Experimental Hematology 24:660-669 (19961 @ 1996 International Society for Experimental Hematology

Rapid Communication

Insights into the cellular mechanisms of erythropoietin-thrombopoietin synergy Thalia Papayannopoulou, Martha Brice, Denise Farrer, Kenneth Kaushansky

University of Washington, Department of Medicine, Seattle, WA

Offprint requests to: Thalia Papayannopoulou, MD, DrSci, University of Washington, Division of Hematology, Box 357710, Seattle, WA 98195-7710

(Received 24 January 1996; revised 14 February 1996; accepted 16 February 1996)

Abstract Using suspension cultures of purified bone marrow CD34+ cells, we have analyzed the effects of the combination of ery­thropoietin (Epo) and thrombopoietin (Tpo) on the in vitro differentiation toward erythropoiesis and thrombopoiesis. The number of CD41+ cells that accumulated over 2 weeks of culture, as well as the number of globin+ cells in the same cul­tures, was found to be significantly higher with the Epo+ Tpo combination compared to either cytokine alone. No evidence was found that Tpo affected the differentiative action of Epo. Instead, there was a significant expansion of erythroid pro­genitors, both erythroid colony-forming and burst-forming units (CFU-E and BFU-E), by 7 days in culture, suggesting a proliferative effect of Tpo on erythroid cells in vitro. To deter­mine the phenotypic features of erythroid progenitor cells which were targets of Tpo's action, and specifically to inquire whether the effect was directed mainly toward bipotent ery­throid/megakaryocytic (E+Mk) progenitors, we isolated sub­sets enriched for both erythroid and megakaryocytic pro­genitors from CD34+ cells. We found that 1) BFU-E and CFU-Mk co-segregate in the subset of CD34+ cells that is nega­tive for the phosphatase isoform CD45RA; 2) the presence of CD41 on this subset appears to segregate late erythroid and late CFU-Mk from early erythroid and early CFU-Mk, which are CD41-negative; 3) bipotent erythroid/Mk progenitors, studied by single-cell culture assays, were found mainly in the CD41+ and rarely in the CD4r subsets, which included more multipotent progenitors; 4) by comparing the frequencies of pure erythroid or pure megakaryocytic progenitors to that of bipotent E+Mk progenitors, we conclude that the erythroid­enhancing effect of Tpo is directed mainly toward pure ery­throid progenitors expressing CD41 and Mpl, as suggested by independent experiments employing anti-Mpl antibody, rather than only on bipotent E+Mk progenitors.

Key words: Erythropoietin-Thrombopoietin-Bipotent erythroid/Mk progenitors-CFU-MK and BFU-E purification

Introduction In hematopoiesis, differentiation along specific lineages is under the control of late-acting cytokines responsible for amplification of progenitor and precursor cells of each lineage

and for the expression of a well-coordinated program in the end-stage, functional cells [1]. The cloned ligand for the c-Mpl receptor, Tpo, appears the role of a lineage-specific cytokine for the mE'l?:akar lineage [2-6]. Its pivotal role in m<~ga.ka.rvc>cvtOJJOie~ platelet production is supported by several in vitro vivo studies [7,8], especially those with c-Mpl-null and more recently with Tpo-null mice [10]. The effects on in vitro megakaryocytopoiesis are amplified combined with other cytokines in a synergistic (KL additive (IL-3) fashion [11], but its exact role in from megakaryocytes remains to be clarified. the detailed spectrum of progenitors on which effects and their composite phenotype have not defined. As a late-acting cytokine, Tpo's in early-acting cytokines (IL-3 or KL) is not exrJecitedN significant synergy with Epo, another lin cytokine, observed both in in vitro and in [12, 13], is of special interest and has relied on undocumented but likely-expression of a receptor by megakaryocytic progenitors, as megakaryocytes [14].

The focus of the present work was to analyze on erythroid differentiation in vitro. Using tures of purified human bone marrow CD34+ ed a detailed quantitative assessment of the Epo+ Tpo combination on both megakaryocytic differentiation over time within the same ' ulluJ'""

that Tpo has a bidirectional effect on both megakaryocytopoiesis, as erythroid and m progeny are enhanced in its presence. Tpo's thropoiesis do not result from an enhancing ferentiative effect of Epo, but are mediated tive effects on early and late erythroid • "''"'c;u<·•~ these cultures. Using purified CD34+ subsets cultures, we concluded that the great throid progenitors that are targets for Tpo's effects are not bipotent (E+Mk), but are rather progenitors carrying the c-Mpl receptor and showing common surface characteristics of are targets for both Epo and Tpo, provide the long-standing relationship between megakaryocytopoiesis.

Papayannopoulou et al.: Epo and Tpo Synergy

cadaveric bone marrow cells obtained from Northwest Center, Puget Sound Blood Bank (Seattle, WA), were

washed, and incubated overnight in IMDM with 10% calf serum on tissue culture plates to remove adherent From the nonadherent cells, CD34+ cells were isolated

direct immunoadherence on anti-CD34 monoclonal anti­(mAb)-coated plates, as previously described [15]. Purity

isolated CD34+ cells ranged from 80 to 96% by this od. Peripheral blood CD34 + cells from granulocyte

ulating factor (G-CSF)-mobilized normal donors provided by Dr. Scott Rowley from Fred Hutchinson

Research Center (Seattle, WA) following approved pro­Fetal liver cells were prepared from samples obtained

the Central Laboratory for Human Embryology, Depart­of Pediatrics, University of Washington. Nucleated fetal cells were used after the removal of erythroid cells

and red cells) by direct immunoadherence using anti-erythroid mAb 23.6 [16].

anti-CD34 antibodies were used: QBEND/10 (AMAC, ME); HPCA-2 (Becton Dickinson Immunocytome­

Jose, CA); and mAb 12.8 [17], kindly donated by Dr. FHCRC. Two antiplatelet antibodies against

(CD41) were used: 13.1 (kindly provided by Dr. D. and Tab [18] (a generous gift from Dr. Roger McEver,

of Oklahoma, Oklahoma City, OK). A directly con-1-FITC was purchased from Immunotech (Miami,

purchased antibodies included FITC-conjugated 'z antimouse IgG secondary antibody (American La Mirada, CA), biotinylated secondary antibody Foster City, CA, or Tago, Burlingame, CA), irrele­of the same isotype as test antibodies (Caltag Labo­

South San Francisco, CA), anti-CD45 RA (AMAC), · (PE)-conjugated streptavidin (Biomeda).

antibodies (anti-'{, 51.7; anti-[3, 16.2 and [337) were in our laboratory and have been described previous­

