I NVITRO Volume14, No. t, 1978 All rightsreserved@

D I F F E R E N T I A T I O N O F MURINE ERYTHROLEUKEMIC (FRIEND) C E L L S : AN IN V I T R O M O D E L O F E R Y T H R O P O I E S I S ' STUART H. ORKIN Division of Hematology-Oncology, Department of Medicine, Children's Hospital Medical Center, and the Department of Pediatrics, Harvard Medical School, Boston, .Massachusetts 02115

SUMMARY Normal erythropoiesis involves differentiation of uncommitted stem cells through committed erythroid precursors into cells specialized for hemoglobin synthesis. Several aspects of this developmental sequence may be studied in murine erythroleukemic cells infected with Friend virus complex. These cells are arrested at the proerythroblast stage, yet capable of continuous growth in vitro. Maturation along an erythroid pathway is induced after treatment with a variety of agents (e.g. dimethylsulfoxide, butyric acid, heroin, ouabainL Following induction, the cells morphologically resemble normoblasts, accumulate globin mRNAs and strain-specific globins, increase heme synthesis and acquire erythrocyte membrane antigens. Cloned populations of erythroleukemic cells mature in a nonhomogencous fashion upon induction, indicative of a stochastic response in the inductive process. This "probability of differentiation" phenotype is formally analogous to stem cell development in which hematopoietic precursor cells form a constant, dividing population from which cells are continuously maturing. Although the sequence of events involved in triggering differentiation is uncertain, cloning and cell hybridization experiments demonstrate that this phenotype is under rather stable genetic (or epigenetic) control. Recent molecular analysis shows that induced differentiation is accompanied by transcriptional activation of the globin genes rather than posttranscriptional stabilization of the globin RNAs. Further application of cellular, molecular and genetic approaches in this system may help to define specific control mechanisms in erythroid development. K e y words: Friend erythroleukemia; variant clones; globins; globin mRNAs; erythroid differentiation.

INTRODUCTION Normal erythropoiesis is a complex, exquisitely coordinated series of cellular transitions. In the bone marrow, a pluripotent stem cell becomes committed to erythroid differentiation and gives rise to nucleated precursors specialized for hemoglobin production (1 L Among the major features of this process are (a) the commitment of the stem cell to undergo erythroid maturation; (b) the expansion of the committed precursor population under the hormonal influence of erythropoietin; (c) the specialization of the cell for hemoglobin production; and (d) the coordination of cellular

events during differentiation, including the coordination of expression of unlinked globin genes. The complexity of the cellular and molecular events involved in erythropoiesis necessitates examination of distinct phases of this process in vitro, if possible, under defined experimental circumstances. Indeed, not all features of this sequence are presently accessible for detailed analysis. Considerable attention, though, has focused recently on an in vitro analogue of erythropoiesis seen in cultured murine cells isolated from the spleens of animals infected with the Friend virus complex (2). Such erythroleukemic or Friend cell lines have been isolated independently in a number of laboratories and can be maintained as clonal populations in continuous culture. Upon treatment with appropriate chemicals, these remarkable cells can be induced to mature along an cry-

;Presented in the formal symposium on Mechanisms of Cellular Control at the 28th Annual Meeting of the Tissue Culture Association, New Orleans, Louisiana, June 6-9, 1977. 146

147

DIFFERENTIATION IN FRIEND CELLS throid pathway. This review draws particular attention to those aspects of this cell system that have broad relevance to cellular controls in development and, in particular, erythroid differentiation. BACKGROUND

Mice infected with the Friend virus complex develop erythroleukemia, characterized by proliferation of reticulum cells in the liver and spleen, erythroblastosis and eventual anemia (3). This variety of leukemia resembles a human disorder, Di Guglielmo's disease. Cell lines derived from the spleens of these animals morphologically resemble proerythroblasts, nonhemoglobinized erythroid precursor cells, and in tissue culture were noted initially by Scher, Holland and Friend (4) to have limited maturation along the erythroid pathway as some hemoglobin synthesis was detectable. The major breakthrough in this system came about when Friend et al. (2) added dimethylsulfoxide to the culture medium in an attempt to increase virus infectivity. To their surprise and delight, over 4 to 5 days the cells began to resemble more mature red-cell precursors, normoblasts, and synthesized large amounts of hemoglobin and heme products. Subsequent to these initial pioneering studies, others observed a sequential increase in heme biosynthetic activities during this process (5), as well as prominent red-cell membrane components, spectrin and glycophorin (7). The hemoglobin synthesized is like that of the mouse strain from which the cell line was derived (8); and a- and/3globin chains generally are synthesized in a coordinate manner, although the relative proportion Of ~major and ~minorcomponents may differ from that of the original mouse strain (9) or vary with the method of induction of differentiation (10, 40). Initial molecular studies demonstrated that cellular maturation was accompanied by an accumulation in the cytoplasm of the messenger RNAs for globin (globin mRNAs) (11). These features of in vitro differentiation of Friend cells are listed in Table 1. Induction of differentiation in erythroleukemic cells is dependent on the concentration of the inducing agent. For the most widely used agent, dimethylsuifoxide, maximal induction is achieved at concentrations greater than 200 mM. In efforts to identify inducing agents with lower-concentration optima and, perhaps, greater target specificity and less cytotoxicity, many compounds have

