EP—Erythroid progenitor

Skip to main contentSkip to article

Download

Blood

Volume 52, Issue 6, December 1978, Pages 1196-1210

ARTICLES

Erythropoietin (Ep) Dose-Response Curves for Three Classes of Erythroid Progenitors in Normal Human Marrow and in Patients with Polycythemia Vera

Author links open overlay panelConnie J.EavesAllen C.Eaves

Outline

Cite

https://doi.org/10.1182/blood.V52.6.1196.1196Get rights and content

Under an Elsevier user license

open archive

Three classes of normal human erythropoietic progenitor cells were investigated with respect to their erythropoietin (Ep) sensitivity under various culture conditions. Dose-response curves extending from 0.001 to 10 units Ep/ml indicated that under optimal conditions the formation of very large erythroid colonies or "bursts" (from primitive BFU-E) required about three times the Ep concentration needed for small bursts (from mature BFU-E) and about five times that needed for the production of small erythroid clusters (from CFU-E). These findings suggest that by the primitive BFU-E stage erythropoietic cells in man have the capacity to respond to Ep and that this responsiveness then progressively increases. Removal of the adherent marrow cell fraction and/or omission of leukocyte-conditioned medium from the cultures resulted in failure to achieve maximum burst formation by primitive BFU-E even in the presence of 10 units Ep/ml. Elimination of these sources of burst-enhancing factors had less effect on the Ep dose-response curve for mature BFU-E, and the effect on CFU-E detection was minimal. These data show that red cell precursors at sequential stages of differentiation also show progressive changes in their response to factors other than Ep that affect burst formation in vitro. Ep dose-response data were also obtained for primitive and mature BFU-E and CFU-E present in specimens from five patients with polycythemia vera (PV). Simple analysis of the curves obtained indicated two coexisting populations in this disease: (1) a population of abnormally (hyper) Ep-responsive cells (sensitive to <0.001 units/ml) and (2) a population of cells whose Ep sensitivity in vitro fell within the appropriate normal range. Quantitation of the ratio of phenotypically normal and abnormal progenitors defined in this way showed a progressive increase in favor of the abnormal cells in the later, mature BFU-E and CFU-E compartments. This could reflect a growth advantage in PV of the abnormally Ep-sensitive line proliferating in a low Ep environment. Such an interpretation would also be consistent with the concept that Ep can influence the production of early red cell precursors in man.

THE REGULATORY ACTION of erythropoietin (Ep) on terminal erythropoietic differentiation events is well established.1 In contrast, very little is known about the possible role of Ep at earlier stages because until recently methods to identify primitive erythroid cells were lacking. Colony assays for a variety of erythropoietic progenitor cell types are now available.2, 3, 4 In the mouse it has been possible to show that these various progenitor cell types correspond to sequential stages of differentiation along the red cell pathway, and the size and time of maturation of the colonies to which they give rise provides a convenient marker for distinguishing them.5, 6 The most mature cells (CFU-E, colony-forming unit-erythroid2) give rise to the smallest colonies that also hemoglobinize first; more primitive cells (BFU-E, burst-forming unit–erythroid3) give rise to progressively larger colonies (called bursts) that hemoglobinize at correspondingly later times.

We previously showed that the concentration of Ep required to stimulate mouse precursors is highest for the most primitive cell types and is progressively less for later cell types, suggesting that the capacity to respond to Ep is acquired early and then increases.5 It has also been shown by studies of the hypertransfused mouse that Ep in vivo may not be obligatory for cells to progress from the primitive BFU-E stage to the CFU-E stage, although it may influence this process.7, 8, 9

The objective of the present studies was to obtain similar information regarding the role of Ep during the early phases of human BFU-E differentiation. We recently described three classes of human erythroid precursors designated as primitive BFU-E, mature BFU-E, and CFU-E on the basis of scoring criteria originally developed for murine precursors.4 Additional experiments10 established a close similarity between each of these cell types and their counterparts in the mouse, suggesting the detection of analogous stages of differentiation, extending from a position close to the pluripotent stem cell down to a position close to the proerythroblast. In the present experiments we examined the Ep sensitivity of each of these three types of precursors in human marrow. In view of the demonstration of burst enhancement attributable to factors released in vitro by leukocytes and adherent marrow cells,11 we also investigated the effect of varying conditions of burst enhancement on the apparent Ep sensitivity of each cell type. The results obtained confirm previous findings in the mouse and provide support for a model of erythropoiesis in man in which erythropoietic cells acquire the capacity to be stimulated by Ep by a very early stage in their differentiation.