Anti-glycophorin A-PE conjugated to PE was pur­Dako (Carpinteria, CA).

labeling and FACS sorting were assayed for the presence of develop­by indirect live cell immunofluorescence

anti-CD41 antibodies (13.1, Tab), followed by antimouse IgG. Analysis was carried out with

microscopy (Zeiss Universal) or by flow cytometry FACStar; Cell Analysis Facility, Department

, University of Washington). Purified bone were labeled with anti-CD34 (QBEND/10 or

-CD45RA (all directly conjugated), and subsets +/CD45RN or CD34+/CD45RK were sorted

con~itions using appropriate positive and nega­Tnpie-labeling of purified CD34+ cells with anti­

,. anti-CD41-FITC (13.1), and anti-CD45RA-done for sorting into CD41+/CD45RA-,

' or CD41+/CD45RA+ subsets of CD34+ cells. sorting was verified by examination of each pop-

661

ulation with fluorescence microscopy. Purified subsets were grown in plasma clot and methylcellulose clonal cultures and in suspension cultures using the combinations of cytokines described in the text. Single cells from the different subsets were. also deposited (by FACS) on 96-well plates containing medmm and cytokines. Clonal growth from single-cell wells were double-labeled with antiglycophorin A-PE and anti­CD41- FITC between days 10 and 19.

Immunocytochemistry

For immunocytochemistry, either plasma clot or cytospin cell preparations were used. These were fixed at days 6-7 and 12-13 with pH 6.5 Histochoice (Amresco, Solon, OH) and stained at room temperature in the following sequence: anti­CD41 antibody; biotinylated goat Fab' 2 antimouse IgG (Tago, Burlingame, CA); streptavidin-conjugated alkaline phos­phatase (Vector Laboratories, Burlingame, CA); and alkaline phosphatase substrate kit I (Vector® Red; Vector Laboratories).

Suspension cultures Liquid cultures of CD34+ purified cells from marrow or from fetal liver were carried out for up to 3 weeks. CD34+ purified cells were inoculated at 104-105/mL. Medium for suspension cultures consisted of IMDM with 1 o/o bovine serum albumin (BSA), 10% normal human plasma or serum, 10-4 M 2-mer­captoethanol (Eastman Kodak, Rochester, NY), and a comple­ment of desired cytokines. IL-3 (Genetics Institute, Cam­bridge, MA) was used at 10 U/mL, and KL (SCF, Amgen, Thousand Oaks, CA) was used at 50 ng/mL. Epo (Genetics Institute, Cambridge, MA) was used at concentrations from 0.2 to 10 U/mL. Human recombinant Tpo was used at 1200 units/mL (optimal concentration determined in preliminary experiments; 50 units was defined as the amount supporting half-maximal proliferation in a BaF3/mpl MIT assay) [20] or at 15 ng of recombinant protein per mL (R&D Systems, Min­neapolis, MN). Cultures were fed every 2-3 days by replacing 80% of the medium, diluting the suspension, or removing cells when necessary to keep the cell concentration below 106/mL. Aliquots were removed at days 6-7 and 12-13 for replating and evaluation.

Semisolid assays

Cells were cultured using the methylcellulose method to evalu­ate BFU-E, CFU-GM, and CFU-Mix (mixed erythroid/myeloid/ macrophage) colonies and the plasma clot technique to evalu­ate either CFU-E or colonies derived from CFU-Mk. Methylcel­lulose medium was as previously described [14]; plasma clot composition was adopted from Mazur and South [20] for CFU­Mk and included 1 o/o BSA, 10% normal human serum, 10-4 M 2-mercaptoethanol, selected combinations of cytokines, 10% bovine citrated plasma, thrombin 0.25 U/mL, fibrinogen 500 ]lg/mL, and 2 ]1M CaCl2 • All cultures were set up in duplicate or triplicate plates, incubated in a fully humidified atmosphere with 5% C02 in air, and studied after 1 or 2 weeks. Mk colonies were identified by immunohistochemical labeling with anti­CD41 antibody. A CFU-Mk colony was defined as a cluster of three or more CD41 + cells. Other colonies were identified by morphologic criteria, direct microscopic observation of live cul­tures, or staining with benzidine and hematoxylin (fixed slide preparations of flattened plasma clots).

662

I'

..

' .

Experimental Hematology vol. 24 119961

CD::l-1 POSlfiVf Gl\ 1 1:

d 'th anti· . a stain) B FACS analysis of C034+ cells labele WI

Fig. 1. A. Purified bone marrow C034+ cells at day 2 of cul~ure (Giems n . C~45RK cells; only 0.07% of the cells were C041 and anti -C045RA. Note that virtually all C041 cell_s are amo g of Tpo alone. Usually over 70% of the cells are C041./C045RA•. C. Oay-12 C034+ cells in suspension cultures_m the pr~senceltures in the presence of Epo+Tpo. The only pop-

. 0 0 12 C034+ cells m suspens1on cu from C041+ at different stages of maturation. • ay- . II (G' msa and benzidine stain). E. Plasma clot cultures 1)

d II d megakaryocytic ce s 1e t' (04 · Ulations present are erythroblasts, re ce s, an t (CFU Mk colonies labeled by an I· . d k ytic colonies were presen · ' the C041./C045RK subset. Only erythroid an mega aryoc F. Putative bipotent E+Mk-derived colony in plasma clot cultures.

Th Papayannopoulou et al.: Epo and Tpo Synergy

Table 1. BM-CD34+ cell suspension cultures: increase in both CD41 + cells and g lobin• cells with Epo+ Tpo combination

Cytokine Nucleated cells C041 + cells Globin• cells x105 x105 x105

FL-MC Tpo 10.0 3.5 ± 0.24 0 Epo 130.0 3.0 ± 0.4 11.1 ± 2.1 Epo+Tpo 195.0 16.0 ± 2.3' 160.0 ± 3.1'

BM-CD34• Tpo 5.0 2.4 0.29 Epo 70.0 0.17 ± 0.08 64.0 ± 1.0 Epo+Tpo 106.0 8.3 ± 0.12' 111.0 ± 3.0'

BM-CD34• Tpo 10.0 Epo 19.0 0.6 ± 0.07 11 .0 ± 2.0 Epo+Tpo 80.0 9.9 ± 0.53' 32.0 ± 4.1'

PB-CD34+ Tpo 4.2 2.13 ± 0.36 0 Epo 0.93 0.02 ± 0.002 0.524 ± 0.04 Epo+Tpo 5.5 2.4 ± 0.33 0.69 ± 0.07

PB-CD34+ cells are from a G-CSF-mobilized normal donor, and evaluation was done at day 7. In all other experiments, assessment of C041 + and globin• cells was done at day 12 in culture, where maximal differences are observed. 'p < 0.01.