TABLE 1 ERYTHROIDDIFFERENTIATIONOF FRIEND CELLS

1. Morphologic progression from proerythroblast to normoblast stage 2. Increased synthesis of hemoglobin: (a) Accumulation of globin mRNA (b) Balanced a- and/~-globin chain synthesis 3. Increased heine synthesis 4. Acquisition of RBC membrane antigens, spectrin, glycophorin 5. Limitation of proliferative capacity been treated in several laboratories. From these efforts a number of different kinds of inducers are now known (Table 2). Selected organic solvents (12), the fatty acid butyric acid (13), polymethylene bisacetamides (14), and purines (151 are satisfactory inducers in several cell lines, often in the mM range of less. Ouabain, a cardiac glycoside that specifically binds to membrane Na-K ATPase, is a potent inducer of differentiation, particularly in cells resistant to the cytotoxicity of the drug (16). Heroin, a naturally occurring substance, is an active inducer in some cell lines and may potentiate the response to dimethylsulfoxide (17, 18). Erythropoietin, a hormone for erythropoiesis in vivo, is ineffective alone in inducing these cells. Two possibilities exist: first, these cells are arrested in maturation at a point past erythropoietin sensitivity; or second, these virus-infected cells have lost normal erythropoietin sensitivity. In selected instances, a particular cell line or clone may be responsive to one inducer but not another. These observations suggest that the targets or sites affected by these agents may be varied. At the present time it is not possible to formulate a precise mechanism of action for any inducer of differentiation in these cells. A number of these agents (dimethylsulfoxide, fatty acids and ouabain) quite likely interact with the cell membrane to stimulate differentiation in some unspecified manner. Difficult to reconcile with any model is the purine class of inducers, which may be active TABLE 2 INDUCERSOF ERYTHROIDDIFFERENTIATIONIN VITRO

1. 2. 3. 4. 5.

Dimethylsulfoxide and other organic solvents Butyric acid Heroin Purines Ouabain

148

ORKIN

even in cells metabolically unable to utilize these compounds ( 15~. A convenient, often used simple assay of differentiation in this system is that of scoring the appearance of benzidine-positive cells following treatment with an inducer I19L Benzidine positivity reflects the heme-peroxidase activity of hemoglobin and therefore the presence of hemoglobin. During culture with dimethylsulfoxide, for example, benzidine-positive cells begin to appear on day 3, and increase markedly over the next 2 days. In some cell lines, greater than 90~ of the cells may stain positively after induction. During treatment with hemin, butyric acid or ouabain, benzidine-positive cells may appear somewhat earlier in time. Whether these differences actually reflect different mechanisms of action of these agents is entirely unknown. A more direct and quantitative measurement of erythroid differentiation is determination of the percent of total protein synthesized which is globin. At a still more molecular level, measurement of globin m R N A accumulation in the cells following exposure to an inducer provides elegant quantitation of the extent of erythroid differentiation. In many instances, mere quantitation of benzidine-positive cells is sufficient to monitor erythroid differentiation. In situations where heine and globin synthesis may be uncoupled or heine has been added

exogenously, benzidine staining cannot be relied upon to monitor cellular maturation. THE PHENOTYPE: THE COMMITMENT TO ERYTHROID DIFFERENTIATION In initial comparative studies we were struck by the different extents of maturation achieved by several clones of erythroleukemic cells under identical culture conditions (19) (Table 3). Whether this reflected differences in the extent to which cells in these populations could be induced in a homogeneous fashion or relative heterogeneity in the response to induction was of considerable importance in our understanding of differentiation in this system. Benzidine staining of cultures clearly showed heterogeneity IFig. 1). As this crude measure might represent merely a TABLE 3 COMPARISON OF SELECTED FRIEND CELL CLONES

Cell Line

T3C12 Original stock Long-term passage Clone TGD-3 745 (GM 86}

Globin in RNA as % of CytoplasmicRNA

0.01% 0.04% 0.004% 0.06%

F[6. 1. Erythroleukemic cells of the line T3C12 stained with acid benzidine reagent (t9~: uninduced cells (lefi~ and induced ceils (right L Black-appearing cells are benzidine-positive.