The detailed Ep dose-response curves established for normal human CFU-E, mature BFU-E, and primitive BFU-E have also been used to analyze the type of data that is obtained in polycythemia vera (PV). The results in four of five patients investigated in this study are consistent with a two-population model as previously suggested by both cytogenetic12 and isoenzyme13 data. Both of these latter markers, however, are restricted to a small proportion of patients with this disease. The dose-response curves reported here indicate that it may be possible in all patients with PV to quantitate the number of normal and abnormal precursors identified as such by their corresponding normal and abnormal Ep sensitivity phenotypes. In the present study we used this approach to follow changes in the relative and absolute numbers of both of these phenotypes at various stages of progenitor cell differentiation. The results suggest that this approach may be useful for defining parameters that affect normal and abnormal regulation of erythropoiesis in PV.

MATERIALS AND METHODS

Bone marrow specimens

These were all obtained from routine bone marrow samples obtained for clinical evaluation with appropriate informed consent. Cells from five patients without myeloid disease and five patients with PV were studied. The clinical findings on all ten patients are summarized in Table 1. The marrows of all five "normal" patients and of two of the five patients with PV were taken at the time of presentation, prior to the initiation of any treatment. In one of the two untreated PV patients full clinical studies were performed, and the results met all of the diagnostic criteria set down by the Polycythemia Study Group.14 The second (patient 58) did not have red cell mass or pO2 studies performed but had pruritus, splenomegaly, and panmyelosis. Of the three previously established PV patients, one (patient 66) had been diagnosed 1 yr before and had been managed with phlebotomy alone. Another (patient 47) had been diagnosed 12 yr ago and managed with phlebotomy initially but more recently with two courses of 32P. The third (patient 67) had been diagnosed 20 yr ago and managed with phlebotomy, 32P, and busulfan.

Table 1.. Clinical Details of Marrow Specimens Used

PatientAge (yr)/SexDiagnosisHb (g/dl)PCV (%)RBC (per µl × 106)WBC (per µl)Platelets (per µl × 103)5230/FHodgkin11.635.24.8710,8003906255/MHodgkin14.645.05.7815,8003456340/FHodgkin13.440.04.465,0006006841/FOat-cell carcinoma12.938.94.665,0002176944/MIdiopathic splenomegaly16.145.35.536,6001657568/MPV18.655.96.458,6003925840/FPV20.358.57.3216,2004936647/MPV14.6466.2513,70010206765/MPV14.544.47.0510,3002644751/MPV14.846.28.0812,600545

Cell preparation

Bone marrow aspirates were collected in 800 units of sterile preservative-free heparin (Connaught Laboratories, Toronto, Ont.) and then centrifuged lightly to allow removal of the buffy coat and plasma. Usually this was followed by one or two sedimentations to get rid of excess red cells. The leukocyte-rich plasma was then centrifuged again and the cells finally resuspended in 2% fetal calf serum (FCS) (Flow Laboratories, Inglewood, Calif.) in α-medium (Connaught). Cells were then adjusted to a concentration of 2 × 106 cells/ml for plating directly in methylcellulose cultures. Such cell suspensions are referred to as "unseparated" cells.

In some experiments "nonadherent cells"15 were also used. For this procedure cells at a concentration of 3 × 106 cells/ml (in 2% FCS) were incubated for 1 hr at 37°C in plastic tissue culture flasks (Flow) using volumes of 1 ml/10 cm2 flask surface. Following incubation the flasks were gently inverted and the nonadherent cells removed. These were diluted where necessary to a maximum concentration of 1.6 × 106 cells/ml and then plated as described below.