Statistical analysis Statistically significant differences among sets of samples were determined by Mann-Whitney test.

Results

Generation of CD41+ cells in suspension cultures of CD34+ cells rith the Epo+Tpo combination The generation of Gpiib-IIIa (CD41) positive cells from bone marrow CD34+ cells during culture in the presence of Epo+Tpo was compared to cultures with each cytokine (Epo or Tpo) alone. The cumulative number of CD41+ cells up to 2 weeks was calculated from the proportion of CD41 • cells and the total number of nucleated cells at each point (positive cells, usually between 1 and 4% at the first day of culture, were screened with the fluorescence microscope to exclude positivity from platelet adherence to cells). When only Tpo was present, there was an initial decline (by the end of the first week) in the total nucleated cell number but a progressive increase in the frequency of CD41 + cells, so that by 2 weeks in culture the cell population consisted of more than 70% CD41• cells (Fig. 1C). The majority of these cells were large­over 20 micron s in diameter- with a lobulated nuclear appearance and cytoplasmic maturation of varying degree. In the presence of Epo alone, CD41 + cells present at the begin­:: of the culture quickly declined. In the presence of both

Tpo, however, CD41+ cells accumulated in much higher ~Umbers than with Tpo alone (Table 1 and Fig. 1D). This CO+Tpo synergy was observed at all doses of Epo tried

rA..'l-10 U/mL, data not shown). To compare the magnitude er the Epo+ Tpo synergy to that of Tpo combined with anoth­lll:okme (IL-3 or KL), we made similar assessments of accu­~ted CD41• cells in Tpo+IL-3 and Tpo+KL cultures. The caJl ts :Vith IL-3+ Tpo in three experiments were not statisti­J ~Ifferent from those with Epo+ Tpo (data not shown), lttal r th combinations were better than Tpo+KL (Fig. 2) . In ~Ver samples, however, the combination of Tpo+Epo

ed to be superior to that of Tpo+IL-3 (Fig. 2) .

Lo Gl .c E :::J z

IL-3 Epo KL

Tpo: + + + +

BM-CD34 + Cells

IL-3 Epo

+ +

FL-MNC

Fig. 2. Generation of C041 + cells from bone marrow C034+ cells in suspension cultures in which Tpo was used either alone or in combination with one other cytokine (IL-3, KL, or Epo). The Epo+ Tpo combination gives similar results to IL-3+ Tpo in BM samples, but it appeared superior in fetal liver samples (fetal mononuclear cells from which the nucleated erythroid cells were removed by immunoad­herence were used in this experiment).

Generation of globin• cells in suspension cultures of CD34+ cells with the Epo+Tpo combination

663

In addition to the generation of CD41 + cells, we also wanted to examine whether Tpo had any enhancing effect on the dif­ferentiative action of Epo in the same cultures. For this pur­pose, we assessed the generation of globin• cells (by immunofluorescence of fixed smears). By the end of 2 weeks, no globin• cells were observed when only Tpo was present in these cultures. In contrast, as seen in Table 1, the generation of globin• cells was significantly higher when the two cytokines were combined compared to Epo alone. To test whether Tpo enhances the differentiative action of Epo in vitro, we compared the proportion of globin• cells not only in the Epo/Tpo combination, but also in the combinations of Tpo with other cytokines, especially KL, since we had previ­ously shown that globin• cells (presumably proerythroblasts) are present in these cultures in the absence of Epo [14]. As seen in Table 2, the relative proportion of globin• cells present with either KL alone or IL-3 alone does not change signifi­cantly in the presence of Tpo. In contrast, the proportion of globin• cells with the Epo+Tpo combination is much lower than with Epo alone, although the total numbers are higher. These results suggest that Tpo's enhancing effects are more proliferative than differentiative. Indeed, when the effects of the Epo+ Tpo combination were assessed at the progenitor level (BFU-E, CFU-E, and CFU-GM), we found that a combina­tion of two cytokines had significant effects on the expansion

'I I

664

Table 2. BM-CD34+ cells in suspension cultures: accumulation o; globin+ cells by day 7 in culture in the presence of IL-3, KL, or Ep with and without Tpo

Globin+ cells

Cytokine• Percent No. x105

Tpo 0 0 KL 3.6 ± 0.005 13.7 ± 2.0 KL+Tpo 2.9 ± O.ol 25.4 ± 10.6 IL-3 0.9 ± 0.002 7.1±12.3 IL-3+Tpo 1.3 ± .003 17.7±4.1 Epo 56.3 ± .05 52.4 ± 4.3 Epo+Tpo 12.5 ± 0.02 69.0 ± 7.9

•All cytokine combinations were initiated with the same inoculum from the same pool of CD34+ cells.

of these progenitors in culture, but only if Tpo was ~resent from the outset of culture. Both early and late er:thrmd pr~­genitors were significantly amplified after 7 days m culture I~ the presence of Epo+ Tpo compared to Epo alo~e (Table 3). It IS

of note that significant differences (p < 0.01) mother types of progenitors such as CFU-GM, were found in three of four experiment; with Epo+ Tpo. In fact, the effect of Tpo on o_ther progenitors, both erythroid and nonerythroid, wa~ not umque to the Epo+Tpo combination and was observed With IL-3+Tpo and KL+ Tpo (Table 3). Enhancement of CFU-E, h~wever, was seen only with the Epo/Tpo combination and With none of the other combinations. These experiments clearly sho:Ved that Tpo affected late stages of erythropo_iesis in vitro mamly through proliferative effects at the progemtor level.