DIFFERENTIATION IN FRIEND CELLS

FIG. 2. Heterogenicity of differentiation at the single cell level in induced populations. Cultures of T3C12 cells were treated with 0.75%, 1.0% and 1.5% dimethylsulfoxide IDMSOI for 5 days. Absorbance at 415 nm was measured in individual cells from each population, and a histogram constructed for each culture. The average absorbance for a posith~e cell in each population is indicated (arrows~. Cells with less than 1% absorbance at 415 nm were scored as negative. The average hemoglobin concentration of positive cells of these cultures did not correlate with the overall extent of induction of the mass culture. The globin mRNA content of cytoplasmic RNA from these populations and the percent benzidinepositive cells by the acid benzidine staining method are shown above each histogram. threshold phenomenon, we examined the hemoglobin concentration in individual cells in induced populations by microspectrophotometry in collaboration with Dr. F. I. Harosi. This more quantitative assessment demonstrated the heterogeneity

FIG. 3. Benzidine positivity of individual colonies of two erythroleukemic cell lines cloned in the presence of dimethylsulfoxide. Clone B ~clone745} differentiates extensively in dimethylsulfoxide, whereas clone A {T3C12 TG D-3} (19~displays only limited induction under identical growth conditions. Individual subclones of these lines were stained with benzidine reagent after 14 days growth in 1.0% dimethylsulfoxide. All colonies of clone A had scattered positive cells.

149

quite directly {Fig. 2 }. The extent of induction of a population reflected the number of cells undergoing differentiation rather than the extent to which the individual ceils in the population could be induced. If this property were dependent on heritable factors, one would expect to observe this heterogeneity within subclones of a cloned population. Such was indeed the case (Fig. 3). When a clone capable of marked erythroid maturation in vitro in response to dimethylsuifoxide is cloned in the presence of dimethylsulfoxide, the colonies arising each have large numbers of benzidine-positive cells and overall fewer cells than colonies grown in the absence of the inducer. When a cell line with more limited capacity to differentiate is similarly cloned, aH colonies have scattered benzidinepositive cells rather than a few colonies with very many positive cells and many without any positive cells. Thus the heterogeneity of cell maturation in mass culture is observed at the clonal level as well. To describe this unusual phenotype, we applied the term "probability of differentiation" to this pattern; i.e. cell lines under specific induction conditions have varying capacities for cellular maturation. Implicit in this description is the notion that the probability of differentiation depends on the cell clone chosen, the inducing agent and its concentration. The probability of differentiation is a stochastic mechanism whereby a fraction of cells under specific conditions mature in response to an inducer. In the absence of inducers, the "basal" globin synthesis in certain lines seems to reflect a "spontaneous" ability of cells within these populations to differentiate at a low rate. Formally this pattern of decision-making for differentiation is similar to that which must occur in the maturation of stem cell population. Stem cells are representative of a class of cells which must give rise to differentiated cells and additional stem cells. The probability of differentiation quantitatively determines the fraction of cells which matures and thereby the fraction of ceils in a total population which becomes differentiated. Although the factors controlling this decisionmaking at the cellular level are unknown at present, it is possible that the mechanisms operating in erythroleukemic cells are those which are utilized by stem cells in vivo in normal erythropoiesis. Differentiation in at least one other in vitro culture system, that of the lipogenic 3T3 fibroblasts, behaves in an analogous stochastic fashion (20~. Recently Gusella et al. ~21} performed a quantitative analysis of differentiation in Friend cells