Colony assays

The basic methylcellulose assay mixture used was as previously described4, 5 and contained 0.8% methylcellulose, 30% FCS, 1% deionized bovine serum albumin, 10–4 M 2-mercaptoethanol, cells, and α-medium. Step III Ep (Connaught) was added to give the desired final concentration, and in some instances the final culture mixture also included 9% (v/v) leukocyte-conditioned medium (LCM). To reduce interexperimental variation a single batch of FCS, a single batch of Ep (lot 3013), and a single large pool of LCM were used in all of the experiments reported here. LCM was prepared by the standard agar-medium overlay procedure16 with the following modifications: the cell concentration in the agar was increased to 4 × 106 cells/ml and the medium was supplemented with 10–4 M mercaptoethanol as well as 10% FCS.

All methylcellulose assays were performed in replicate 35-mm Lux Petri dishes (Flow), each dish containing a volume of 1.1 ml and either 2 × 105 unseparated cells or 0.5–1.6 × 105 nonadherent cells (depending on the experiment). Cultures were incubated at 37°C in an atmosphere of 5% CO2 in air with high humidity.

Dishes were removed from the incubator and examined at three different times. Single clusters or pairs of small clusters of erythroblasts recognized by their clearly orange-red coloration were scored on the eighth or ninth day after plating. Such colonies are considered to be derived from a progenitor population analogous to that identified as CFU-E in the mouse.4, 10 Bursts, i.e., colonies derived from more primitive precursors and characterized by their larger size, usually arranged in clusters or subcolonies, were scored first at 10–13 days after plating. At this time the number of small bursts (3–8 clusters) detectable reaches a peak. Burst counts were again recorded 18–21 days after plating, at a time when the largest bursts (>16 clusters) reach a peak.4 Although different sizes of bursts may be present from day 10 onward, the choice of these two different scoring times was based on the finding that large bursts are underscored prior to 2.5 wk incubation because of lack of hemoglobin synthesis in them at earlier times and, similarly, small bursts may be underscored at later times because of mature red cell lysis.4 We refer to the progenitors of the smallest (3–8 cluster) early-maturing bursts as mature BFU-E and to the progenitors of the largest (>16 clusters) late-maturing bursts as primitive BFU-E. All scoring was done directly on live preparations using an inverted microscope. The validity of using red or orange as a discriminating cytoplasmic marker for erythroblasts was confirmed by benzidine staining.

RESULTS

Results for normal marrow

Ep dose-response curves for CFU-E, mature BFU-E, and primitive BFU-E were obtained under three different conditions: (1) unseparated marrow buffy coat cells (2 × 105 nucleated cells/dish) plated in methylcellulose medium containing 9% LCM, (2) nonadherent marrow cells (approximately 105 nucleated cells/dish) plated with 9% LCM, and (3) nonadherent marrow cells plated without LCM. Data for each curve included 14 serial twofold dilutions of Ep plus a measurement of colony formation in cultures to which no Ep was added. This covered a range of Ep concentrations in the culture medium of 0.001–10 units/ml. The lower 0.001 unit/ml limit represented the Ep content of cultures containing 30% FCS (but no added Ep) as determined previously in mouse CFU-E dose-response experiments using the same culture reagents.5 These showed an increasing response to added Ep from a minimum at 0.001 units/ml to a maximum at 0.03 units/ml. In the present experiments complete dose-response curves for five normal human marrow specimens were obtained. The clinical findings on these patients are given in Table 1.