Enrichment of erythroid and megakaryocytic progenitors in specific

CD34+ subsets . . To analyze the phenotypic features of erythroid progemt_ors on which Tpo exerted its effects, we atte_mpted to ennc~ either erythroid or megakaryocytic progemtors from _CD34 cells. Previous experiments [22] showed that erythrm~ pro­genitors can be greatly enriched in the CD34~/CD45RA_ sub­set and similar observations have been descnbed prevwusly [23 24]. The results of seven experiments showing the degree of BFU-E enrichment in these subsets are shown in Figure 3. A wide spectrum of BFU-E sizes was observed. Thus, the absence of CD45RA phosphatase does not discriminate between early and late erythroid progenitors. To test whether t_h_ese ery­throid progenitors display any megakaryocyte-specifiC mar~­ers we used two antibodies, anti-CD4I and a polyclonal antl­m~l receptor antibody. First, we examined the presence of CD41+ cells among the CD34+/CD45RA- subset. We ~ound that the great majority of CD4I+ cells were present m ~he 45RA- subset, compared to 45RA+ or the unsorted populatwn (Fig. IB). Less than 6%, 1. 7%, and <1% were_ observe~ to co_ex­press CD4I and CD45RA in thre_e expenments _m which CD34+ cells were labeled with antl-CD4I and anti-CD45RA antibodies (Table 4). These results predicted that megakaryo­cytic progenitors could be found in the 45RA- rather tha~ the 45RA+ subset. Indeed, when we first tested the 34+/45RA sub­set for the presence of megakaryocytic progenitors, more than 80% were found in the CD45RA- subset (data not shown). In subsequent experiments, therefore, we specifically sorted,

Experimental Hematology vol. 24 (

Table 3. CD34+ cell suspension cultu~es: expansi.on in progenit~r cells (CFU-GM, BFU-E, CFU-E) by day 7 1n culture With Epo+ Tpo corl'l· bination

Cytokine CFU-E x 103 BFU-E x 103

FL-MC Epo 8.7 ± 1.3 2.0 ± 0.4 Epo+Tpo 17.2 ± 1.2b 9.9 ± 0.8b

BM-CD34+ Epo 1.8 16.0±1.9 Epo+Tpo 3.3 ± 0.2b 37.0 ± 4.9b

BM-CD34+ Epo 5.1 ± 0.5 4.0± 0.6 Epo+Tpo 9.4 ± 0.7b 16.8 ± 4.2b

BM-CD34+ Epo 4.9 ± 1.2 26.4± 1.6 Epo+Tpo 15.4 ± 2.8b 66.2 ± 4.6b

PB-CD34+ Epo 1.3 ± 0.07 9.2 ± 0.2 Epo+Tpo 2.9 ± 0.4b 45.8 ± 4.9b Epo+Tpo

(delayed)" 2.3 ± 0.1 13.3 ± 1.6 IL-3 0 29.2 ± 4.0 IL-3+Tpo 0 70.1 ± 4.4b KL 1.6 ± 0.2 32.2 ± 1.7 KL+Tpo 1.2 ± 0.07 54.8 ± 5.5b

IL-3 and KL with and without Tpo were included in experiment for comparison. •Addition of Tpo was delayed by 3 days. bp < 0.01.

among CD34+ cells, subsets that were 41+/45~-, 4 or 41-/45RK and subjected them to clonogemc suspension cultures. In three such experiments, we _ virtually all BFU-E again resided in the CD45RA whether they were CD4I+ or CD4I- (Table 4). Of BFU-E present in the CD41+/CD45RA- subset were late or mature type, since less than 10% of the total were large bursts. In contrast, over a third of_ the bursts present in the CD41-/CD45RK subset_ m ment and over 40% in another were categonzed as extra-large bursts. Thus the presence of CD:l 45RA- cells, which include virtually all erythrmd seems to segregate the early from the late BFU-E. subsets were examined for the presence of C genitors, it was found that +CFU-}vfk-derived mainly present in the CD4I /45RA subset, but a number were also found in the CD4r/45RA-. As with BFU-E, the majority of CFU-Mk colomes CD4r subset were larger (>seven cells per colony) present in the CD41+ subset (few colonies_wi_th The fact that earlier CFU-Mk and more pnmltlve present in the CD4I- subset was secured by suspension cultures of these subsets. In these fication of CFU-Mk or of BFU-E was evaluated replatings after the first week in culture, and th; tion of CD4I + and globin+ cells was assessed ove weeks in culture. As seen in Table 5, by the end of week in culture, more CD4I+ and globin+ cells from the CD4r subset than the CD4I+ subset, that the former subset had 0% CD4I+ cells by CD4I-/CD45RK) when the culture was more, there was substantial BFU-E expansion from the CD4I- subset when evaluated at day 7 culture compared to minimal or no expan CD4I+ cells (0.7- to 2.4-fold). Similarly,

Th Papayannopoulou et al.: Epo and Tpo Synergy

BFU-E% CFU-GM%

30 20 10 0 10 20

Fig. 3. Colony yield (BFU-E or CFU-GM) from subsets of CD34+ cells that are either positive or negative for CD45RA. Length of each bar indicates number of colonies expressed as percent of cells plated. Note the reciprocal enrichment of the CD45RK subset in erythroid colonies and of the CD45RA+ subset in CFU-GM colonies.

------------------------------------------

30

+subset showed many scattered megakaryocytes but few erived colonies. By contrast, many CFU-Mk

were generated when cells from the CD4I- subset in for 7 days were replated (Table 5). also did two experiments in which we made use of an

polyclonal antibody (kindly provided by Immunex). 1 +-selected cells, erythroid progenitors were found

the positive (Mplhi) and negative (Mpl10) fractions. as with CD41, the c-Mplhi_selected subsets were

enriched in BFU-E compared with CFU-GM. In 'l"'"H'""'·'' for example, the Mplhi subset had BFU-E =

5% and GM = 0.3 ± 0.02% (E:GM ratio ~8:1). The had BFU-E = 3.0 ± 0.7% and GM = 8.1 ± 0.8%

1:2.6, a reversal of the Mplhi E:GM ratio). In a sec­the ratio was also high in the Mplhi subset:

1.77% and CFU-GM = 0.03% (E:GM = 57:I). These that erythroid progenitors found among both

c-Mpl+ subsets share some phenotypic markers progenitors.

amd frequency of bipotent erythroid/megakaryocytic

experiments described in the previous section, it erythroid progenitors were found in the CD41 + -; subsets, and that these subsets contained a

number of CFU-Mk progenitors. Since there were ( ~ 15-fold) more erythroid progenitors present in compared to megakaryocytic progenitors, it was an progenitors present could be bipotent ery-

<lK>Irv,-.r,,r;· progenitors. It was not clear, however, megakaryocytic progenitors recovered in specific in fact bipotent rather than pure megakaryocytic and these were copurified in the CD4I+/CD45RA----"""''UO" the presence and frequency of bipotent

aryocytic progenitors in the CD4I + and Wt initiated plasma clot cultures from these inocula (500-1000 cells/mL) or did single-cell

665

Table 4. Hematopoietic progenitors (BFU-E, CFU-GM, and CFU-Mk) present in different CD34+ subsets

Subset BFU-E CFU-GM CFU-Mk Experiment (41/45RA) 0/o % %

+/- (20%) 29.9 ± 3.6 0.6 ± O.Ql 1.6 ± 0.5 -1- (48.6%) 13.3 ± 3.5 3.5 ± 0.6 0.3 ± 0.1 +!+ (6.1%) 0.58 ± 0.04 12.3 ± 1.8 0.2 ± 0.1 2a +/- (17.7%) 24.36 ± 3.1 0 3.6b

-1- (69.4%) 9.53 ± 1.0 3.2 ± 0.1 1.5b 3a +/- (14.45%) 37.9 ± 2.3 0.4± 0.4 7.7 ± 1.3 -1- (48.74%) 29.0 ± 3.4 2.0 ± 0.2 0.8 ± 0.3

Numbers in parentheses indicate the representation of each subset among the total CD34+ population. •The proportions of+!+ in experiments 2 and 3 were 1.7 and 0.07%, respectively. bSingle plate count.

