In plasma-clot cultures, 'BFU-E-derived colonies' (erythroid bursts) from normal, human peripheral blood were originally defined by Housman et al. as clusters of 3 to 20 aggregated subcolonies with 100 to 1,000 cells total (1).  These clusters were similar to those previously observed with murine BFU-E colonies (2), and the colony number was observed to vary linearly with the cell concentration plated, suggesting that a single entity, the BFU-E, was responsible for each cluster (1).  These colonies could be detected by d10 of culture and reached their peak of development by days 13-14 (1).  Normal individuals were estimated to have an average of 100 to 200 BFU-E per ml of blood; and under the culture conditions employed, a plating efficiency of  about 12 BFU-E/2.5 x 105 cells was obtained with 4 U/ml of purified Epo (1). The reported mean number of subcolonies/burst was 6.5, with an average of 500 cells/burst. 

Eaves and Eaves reported that by days 11 to 12, the bursts obtained in SC methylcellulose had between 3 and 8 subcolonies, and that by days 18 to 21, the bursts scored fell into two types: large bursts with more than 16 subcolonies, and smaller bursts with between 9 and 16 subcolonies (3).  Whereas the day 11 to 12 bursts were classified as "mature" bursts, the day 18 to 21 bursts were considered to be "primitive" bursts derived from the earliest BFU-E; and only the latter type were subdivided into large and small (3).  The same authors reported an average of 30 bursts total /2 x 105 PB MNC (3).  In contrast, using an identical SC assay, Ogawa et al.  reported that between days 11 and 14 two types of bursts could also be distinguished:  type A was formed of two or more subcolonies, each with more than 50 cells, and type B, which had a compact, tight morphology and constituted a single colony (4).  Overall, both types of day 11 to 14 bursts detected had 500 or more cells total (4).  An average total number of 60 bursts (47 type A and 13 type B)/2 x 105 PB MNC was also reported (4). 

Subsequently, using SC assays, other workers reported varying ranges of burst-forming efficiencies: 0/66 (5-6), 9/45  (7), 2/46 (8), 8/16 (9), 18 (10), 9 (11), 29/93 (12),  20/110 (13), 65 (14), 6000 (15), 10/300 (16) 82 (17) and 16/26 (18), where all reported values are expressed per 2 x 105 PB MNC.  All of these assays were carried out in the presence of 1 to 3 U/ml Epo, BPA-containing albumins and 10-30% fetal calf serum. Under optimal 'SF' conditions, others have reported burst-forming efficiencies of 180 (19), 82 (17), 37/70 (20), 61 (21), 5/133 (22), 4/20 (18),  12 (23), 72 (24) and 47 (25) with the values equally expressed per 2 x 105 PB MNC.  Some authors report that all bursts are composed of 3 or more subcolonies, irrespective of subcolony cellularity (18), others count bursts as one or more clusters of ≥50 benzidine-positive cells or alternatively, three or more aggregates of 10 to 49 such cells (23) while still others appear to count subcolonies as bursts (14-15).  To make these discrepancies yet more poignant, some authors have distinguished between small (<8 subcolonies) and large (>8 subcolonies) bursts at day 14, and wrongly named these, respectively, "mature" and "primitive" bursts (25).  Furthermore there are also discrepancies in the total number of cells per burst (and, implicitly, per subcolony) according to the cellularity cut-off point employed.  A burst may have as few as 50 cells (12, 23), or 100 (9), or 500 (4), and averages of 100 (9), 100 to 1000 (1,15), 500 (4), 500/1000 (24), 5700 (21) hemoglobinized cells.  With the exception of Eaves' group (3), all the studies referred to scored bursts at days 12 to 14.