Figure 1 shows the counts obtained in a representative experiment (patient 52). To facilitate comparison of the various curves obtained, a dotted line was drawn to indicate the concentration of Ep that gave half-maximum counts. Comparison of the data in the two bottom panels of Fig 1. shows the effects of different culture conditions on the formation of large, late-appearing bursts. As shown by the open squares (bottom right panel) this type of burst could often not be detected at all even in the presence of high concentrations of Ep when nonadherent marrow cells were cultured in the absence of LCM, although in the presence of LCM a small number of large bursts were always seen (bottom right panel, half-open squares). When unseparated cells were cultured in the presence of LCM (bottom left panel, solid squares), large bursts were detected at much lower Ep concentrations, and the Ep sensitivity of primitive BFU-E thus appeared to be increased. The solid curve and dotted line shown for primitive BFU-E in cultures of nonadherent cells were drawn based on the assumption that the recovery of primitive BFU-E in the nonadherent cell fraction was the same as that obtained for CFU-E and mature BFU-E in the same experiment. This gave a maximum possible number of primitive BFU-E in the nonadherent cell fraction that was approximately twice as high as the number detected with 9% LCM and 10 units Ep/ml in this experiment.

Download : Download full-size image

Fig 1. Representative set of Ep dose-response curves for erythroid precursors in normal human marrow showing raw data (patient 52). Each point, mean of two counts from replicate cultures. Dotted lines, concentration of Ep that gave half-maximum values. Solid symbols, unseparated cells plus 9% LCM; half-filled symbols, nonadherent cells plus 9% LCM; open symbols, nonadherent cells, no LCM.

For small, early-appearing bursts, the effect of omitting the adherent cell fraction and/or LCM was less. Thus, when nonadherent cells were cultured with versus without LCM (Fig 1., middle right panel) there was simply a fourfold shift to the right in the mature BFU-E Ep dose-response curve, indicating that for these cells a measurable increase in Ep concentration could compensate for the lack of LCM. Moreover, when the adherent marrow cell fraction as well as LCM were included in the assay cultures, the Ep dose-response curve for mature BFU-E was not shifted further to the left. This can be seen by comparing the solid triangles and the half open triangles in the two middle panels of Fig 1.. For CFU-E the Ep dose-response curve remained the same or similar even when both LCM and the adherent marrow cell fraction were omitted (Fig 1., two upper panels).

These results suggest that the normal differentiation of erythroid precursors from primitive BFU-E to CFU-E is accompanied by changes in their requirement in vitro for factors derived from adherent marrow cells and peripheral blood leukocytes. Both appeared necessary to minimize the Ep requirement of primitive BFU-E, whereas LCM alone was sufficient to reduce the Ep requirement of mature BFU-E to a minimum, and neither was found to affect the Ep sensitivity of CFU-E in a significant fashion. It remains to be established whether these effects of adherent cells and LCM represent the same or different activities and hence whether these differences in primitive and mature BFU-E responses are qualitative or quantitative. Failure to detect an effect of either adherent cells or LCM on cluster formation by CFU-E may not necessarily be due to a lack of requirement by CFU-E for factors that here appear to be selective for BFU-E. Indeed, the present findings would be quite compatible with an exquisite sensitivity of CFU-E to the residual levels of such factors that were probably present even in cultures of nonadherent cells.

Comparison of the dotted lines in the three left-hand panels in Fig 1. shows an increasing sensitivity to Ep with primitive erythroid precursor differentiation. For CFU-E and mature BFU-E this difference in Ep sensitivity is probably real; at least it cannot be explained as a differential response of mature BFU-E to burst-enchancing factors in the Ep itself, since the differential sensitivity to Ep was shown under conditions already optimal for small burst enhancement, as discussed above (see also Fig 2., below).

Download : Download full-size image

Fig 2. Average Ep dose-response curves for erythroid precursors in normal human marrow. Data from five experiments were normalized and pooled as described in text. Error bars, ± 1 SE. Symbology as in Fig 1..