--------------------- --- ---------------·----------------~-

plating in 96-well plates using the automatic FACS sorting device. In the plasma clots, three categories of colonies were scored: pure erythroid, pure megakaryocytic, and bipotent erythroid/megakaryocytic. The plasma clots were stained with anti-CD4I by immunohistochemical means and counter­stained to reveal the characteristic morphology of erythroid colonies (Fig. IE). The frequencies of erythroid colonies in these plasma clot cultures were about I4.6% (Table 6) and 22% (data not shown) in two experiments, respectively; the frequencies of pure megakaryocytic colonies were about 1.6% (Table 6) and 7.7% in the 4I+/45RA- cells (Table 4). As both pure erythroid and pure Mk colonies were enriched in these plasma clots, there was ample opportunity for an overlap between these two types of colonies in plasma clots. This was observed several times (Fig. IE), but CD4I +cells also appeared to sprout from some erythroid colonies rather than represent­ing overlapping unipotent colonies (Fig. IF). These colonies were scored as putative bipotent E+Mk colonies. The frequen­cies of these colonies in the plasma clot culture were 0.3% (Table 6) or <O.OI in experiment 2 in plasma clot cultures of 4I+/45RA- cells (data not shown). To secure the single-cell ori­gin of this putative bipotent progenitor, we carried out single­cell cultures (in medium containing IL-3, KL+G-CSF+ Tpo, and Epo) and scored single-cell growths for presence of erythroid (glycophorin A+) and megakaryocytic (CD4I +) cells. Wells with more than 100-200 cells each were individually double­labeled with anti-CD4I and anti-glycophorin antibodies directly conjugated with different fluorochromes and exam­ined by fluorescence microscopy. (In preliminary experi­ments, we attempted cytospin preparations from wells with enough cells, but the yield of evaluable samples was low.) When all cells were positive with glycophorin A, clonal growths were scored as pure erythroid. When cells both posi­tive and negative for glycophorin A were present, wells were scored as E+Mix. On some occasions, these mixed colonies had CD41+ cells. Wells that had only glycophorin+ and CD4I+ cells were scored as E+Mk. The overall plating efficiencies and proportions of pure erythroid or bipotent E+Mk colonies and other types of colonies in three experiments are shown in Table 6. It was of interest that although the plating efficiency was not very different between the CD4I+ and CD41- subsets

666 Experimental Hematology vol. 24

1 • f CD41+ cells and globin+ cells by 2 weeks in culture, and BFU-t and Table 5. Suspension cultures initiated from CD34+ subsets: accumu atron o <

CFU-Mk expansion after 1 week in culture

Experiment

2

3

Subset (41/45RA)

+/­-/-

Cell no.

108.0 192.0

+/- 4.8 -1- 10.0 +/+ 2.6

+/- 6.9 -1- 58.0

Cells x105

6.1 ± 0.74 22.0 ± 1.5

0.31 ± 0.05 1.3 ± 0.3

0.05 ± O.ol

0.68 ± 0.08 1.67 ± 0.26

Globin+

[58.0 ± 6.2]" [23.0 ± 1.6]"

2.6 ± 0.74 3.4 ± 0.5 0.5 ± 0.07

5.34 ± 0.13 24.7 ± 0.2

BFU-E

2.4 X

29.2 X

0.7 X

5.4 X

0.71 X

8.8 X

Expansion

CFU-Mk

11.7 X

25.4 X

1.36 X

2.65 X

1.6 X

105.0 X

• IL 3 IL 6 KL and Tpo but not Epo. The initial inoculum in experi-"ln this experiment the cytokines included in s~spensron ":ere . .' . ; 3 ~ x 104 cells· it was the same for all the subset cultures. ment 1 was 3 x 104 cells, in experiment 2, 1 x 10 cells, and rn experrmen , ,

(56 and 36o/o in 41+ compared to 53 and_ 42o/o in 4r),_ there were more cells in each well in the CD41 subset than m the CD41+ subset. Because of this, more wells from CD4:- were evaluable. In general, wells in which both eryt?rmd and CD41 + cells (derived from bipotent E+Mk progemtors) were observed were much less frequent (~16-fold) than the pu~e erythroid. More bipotent E+Mk progenitors :vere obs~rved m the CD41+ population (6.2 and 5.4o/o) than m CD41 (0 _and 2.8%), indicating that this bipotent progenitor comes mt~ play more frequently after the appearance of the C~41 anti­gen. In all bipotent E+Mk cells, there was a predommance of erythroid cells compared to CD41 + cells (the latter ranged from 1 to 15o/o of total cells). More of the clonal growths _co~­taining megakaryocytic cells (CD41+) were part of CFU-Mix m the CD4r subset, in contrast to the CD41+ subset.

Discussion . . The existence of a close relationship between erythropOiesis and megakaryocytopoiesis rests on numerous experime~tal observations spanning more than 15 years. Thus, ~rogemtor cells from both lineages share common features, either tran-

E-Mix) derived colonies observed in

Plasma clot cultures

Experiment Subset E%

scription factors (e.g., GATA-1, ets, NFE2) [25,26] or antigens or receptors (e.g., presen~e o_f Ep? rece megakaryocytic cells) [14], and certam signalmg (e.g., JAK2 kinase) [27]. Also, it is not a random nnPnnm

that virtually all erythroleukemia cell lines that have made today also display megakaryocytic features Papayannopoulou, unpublished data]. Furthermore, it was shown that Gplla-targeted suicide genes influence both erythroid and megakaryocytic cells [35]. Erythropoietin was known to enhance CFU-Mk growth [36], a finding confirmed when . . became available [11]. Additional in vivo data either m after treatment with high doses of Epo [37] or in mals [38] also show effects in both lineages. Of thropoietin and thrombopoietin share structural and qualitatively similar biologic acti~i~ies ~7].