If it is likely that different culture methods (e.g. SC vs. 'SF' culture, etc) and conditions (e.g. purity of preparations employed, different concentrations of Epo, of serum, of albumin utilized, etc), as well as different histochemical and in situ  methods of scoring account, in all likelihood for these reported discrepancies, it is equally apparent that there is little agreement on morphological and cellularity criteria used to score bursts and their component colonies.  Moreover, this lack of unanimity extends as far as the conceptual terminology employed to designate or refer to bursts and their subcolonies.  First, a number of authors have used the same term 'BFU-E' to designate both the product and the progenitor itself.  This is an obvious error and should be strictly avoided.  Second, there are at least a dozen terms that have been used for the entities that arise in culture from BFU-E.  The frequently used term 'BFU-E-derived colony' is in fact a misnomer because it could refer either to the whole group of colonies derived from the BFU-E or to any single member of the group.  Nor is 'BFU-E-derived colony' an operational term, for it assumes a priori knowledge that what is being scored actually came from a BFU-E, when it is obvious that the BFU-E itself cannot be identified directly.

Thirdly, another reason for this confusion stems from the way in which the progeny of the BFU-E become distributed in the course of their growth and differentiation in culture.  The progeny of a single BFU-E appear either as a single colony or as a group of colonies developing in the same 'neighborhood'. This configuration in murine marrow cultures originally gave rise to the term 'erythropoietic burst', one of the terms used at present also to designate the progeny of human BFU-E, but abandoned by some authors because it does not adequately describe those instances in which the progeny of the BFU-E remain together as a single colony.  There is another difficulty with the concept of the 'burst' based on the proximity of colonies to one another: since nothing defines the dimensions of the 'neighborhood', we cannot know where its limits lie.  Accordingly, we are forced to score these entities arbitrarily.  This problem becomes the more critical as the density of colony growth is greater, whether by increased cell numbers plated or by improved culture conditions.  Moreover, other factors such as viscosity of the medium and duration of culture have a profound influence on the numbers scored as they alter the density of colonies per unit area.

However, if cell concentrations are kept low (=/< 10-5 PB MNC plated) it is possible to discern the discrete limits of BCC clusters and the results are reproducible within 10 to 15% of the total mean number of bursts scored.  As an example, the following comparisons of d16 burst-scoring were carried out with a control score performed by Ms. D. Eskinazi:

1) IGF-I + Epo: M +/- SEM

Ours: 98,109,88 98 +/- 7

Control: 77,86,95 86 +/- 6

2) IGF-I:

Ours: 122,95,87 101 +/- 13

Control: 80, 75, 90 82 +/- 5

Using a two-tailed t test, the mean values obtained may be seen not to differ significantly.  This suggests that, aside from consistency in scoring by the same observer, the burst boundaries must have a relatively objective reality.

In light of our results, we have come to operationally define the burst as a cluster of 2 to 10 subcolonies, each subcolony having =/> 50 cells. We refer to these subcolonies as burst-component colonies, BCC.  However, if a single colony of =/> 50 cells is found in isolation it will also be scored as a burst.  This is the limit case where 1 BCC=1 burst, as indicated in the top left diagram of Fig. A1.  As the number of subcolony-clusters appeared to vary linearly with  the cell concentrations used, and as the number of subcolonies did not, it would seem that BFU-E colonies are clonal entities whereas subcolonies are not necessarily so.  But there is an alternative interpretation to our results: it may be that the reason why the production of day 12 to 16 erythroid subcolonies does not vary linearly with the cell concentrations plated is due to the dilution of critical accessory cells that are producing necessary growth factors not being directly provided or, even more simply, due to mixing of subcolonies (occasioned by the proximity of different BCC progenitors).

As shown in Fig. A1., other typical cases of difficult scoring are encountered, and our criteria for counting bursts and BCC is described in the legends of each diagram.

1.  Clarke BJ, Housman D (1977) Characterization of an erythroid precursor cell of high proliferative capacity in normal human peripheral blood.  Proc Natl Acad Sci (USA) 74:1105.

2. Heath DS, Axelrad AA, McLeod DL, Shreeve MM (1976)  Separation of the erythropoietin-responsive progenitors BFU-E and CFU-E in mouse bone marrow by unit gravity sedimentation.        Blood 47:777.

3.  Eaves AC, Henkelman DH, Eaves CJ (1980) Abnormal erythropoiesis in the myeloproliferative disorders: an analysis of underlying cellular and humoral mechanisms.  Exp Hematol 8:235.

4.  Ogawa M, Grush OC, O'Dell RF et al. (1977) Circulating erythropoietic precursors assessed in culture: characterization in normal men and patients with hemoglobinopathies.  Blood 50:1081.