150

ORKIN

by cloning in plasma clots. Their studies demonstrated that the probability of differentiation accurately describea the observed behavior of these cells in response to an inducer. Furthermore, they showed directly that differentiation is associated with a limitation of proliferative capacity, similar to that which must occur in vivo during terminal erythroid development. CONTROL OF ERYTHROID DIFFERENTIATION The cloning experiments suggested quite directly that the probability of differentiation phenotype was controlled by rather stable mechanisms. Whether epigenetic or genetic factors are responsible for phenotypic differences among clones of erythroleukemic cells is not known. Nevertheless, the interaction of cellular factors capable of regulating cellular differentiation in vitro may be studied in somatic hybrid cells in which the parental lines differ in phenotype. Somatic hybrids formed between erythroleukemic cells and fibroblasts of either mouse or human origin do not exhibit erythroid properties (19, 22, 23L When assayed for hybridizable globin mRNA, such hybrid cells have been uniformly negative to data. Even the "basal" globin m R N A content of uninduced erythroleukemic cells is absent in such hybrids. Although unambiguous interpretation of such somatic hybridization experiments is often difficult, it appears likely that the extinction of differentiation in these hybrids reflects a true negative control by the fibroblast on erythroid differentiation in vitro. It should be emphasized, however, that this negative control may be exerted at any step prior to accumulation of globin m R N A and most probably reflects regulation at a very early site in the pathway of differentiation. For example, fibroblast cytoplasmic or membrane factors might interfere with entry into erythroid differentiation itself. Such a negative control would not necessarily have any relevance to the cellular mechanisms responsible for transcription of globin genes. Somatic hybrids formed between erythroleukemic cell lines with differing phenotypes illustrate additional features of the control of the differentiation phenotype in vitro. In our studies ~19~, one clone with both low basal and induced globin m R N A levels was hybridized with a clone with considerably higher globin m R N A levels in both situations. The hybrids demonstrated an intermediate level of differentiation in the uninduced state, but were barely inducible with dimethylsulioxidc treatment IFig. 4}. The probability of differentiation in the uninduced and in-

duced states therefore may be regulated independently. In view of the complexity of the events involved, it is not possible to describe in more detail the mechanisms at a molecular level. Observations of the phenotype of differentiation suggest that erythroleukemic cells are poised at a stage of differentiation under usual growth conditions and may be "triggered" to differentiate further in response to a number of seemingly unrelated agents. The wide range of active inducing agents, on the one hand, suggests some nonspecificity: the same end result (erythroid differentiation) may be achieved by treatment with various chemicals. On the other hand, the fact that some clones respond better to one agent than another implies that many different sites or targets are specifically involved. In the future a somatic cell genetic approach may be required to dissect the sites and mechanisms involved. There are however, great technical and theoretical impediments to progress in this area. Although it is feasible to isolate noninducible erythroleukemic clones by selecting in dimethylsulfoxide-containing medium (24, 25}, it may be difficult to determine the true defect in such variants. Measurement of globin R N A in either cytoplasm or nuclei of these cells, though

FI6. 4. Globin mRNA content of cytoplasmic RNA from parental erythroleukemic cells and somatic hybrid cells. Independently isolated hybrid clones were assayed for globin RNA untreated ileft} and after treatment with 1% dimethylsulfoxide for 4 days ~right}. In the uninduced state, the hybrids are intermediate between the parental lines. After treatment with dimethylsulfoxide, there is only a modest increase in globin RNA content and the extent of differentiation.

DIFFERENTIATION IN FRIEND CELLS sophisticated, cannot pinpoint the defect, unless large quantities of untranslated globin RNA are present in a noninducible variant. If specific globin RNA is absent in such variants, it may only indicate that a subtle membrane alteration, or similarly ill-defined change, prevents the cell from entering into erythroid maturation in vitro. That membrane changes or ion fluxes may be involved in controlling the differentiation of these cells is suggested by the experiments with ouabain as a potent inducer in culture H6). Possibly rewarding avenues for further work might involve examination of the rates of appearance of particular phenotypic variants with different mutagenic agents and development of novel, selective techniques for the isolation of variant cells. GLOBIN GENE EXPRESSION AND ITS COORDINATION During normal erythropoiesis, maturation of erythroid precursors is accompanied by cytoplasmic accumulation of globin mRNA's (26L At the reticulocyte stage there are, normally, nearly equal amounts of a- and/3-globin RNAs or, perhaps, a slight excess of a-RNA (10 to 30%) {27-29). How the cell normally coordinates the accumulation of globin mRNAs is a formidable problem, especially since the genes for the a- and /3-globins are unlinked and on different chromosomes (30). The erythroleukemic cell system serves as a model with which to explore cellular controls over the accumulation and coordination of specific mRNAs. We are, indeed, fortunate to have sensitive molecular techniques to bring to bear on this problem. As in normal erythropoiesis, differentiation of erythroleukemic cells in vitro is accompanied by cytoplasmic accumulation of globin R N A (Fig. 5L Although this often has been assumed to reflect primarily transcriptional control, direct demonstration of increased globin transcription after induction has been largely lacking. After synthesis of a RNA species within the nucleus, the newly synthesized transcript is subject to factors affecting its stability (posttranseriptional stabilizationj, processing mechanisms which include both shortening as well as modifications on both ends of the m R N A molecule, and transport to the cytoplasm where m R N A stability may be quite important (31, 32). The accumulation of a specific R N A in the cytoplasm may be controlled at any of these potential sites within the cell. One of our recent goals has been to examine several of these potential control sites in a cell-free

151

0.10