Our present studies, showing a ~Iduectwn~l +effect on the accumulation of both CD41 and globm cells pension cultures in the presen~e.of E?o, extend these observations. To explain the biduectwnal effect of Tpo generation of CD41 + and the increase of globin+

clot cultures and cultures of different CD34'

Mk% E+Mk%

41+/45RA- 2012" 4r/45RA-1251"

14.6 ± 0.4 9.96 ± 0.3

1.6 ± 0.3 0.3 ± 0.07

0.3 ± 0.07 0.075 ± 0.02

Single-cell cultures

Experiment Subset Wells with growth/total wells

41+/45RA- 109/192 4r/45RA- 101/192

2 41+/45RA- 164/485 41-/45RA- 206/485

3' 34+/45RA- 124/192 4' 34+/45RA- 95/290

Wells labeled

48 68 55 71 35 30

E

38 40 41 49 13 11

E-Mix

5

Mk

0 0

0 5 2

k b' t t throid/megakaryocytic colonies; E = pure erythroid colonies; Mk = pure mega~aryocytic colonies; E+M = 1P0 en ery mixed colonies with erythroid and nonerythrord components. "Total colonies scored. +

bNumbers in parentheses indicate E+Mix colonies including CD41 cells. I I' bl than the live cell 'Evaluated by immunocytochemical means which, because of cell losses, were ess re ra e used in experiments 1 and 2.

Papayannopoulou et al.: Epo and Tpo Synergy

667

--acc•rwp of Epo, we entertained the following possibilities: that Tpo enhances the differentiative action of Epo; sec­that Tpo has only a proliferative effect on all erythroid

-~~a,,,,.,..,,<. and third, that Tpo mainly affects bipotent ery-. megakaryocytic progenitors (E+Mk) present in these

. Combinations of two or three of these possibilities also considered. We found that enhancement of the dif­

tive for these two molecules were found) were indeed present among both CD41+ and CD4r subsets, albeit with different frequencies and different characteristics. Truly bipotent E+Mk progenitors were more frequent in the CD41 + subset. The

action of Epo by Tpo was unlikely on the basis of data presented in Table 2. Tpo alone generates no globin+ and does not enhance the proportion of globin+ cells

in the presence of KL (or IL-3) but in the absence of . Most important, the proportion of globin+

with the Epo+Tpo combination was much lower than Epo only, and a delayed addition of Tpo abrogated much

effect. When we tested the effect of Tpo on the expan-of erythroid progenitors (CFU-E and BFU-E), however, we

a significant enhancement in the presence of Tpo. data confirm previous in vitro and in vivo data in the

model [11] and recent observations with human cells Further, we were interested to know whether the ery­

progenitors on which Tpo exerts its effect display markers, specifically c-Mpl or CD41. When

CD34+ subsets (34+/45RA-) that were highly in BFU-E with anti-CD41, we found that virtually all

cells were included in this subset (Fig. 1B). As expected, were also enriched for megakaryocytic progenitors. As

of progenitors present in these CD41 + subsets erythroid, with characteristics of late BFU-E, this

formally demonstrates that these erythroid pro­display CD41 on their surface. Megakaryocytic pro­

highly enriched in the same subset ~~·-rvnn-), were much less frequent than erythroid

and were also mostly of late type, giving rise to colonies than those present in the CD41- subset. Sev­

pieces of evidence support the notion that late and late megakaryocytic progenitors were present in

1+ compared to CD41- subset. Suspension cultures with CD41 + or CD41- cells showed greater prolifera­

of total nucleated cells, higher accumulation and globin+ cells, and greater expansion of BFU-E

-Mk in the CD41- subset compared to CD41 + (Table fact that rather late CFU-Mk are present among

cells was also suggested by some previous stud-independent experiments in which anti-Mpl was of anti-CD41 showed that Mplhi/45RK cells were

in BFU-E. Unlike anti-CD41, anti-Mpl anti­a continuing spectrum of positivity without between positive and negative cells. Whether

---..•. ,~ .. depends on the specific Abused or reflects spectrum of Mpl presence among progenitors is

our data taken together strongly sug­progenitors share common phenotypic

CFU-Mk progenitors.

for the presence and frequency of bipo­nu~gaka.rvc>cvti' progenitors, we analyzed sin­

for the presence of cells with erythroid and markers originating from a single cell. In

we were able to demonstrate that pro­giving rise to erythroid only (glycophorin N) and

only (CD41+) progeny (no cells doubly posi-

CD41 + subset also contained pure erythroid, pure megakary­ocytic, and rare or no CFU-Mix that included megakaryocytes, even when analysis was carried out after 17 days in culture. In contrast, the CD4r subset contained fewer true E+Mk and more CFU-Mix with or without Mk and included earlier pro­genitors with high proliferative capacity. Since only single-cell growths of more than 100-200 cells were analyzed, a bias directed against the analysis of progenitors giving rise to small clonal growths is undoubtedly present. This bias likely con­cerns late pure CFU-Mk or late pure erythroid progenitors, and it is unlikely that it is directed toward E+Mk progenitors.

Because the frequency of truly bipotent E+Mk progenitors was very low compared to pure erythroid, and because a significant proportion of the erythroid progenitors displayed Mpl, we believe that the proliferative effect of Tpo is directed toward these erythroid progenitors. Both megakaryocytic and erythroid cells were increased in the same cultures in vitro, and the progeny of both lineages was increased in vivo with­out any apparent competition. This fact can be explained by the action of Tpo on independent progenitors, rather than on the same progenitor (bipotent E+Mk). On the basis of our in vitro data, one can suggest that all three types of progenitors (pure erythroid, pure CFU-Mk, and E+Mk) exist in vivo and are targets for Tpo's action.

The notion that bipotent erythroid/megakaryocytic pro­genitors exist is not new [41-45], but attempts to demonstrate their frequency or enrichment under a phenotypic class of progenitors have not been made previously. Furthermore, it was unclear whether this bipotent progenitor was just an example of a progressive loss of potentialities from a pluripo­tent cell or something beyond this concept. Our present find­ings, showing that the bipotent E+Mk progenitor copurifies with the subset (CD41 +) that contains less multi potent pro­genitors, are certainly compatible with the view that such a progenitor is located downstream of the pluripotent cell in the hematopoietic hierarchy. However, its special relationship with erythroid progenitors rather than a random association with both erythroid and granulocytic progenitors (since no bipotent megakaryocytic/myeloid progenitors were detected) supports the notion that an E+Mk bipotent progenitor has a distinct place in hematopoietic differentiation. In fact, on the basis of in vivo data suggesting competition between erythro­poiesis and megakaryocytopoiesis [46], some investigators suggested that this progenitor may be the most frequent one present in vivo, especially before the two lineages diverge. Our data do not support such a view. If all erythroid and megakaryocytic cells have to go through the stage of a bipo­tent progenitor, competition between the two lineages may be expected, and the higher levels of one cytokine (Epo or Tpo) present at any given time may drive the cells toward one or the other pathway. Alternatively, as our data suggest, ery­throid and megakaryocytic cells can be generated from both bipotent E+Mk progenitors and unipotent progenitors, and the action of each lineage-specific cytokine (Epo or Tpo) is addressed to both pure and bipotent progenitors. Such a view reconciles the contemporaneous enhancement of both lin-

668

eages by either Epo or Tpo treatments in vivo under certain

circumstances [12, 13,37].