Results similar to those shown in Fig 1. were obtained with each of the other four control marrows tested, although variation did occur in the maximum counts obtained for each type of colony. These values are listed in Table 2. Two approaches were taken to look at the extent of interexperimental variation in the Ep sensitivities of normal CFU-E, mature BFU-E, and primitive BFU-E shown under each of the three culture conditions used. The first method was simply to calculate a mean 50% stimulatory Ep dose, together with its standard error, for each type of curve using the values obtained in individual experiments. These were simply read off curves fitted to the data by eye, as illustrated in Fig 1.. The results obtained using this method are presented in Table 3. The second approach represented an attempt to generate an "average" normal curve for each cell type and assay condition. This required normalizing the data for each curve, which was done by setting the maximum count to 100%.7 The normalized values for similar curves could then be averaged at each Ep concentration. The results obtained are shown in Fig 2.. The 50% stimulatory values given by the "average" curves shown in Fig 2. are indicated by the numbers in parentheses in Table 3.

Table 2.. Absolute Precursor Numbers Detected per 2 × 105 Marrow Buffy Coat Cells Plated Under Conditions of Optimum Ep (I) or No Added EP (II)

MarrowPatientCFU-EMature BFU-EPrimitive BFU-EIIIIIIIIINormal521160990150Normal62770570780Normal63740600540Normal681190530320Normal692100780430Mean ± 1 SE119 ± 2769.4 ± 9.644.4 ± 11.8PV752949114144756PV584476019110PV66274104587220PV67284149361150PV471371372823.53313Mean ± 1 SE206 ± 5697.6 ± 27.964.6 ± 22.420.9 ± 7.229.2 ± 13.93.8 ± 2.9

The numbers shown under "I" are the highest counts obtained on the upper plateau of the Ep dose-response curve. The numbers shown under "II" are the values obtained in cultures to which no Ep was added, i.e., where the Ep concentration was ≤0.001 units Ep/ml. As described in the text the numbers under "II" correspond approximately to the number of abnormally Ep-responsive precursors in each PV culture. The number of normally Ep-responsive precursors is then given approximately by the total number of precursors detected minus the number of abnormal precursors (i.e., I – II).

Table 3.. Ep Concentration (U/ml Culture Medium) Required for Half-Maximum Stimulation of Different Classes of Human Erythroid Precursors

PrecursorAssay ConditionsBurst EnhancementCFU-EMature BFU-EPrimitive BFU-E*Unseparated cells + 9% LCMHighest0.14 ± 0.030.23 ± 0.040.75 ± 0.13(0.14)(0.28)(0.65)Nonadherent cells + 9% LCMIntermediate0.21t0.30 ± 0.096.6 ± 2.2(0.29)(4.0)Nonadherent cells + 0% LCMLow0.29 ± 0.11.9 ± 1.28.3 ± 2.0(0.23)(1.2)(>10)

Data like those shown in Fig 1. were obtained in five separate experiments. In each experiment a 50% value was read off a curve fitted by eye to the data for a particular precursor assayed under one of the three conditions tested. The values shown here thus represent the means ± 1 SE of four or five individually determined estimates. The values shown in parentheses are the corresponding values derived from the curves fitted by eye to the pooled data in Fig 2..

†One experiment only.

Minimum values, since evidence of an optimal Ep plateau was not obtained in all experiments.

Results for PV marrow

In 1974 Prchal and Axelrad17 reported that patients with PV possessed cells that would form erythroid colonies in the absence of added Ep. This observation has been confirmed in a number of other laboratories.13, 18, 19 However, the interpretation of this observation has remained controversial. As shown above, Ep dose-response curves vary according to the type of colony or burst scored and the time of scoring. Therefore if counts are pooled the dose-response curve obtained will be highly influenced by the predominant colony or burst type present at the time of scoring. As shown above also, Ep dose-response curves vary according to the assay conditions, particularly with respect to whether LCM and/or the adherent marrow cell fraction are included in the cultures. Moreover, both of these variables affect different classes of precursors differently (Table 3). Ep dose-response data will therefore depend not only on whether exogenous LCM is added and the potency of the LCM used but also on the activity of the adherent cells in individual marrow specimens and whether or not an adherent cell separation is performed. The Ep preparation, FCS and BSA used may also contain with LCM-like factors.

