.M., Melnikov, I.Y., Rassokhin, A.G., 1984. Study on erythropoiesis by modification of method for isolation of erythroblastic islets. Hematol Transfusiol 4, 52–54. Zakharov, Y.M., Melnikov, I.Y., Rassokhin, A.G., 1990. Classification of erythroblastic islets of the bone marrow and the study of their cellular composition. Arkh Anat Gistol Embriol 98, 38–42.