(c-mplligand) acts synergistically with erythropoietin stem cell factor to enhance murine megakaryocyte '-v,,vnv 1

growth and increases megakaryocyte ploidy in Finally, if pure erythroid and pure CFU-Mk progenitors

both express Mpl, it will be intriguing to know how the sig­naling pathways in one case (Mpt/erythroid progenitors) cause only proliferative effects, and in the second (Mpl+/CFU­Mk), induce both proliferative and differentiative effects. Per­haps the different regimen of transcriptional factors present in these two types of progenitors dictates the signaling path-

Blood 85:1719 12. Kaushansky K, Broudy VC, Grossman A, Humes],

Hong P, Sprugel KH, Baily MC,

ways leading to these changes.

References 1. Ogawa M (1993) Differentiation and proliferation of

hemopoietic stem cells. Blood 81:2844 2. de Sauvage FJ, Hass PE, Spencer SD, Malloy BE, Gurney

AL, Spencer SA, Darbonnie WC, Henzel WJ, Wong SC, Kuang WJ, Oles KJ, Hultgren B, Solbert LA, Goedde! DV, Eaton DL (1994) Stimulation of megakaryocytopoiesis and thrombopoiesis by the c-Mplligand. Nature 369:533

3. Lok S, Kaushansky K, Holly RD, Kuijper JL, Lofton-Day CE, Oort P], Grant FJ, Heipel MD, Burhead SK, Kramer JM, Bell LA, Sprecher CA, Blumberg H, johnson R, Prunkard D, Ching AFT, Mathewes SL, Bailey MC, Forstrom JW, Buddie MM, Osborn SG, Evans SJ, Sheppard PO, Presnell SR, O'Hara PJ, Hagen FS, Roth GJ, Foster DC (1994) Murine thrombopoietin: Expression cloning, eDNA sequence and stimulation of platelet production in vivo. Nature 369:565

4. Wendling F, Maraskovsky E, Debili N, Florindo C, Teepe M, Tieux M, Methia N, Breton-Gorius J, Cosman D, Vainchenker W (1994) The Mplligand is a humoral regu­lator of megakaryocytopoiesis. Nature 369:571

5. Miyazaki H, Kato T, Ogami K, Iwamatsu A, Shimada Y, Souma Y, Akahori H, Horie K, Kokubio A, Kudo Y, Maeda E, Kawamura K, Sudo T (1994) Isolation and cloning of a novel human thrombopoietin factor. Exp Hematol 22:838

6. Bartley TD, Bogenberger J, Hunt P, Li YS, Lu HS, Martin F, Chang MS, Sarna! B, Nichol JL, SwiftS, Johnson MJ, Hsu RY, Parker VP, Suggs S, Skrine JD, Merewether LA, Logston C, Hsu E, Hokom MM, Hornkohl A, Choi E, Pangelinan M, Sun Y, Mar V, McNinch], Simonet L, Jacobsen F, Xie C, Shutter J, Chute H, Basu R, Selander L, Trollinger D, Sieu L, Padilla D, Trail G, Elliott G, Izumi R, Covey T, Crouse J, Garcia A, Xu W, Del Castillo J, Biron J, Cole S, Hu MCT, Pacifici R, Pouting I, Saris C, Wen D, Yung YP, Lin H, Bosselman RA (1994) Identification and cloning of a megakaryocyte growth and development factor that is a ligand for the cytokine receptor Mpl. Cell 77:1117

7. Kaushansky K (1995) Thrombopoietin: The primary regu­lator of platelet production. Blood 6:419

8. Broudy VC, Kaushansky K (1995) Thrombopoietin, the c­mpl ligand, is a major regulator of platelet production. J

Leuk Biol57:719 9. Gurney AL, Carver-Moore K, de Sauvage FJ, Moore MW

(1994) Thrombocytopenia in c-mpl-deficient mice. Science

265:1445 10. de Sauvage FJ, Luoh S, Carver-Moore K, Ryan A, Dowd M,

Eaton DL, Moore MW (1995) Deficiencies in early and late stages of megakaryocytopoiesis in Tpo-KO mice. Blood

84:255 [abstr] 11. Broudy VC, Lin NL, Kaushansky K (1995) Thrombopoietin

Forstrom J (1995) Thrombopoietin expands erythroid genitors, increases red cell production and enhances throid recovery following myelosuppressive

Clin Invest 96:1683 13. Fib be WE, Heemskerk DPM, Laterveer L, Pruijt JFM,

D, Kaushansky K, Willemze R (1995) Accelerated tution of platelets and erythrocytes after syngeneic plantation of bone marrow cells derived from bopoietin pretreated donor mice. Blood 86:3308

14. Fraser JK, Tan AS, Lin F-K, Berridge MV (1989) of a specific high-affinity binding site for ne1mopoletiil rat and mouse megakaryocytes. Exp Hematol17:10

15. Papayannopoulou T, Brice M, Blau T (1993) Kit synergy with interleukin-3 amplifies the independent, globin-synthesizing progeny of human burst-forming units-erythroid in suspension tures: Physiological implications. Blood 81:299

16. Das Gupta A, Samoszuk M, Papayannopoulou T, toyannopoulos G (1985) SFL 23.6: A monoclonal reactive with CFU-E, erythroblasts, and "r"cth•·rw•

Blood 66:522 17. Andrews RG, Singer JW, Bernstein ID (1986)

antibody 12.8 recognizes a 115-kd molecule both unipotent and multipotent hematopoietic forming cells and their precursors. Blood 67:842

18. McEver RP, Baezinger LN, Majerus PW (1980) and quantitation of the platelet membrane deficient in thrombasthenia using a monoclonal rna antibody. J Clin Invest 66:1311