In the present studies an attempt to minimize these sources of variability was made by assaying unseparated PV marrow cells (2 × 105 nucleated cells/dish) in cultures supplemented with 9% LCM. (The LCM used was from the same batch as that used for the normal Ep dose-response studies.) We felt that under these conditions the minimum Ep sensitivity of at least CFU-E and mature BFU-E could be measured with confidence, since the adherent cell component of normal marrows had not been found to affect the Ep dose-response curves obtained for either of these two progenitor cell types. Because individual batches of serum may vary in their content of Ep and thereby confuse comparisons of effects attributed to low amounts of added Ep, a single batch of FCS was used in all the experiments reported here. As described above this FCS was estimated to contribute <0.001 units Ep for each 1.1 ml culture.

Marrows from five different patients with PV were investigated using the same range of Ep concentrations used for control marrows, i.e., 0.001–10 units Ep/ml. In each experiment separate counts were performed for CFU-E, mature BFU-E, and primitive BFU-E at the appropriate times. Figure 3 illustrates the type of results obtained, in this case in patient 66. It can be seen that a substantial number of CFU-E derived clusters formed with as little as 0.01 units Ep/ml, and there was no further decrease in this type of colony down to concentrations

Download : Download full-size image

Fig 3. Representative set of Ep dose-response curves for erythroid precursors in PV marrow showing raw data (patient 66). Each point, mean of two counts from replicate cultures containing unseparated cells plus 9% LCM.

≤0.001 units/ml. The results for mature BFU-E were similar except that a smaller proportion of cells in this compartment appeared abnormal in their Ep sensitivity. In this experiment no very large bursts (>16 clusters) were seen at low Ep concentrations, although more than 20 such bursts/dish were obtained when sufficient Ep was added. The "hockey stick" shape of the dose-response curves shown in Fig 3. was characteristic of PV. However, as was found for "normal" marrows, variation did occur in the number of precursors detected at high Ep concentrations (Table 2). In addition, variation occurred in the number of precursors detected in cultures to which no Ep was added (also shown in Table 2).

The simplest interpretation of these data is that in PV two lines of erythroid precursors may coexist and differentiate in vivo to the CFU-E stage. This would be anticipated to include a normally Ep-responsive line as well as a line that was either sensitive to <0.001 units Ep/ml or was in fact autonomous. According to such a hypothesis two variables that would influence Ep dose-response data were considered. These were, first, the percentage of abnormally responsive precursors in a given compartment, and, second, the degree of increased Ep sensitivity (i.e., decreased Ep concentration required for maximum stimulation). For example, if the abnormal CFU-E were characterized by a 400-fold increase in Ep sensitivity, then the simple curves shown in Fig 4.A illustrate the results expected with different PV patients if the relative numbers of abnormal and normal CFU-E were variable. Similarly (Fig 4.B) a series of theoretical curves could be drawn to illustrate the results expected when the relative size of the abnormal CFU-E population remained at 50% but the Ep sensitivity of these cells varied from one patient to another. Figure 5 shows the normalized CFU-E data obtained in exp. 66 together with a theoretical curve derived from the type of modeling shown in Fig 4., using the data obtained on the low Ep plateau to define the percentage of abnormal cells. It can be seen that the curve fits the experimental points well, particularly in the region of normal CFU-E responsiveness (0.02–1 units/ml). Similar fits were obtained when such curves were also generated for mature BFU-E and primitive BFU-E in each of the five PV patients studied, assuming in each case that the relative size of the abnormal population could be defined by the plateau response seen at low Ep concentrations. Our data are therefore consistent with the two-population model proposed.

Download : Download full-size image

Fig 4. Theoretical Ep dose-response curves expected if normal erythroid precursors were mixed with second population of precursors with increased Ep sensitivity. Curve shown for 100% normal cells corresponds to data obtained for CFU-E (see Fig 2.), although similar curves were also generated for mature BFU-E and primitive BFU-E using the reference normal curves obtained for these cell types. (A) Situation actually found in most PV patients, where degree of increased sensitivity of abnormal population was at or beyond detectable limit. (B) Various curves that would be expected if a measurable shift in Ep sensitivity affected half of the precursors.