19. Stamatoyannopoulos G, Farquhar M, Lindsley D, Papayannopoulou T, Nute PI (1983) Monoclonal ies specific for globin chains. Blood 61:530

20. Kaushansky K, Broudy VC, LinN, Jorgensen MJ, J, Fox N, Zucker-Franklin D, Lofton-Day C (1995) bopoietin, the Mpl ligand, is essential for full cyte development. Proc Natl Acad Sci USA 92:3234

21. Mazur EM, South K (1985) Human megakaryocyte stimulating factor in sera from aplastic dogs: cation, characterization, and determination of poietic cell lineage specificity. Exp Hematol 13:

22. Papayannopoulou Th, Brice M, Kaushansky K influence of Mpl-ligand on the development karyocytes from CD34+ cells isolated from ' peripheral blood and cord blood. Blood 84:324

23. Lansdorp PM, Sutherland HJ, Eaves CJ (1990) expression of CD45 isoforms on functional tions of CD34+ hemopoietic cells from human

row. J Exp Med 172:363 24. Bender JG, Unverzagt K, Walker DE, Lee

Williams S, Van Epps DE (1994) Phenotypic characterization of CD34+ cells from normal marrow, cord blood, peripheral blood, and peripheral blood from patients undergoing cell transplantation. Clin Immunol

Th Fapayannopoulou et al.: Epo and Tpo Synergy

Martin DIK, Zon LI, Mutter G Orkin SH (1990) E . f . ' xpresswn o an eryt~rmd transcription factor in megakaryocytic and mast celllmeages. Nature 344:344 R~me~ PH, Prandini MH, Joulin V, Mignotte M, Prenant V:,, Vam~henker G, Marguerie G, Uzan G (1990) Mega­karyo.cytic and erythrocytic lineages share specific tran­scnptwn factors. Nature 344:447 Ih~e JN (1995) Cytokine receptor signalling. Nature 37 I :591 Papayannopoulou T, Nakamoto B, Kurachi S, Tweeddale M, Messner H (1988) Surface antigenic profile and globin phenoty~e ~f two new human erythroleukemia lines: Charactenzatwn and interpretations. Blood 72:1029 Sato T, Fuse A, Eguchi M, Hayashi Y Sugr'ta K N k ,. , , a azawa S, Mmato K, Shima Y, Komori I, Sunami S Ok' y N k ,. H ( , rmoto ,

a ·aJrma 1987) Establishment and characterization of a megakaryoblastic cell line (CMK) from a Down's syn­drome patient with acute megakaryoblastic leukemia. Ex Hematol15:495 p Chiba S, Takaku F, Tange T, Shibuya K, Misawa C, Sasaki

Mlyagawa K, Yazaki Y, Hirai H (1991) Establishment erythroid differentiation of a cytokine-dependent

leukemic cell line F-36: A parental line requiring cyte-macrophage colony-stimulating factor or

-3, and a subline requiring erythropoietin 78:2261 .

DA, Gumucio DL, Brodsky I (1991) Granulocyte­uu'"'"w~•e c~lony-~timulating factor-dependent growth ~rythropmetm-mduced differentiation of a human

lme MB-02. Blood 78:2860 M, Morishima Y, Ohno R, Kato Y, Hirabayashi N, H, Sato H (1985) Establishment of a novel human

blastic leukemia cell line MEG-01 wr'th . Ph. ' , posr-

Jladelphia chromosome. Blood 66:1384 N, Yamamoto M, Fujita H, Miwa A, Hatake K,

T, Oka~o H, Katsube T, Fukumaki Y, Sassa S, Miura y Establishment and characterization of an erythro­-dependent subline, UT-7 /Epo, derived from leukemia cell lines, UT-7. Blood 82:456

T, T~nge T, Terasawa T, Chiba T, Kuwaki T, K, Prao Y-F, Miyazono K, Urabe A, Takaku F

Establishment and characterization of a unique cell line that proliferates dependently on GM-CSF

or erythropoietin. J Cell Physiol 140:323 ' Le Roux D, Roullot V, Schweitzer A, Berthier R,

669

Marguerie G (1995) Suppression of erythro-megakaryocy­t~p~resJs. and the induction of reversible thrombocytope­nia m mrce transgenic for the thymidine kinase gene tar­geted by the platelet glycoprotein a!Ib promoter J E Med 181:2141 · xp

36. Ishibashi T, Koziol JA Burstein SA (1987) H ' uman recombi-nant erythropoietin promotes differentiation of murine megakaryocytes in vitro. J Clin Invest 79:286

37. McDonal? TP, Cottrell MB, Clift RE, Cullen WC, Lin F-K (1987) Hrgh doses of recombinant erythropoietin stimu­late _Platelet production in mice. Exp Hematol 15:719

38. Sullrvan PS, Jackson CW, McDonald TP (1995) Castration decreases thrombocytopoiesis and testosterone restores platelet production in castrated Balb/c mice: Evidence that testosterone acts on a bipotential hematopoietic precursor cell. J Lab Clin Med 125:326

39. Kobayashi M, Laver JH, Kato T, Miyazaki H Ogawa M (1995) Recombinant human thrombopoietin (Mplligand) enhances proliferation of erythroid progenitors Blood 86:2494 .

40. Debili N, Issaad C, Masse JM Guichard J Katz A B t G · . ' , , re on-onus J, Vamchenker W (1992) Expression of CD34 and

platel~t ~lycoproteins during human megakaryocytic dif­ferentiatiOn. Blood 80:3022

41. Nicola NA, Johnson GR (1981) The production of commit­t~d hematopoietic colony-forming cells from multipoten­tial precursor cells in vivo. Blood 60:1019

42. McLeod DL, Shreeve MM, Axelrad AA (1980) Chromo­some marker evidence for the biopotentiality of BFU-E Blood 56:318 ·

43. Nishi N, Nakahata T, Koike K, Takagi M, Nagamura K, Akabane T (199_0) Induction of mixed erythroid-mega­karyoc~te colomes and bipotential blast cell colonies by recombmant human erythropoietin in serum-free culture Blood 76:1330 ·

44. Bellucci S, Han ZC, Pidard D, Caen JP (1992) Identification of a n~rmal human bone marrow cell population co­expressmg megakaryocytic and erythroid markers in cul­ture. Eur J Haematol 48:259

45. Vannucchi AM, Paoletti F, Grossi A (1994) A common megakaryocytic and erythrocytic precursor in murine ery­throleukemia (Friend) cells? Exp Hematol 22:110

46. McDonald TP, Clift RE, Cottrell MB (1992) Large, chronic doses of erythropoietin cause thrombocytopenia in mice Blood 80:352 .