Download : Download full-size image

Fig 5. Same data as in upper panel of Fig 3. (i.e., for CFU-E) expressed as percentage maximum counts. Line, theoretical curve expected for mixture of 60% normally Ep-responsive CFU-E and 40% CFU-E with Ep sensitivity increased more than 750-fold.

Table 4 shows for the five PV patients studied how the proportion of abnormally Ep-responsive precursors varied in each of the three compartments assayed. The abnormal type was always plentiful in the CFU-E compartment, ranging from 35% to 100% of the total. In the mature BFU-E compartment the abnormal type tended to be represented at a relatively lower frequency, and in the primitive BFU-E compartment abnormally responsive cells were often not present in detectable numbers. (This was the case in three of five patients.) Conversely, the proportion of precursors that were normal tended to be lowest in the CFU-E compartment and higher in the more primitive compartments.

Table 4.. Percentage Abnormally Ep-Responsive Precursors in Individual PV Patients

PatientClinical DetailsCFU-EMature BFU-EPrimitive BFU-E75First presentation35401358First presentation6644<566Treated4013<367Treated5540<1047Treated10090≥40

To obtain these values the counts/culture were converted to percentage maximum values. All points falling on the low Ep plateau were then averaged and finally multiplied by 1.09, since the use of maximum counts for normalization results in an 8% underestimate of the true proportional value, as shown by the average dose-response curves in Fig 2., which plateaued at 92%.

Curves similar to A, B, or C in Fig 4. B, indicating an increased Ep sensitivity within the range detectable with our culture reagents, were not found. As exemplified by the CFU-E data shown in Fig 5., abnormal precursors, whenever present, usually grew as well in FCS alone as with a small addition of Ep. However, in 2 of the 12 dose-response curves where some colonies were obtained in FCS alone, the data did suggest a response to Ep up to a concentration of 0.002 units/ml (analogous to curve D, Fig 4.B). This finding clearly requires further documentation, since it would, like the anti-Ep findings of Zanjani et al.,18 support the concept of altered but not autonomous proliferation and differentiation of erythroid precursors in PV.

DISCUSSION

The purpose of the experiments reported here was to investigate mechanisms of early erythropoietic precursor differentiation in man. Because of the large body of historical evidence showing the specificity of Ep for erythroid cells,1, 7 we were interested in the possibility that the production of cellular receptors for Ep might represent an important step in the early stages of erythroid precursor development. Experiments showing a marked increase in Ep responsiveness during the differentiation of murine CFU-E from BFU-E5, 6 suggested that in the mouse, erythroid precursors have acquired such receptors by the primitive BFU-E stage, although the higher concentrations of Ep required to achieve maximum stimulation suggested that the number or availability of such receptors might be initially low and then subsequently increase. In the present studies we extended this observation to the human system (Fig 2., Table 3). Thus even under conditions where the Ep concentration requirement was minimized (by optimizing conditions for burst enhancement) the dose-response curve for CFU-E remained about 1.5–2 times more sensitive than the curve for mature BFU-E and about 5 times more sensitive than the curve for primitive BFU-E.

The Ep dose-response studies with cells from PV patients provide additional support for the view that Ep responsiveness appears at an early stage. These studies showed that most individuals with this disease possess two distinct populations of erythroid precursors, (1) a normally Ep-responsive population and (2) a population of cells similar in their proliferative and maturational behavior in culture but requiring little or no Ep to progress from a primitive BFU-E to a hemoglobinizing erythroblast. It was thus possible to measure the relative numbers of these two phenotypes at different stages of differentiation. In most of the patients studied in this series, very few precursors of the abnormal type were detected in the primitive BFU-E compartment, but by the stage of mature BFU-E production a substantial increase in their numbers was evident, and by the stage of CFU-E they represented about half of the total CFU-E compartment (Table 4). The normally Ep-responsive precursors also increased in numbers from the primitive BFU-E to the CFU-E stage, but less so. Thus the growth of the normally Ep-responsive precursors in vivo in PV could be shown to be at a disadvantage by comparison to the abnormal population.

Although the explanation of this phenomenon must remain speculative until further information is obtained, two possibilities may be considered. As suggested previously,20 there may be an as yet unknown mechanism of suppression of normal erythropoiesis, possibly by cell-cell interactions. However, the normal mechanism of Ep-mediated regulation of erythropoiesis may also contribute to selection in favor of less Ep-requiring cell types. It is well known that patients with untreated PV have reduced levels of circulating Ep due to their polycythemia and that even after the hematocrit is brought down to normal, Ep production, although increased, may not come up to the levels seen in normal individuals.21 It would therefore be expected that in untreated disease, and possibly in some treated patients also, a low Ep environment would confer a growth advantage on the abnormally responsive cell types. The point at which such a selection process became apparent would then indicate the first appearance of the capacity to respond to Ep. The present studies show this to be prior to the mature BFU-E stage.

These findings are also of interest in considering mechanisms of Ep-mediated regulation. In the mouse it has been found that detectable numbers of cells can differentiate down to the CFU-E stage under conditions of plethora7 i.e., in an environment where Ep levels are very low. The finding of substantial numbers of phenotypically normal CFU-E in two patients with untreated PV provides support for the view that this may also occur in man. In vitro it was also shown recently that some erythropoiesis in murine bursts can proceed all the way to completion in the absence of detectable Ep.22 An explanation for these observations lies in the demonstration that other factors derived from leukocytes and/or adherent marrow cells may also play a role in the regulation of erythropoiesis.11, 22, 23 The present studies suggest that responsiveness to Ep is acquired at a very early stage, even though the subsequent differentiation of cells already responsive to Ep may not actually require Ep until the point of hemoglobinization is reached. The role of non-Ep factors appears to be of a more obligatory nature, at least at the earliest stages of erythropoietic cell differentiation, since the absence of such factors totally eliminates large-burst formation. In order to account for these different effects of Ep and LCM or adherent cell deprivation, we recently suggested that the role of burst-enchancing factors might be to stimulate proliferation and the role of Ep to allow cycling cells to survive and undergo further amplification.10

Such a model explains a number of findings. First, it explains the apparent paradox that the concentration of Ep required to stimulate all primitive BFU-E is very high, even though some normal BFU-E differentiation can occur under conditions where Ep levels would be expected to be very low (in PV). Second, it reconciles the fact that "low" levels of burst-enhancing factors (obtained in nonadherent cell cultures without added LCM) suppress the detection of BFU-E more than CFU-E, whereas "low" levels of Ep (obtained in PV) suppress the production of normal CFU-E more than the production of normal BFU-E. Similarly, it would predict that in PV the production of normal RBC would be suppressed more than the production of normal CFU-E, as was recently found.13, 20 Finally, it explains why CFU-E remain in cycle even after hypertransfusion9 and yet do not survive for more than a few hours in vitro unless Ep is present.23

The finding that the majority of primitive BFU-E in four of five patients with PV appeared to be totally normal in their Ep requirement raises interesting questions concerning the genotype of these phenotypically normal cells. It has been reported that two of three PV patients who presented with 100% abnormal metaphases showed some normal metaphases after treatment.12 The preliminary isoenzyme findings of Prchal et al.13 also suggested the presence of genetically normal erythroid precursors in two G-6-PD heterozygotes with PV. The same study suggested that the abnormally Ep-responsive precursors were all derived from a single pluripotent ancestor. More quantitative information is now required to establish whether or not any of the normally Ep-responsive precursors belong to the genetically abnormal clone. The type of Ep dose-response data presented here provides a basis for investigating this question.

ACKNOWLEDGMENT

We are grateful to M. Heppner, S. Thomas, and C. Smith for expert technical assistance and to members of the Cancer Control Agency of British Columbia and the Vancouver General Hospital for facilitating the obtaining of marrow specimens and associated clinical information. We also wish to express our thanks to Drs. S. Naiman, L. Grossman, and N. Buskard for their cooperation in bringing PV patient material to our attention.