5.  Fauser AA, Messner HA (1978) Granuloerythropoietic colonies in human bone marrow, peripheral blood, and cord blood.  Blood 52:1243.

6. Messner HA,  Fauser AA, Lepine J, Martin M (1980) Properties of human pluripotent hemopoietic progenitors.  Blood Cells 6:595.

7.  Fauser AA, Messner HA (1981) Pluripotent Hemopoietic progenitors (CFU-GEMM) in polycythemia vera: analysis of erythropoietin requirement and proliferative activity. Blood 58:1224.

8.   Nathan DG, Hillman DG (1978) Studies of erythropoiesis in culture.  Blood Cells 4:219.

9.  Lacombe C, Casadevall N, Varet B (1980) Polycythemia vera : in vitro studies of circulating erythroid progenitors.  Br J Haematol 44:189.

10.  Varet B, Casadevall N, Lacombe C (1981) Erythroid progenitors in polycythemia vera.  Blood Cells 7:125.

11.  Tchernia G, Mielot F, Coulombel L, Mohandas N (1981) Characterization of circulating erythroid progenitor cells in human newborn blood.  J Lab Clin Med  97:322.

12.  Douer D, Koeffler HP (1982) Retinoic acid enhances growth of human early erythroid progenitor cells in vitro.  J Clin Invest 69:1039.

13.   Lu L, Broxmeyer HE (1983) The selective influence of hemin and products of human erythrocytes on colony formation by human multipotential (CFU-GEMM) and erythroid (BFU-E) progenitor cells in vitro. Exp Hematol 11:721.

14.   de Wolf JT, Beentjes JA, Esselink MT et al. (1989) In Polycythemia vera human interleukin 3 and granulocyte-macrophage colony-stimulating factor enhance erythroid colony growth in the absence of erythropoietin.  Exp Hematol 17: 981.

  15.  Emerson SG, Thomas S, Ferrara JL, Greenstein JL (1989) Developmental regulation of erythropoiesis by hematopoietic growth factors: analysis on populations of BFU-E from bone marrow, peripheral blood, and fetal liver.  Blood 74:49.

16. Umemura T, Constantoulakis P, Papayannopoulou Th, Stamatoyannopoulos G (1990)  Differentiation kinetics and globin gene expression by circulating human BFU-E in suspension cultures. Exp Hematol 18:1116.

17. Konwalinka G, Geissler D, Peschel C et al. (1986)  Human erythropoiesis in vitro and the source of burst-promoting activity in a serum-free system.  Exp Hematol 14:899.

18.  Misago M, Chiba S, Kikuchi M et al. (1989) Effect of recombinant human interleukin 3, granulocyte-macrophage colony-stimulating factor and granulocyte colony-stimulating factor on human BFU-E in serum-free cultures.  Int J Cell Cloning 7:37.

19.  Casadevall B, Vainchenker W, Lacombe C et al. (1982) Erythroid progenitors in Polycythemia vera: demonstration of their hypersensitivity to erythropoietin using serum free cultures.  Blood 59:447.

20. Akahane K, Tojo A, Urabe A, Takaku F (1987) Pure erythropoietic colony and burst formations in serum-free culture and their enhancement by insulin-like Growth Factor 1.  Exp Hematol 15:797.

21. Migliaccio G, Migliaccio AR (1987) Cloning of human erythroid progenitors (BFU-E) in the absence of fetal bovine serum.  Br J Haematol 67:129.

22.  Eridani S, Dudley JM, Sawyer BM, Pearson TC (1987) Erythropoietic colonies in a serum-free system: results in primary proliferative polycythaemia and thrombocythaemia.  Br J Haematol 67:387.

23.  Dainiak N, Kreczko S,  Cohen A, et al. (1985) Primary human marrow cultures for erythroid bursts in a serum-substituted system.  Exp Hematol 13:1073.

24.  Valtieri M, Gabbianelli M, Pelosi E et al.  (1989)  Erythropoietin alone induces erythroid burst formation by human embryonic but not adult BFU-E in unicellular serum-free culture.  Blood 74: 460.

25.  Dudley JM, Westwood N, Leonard S et al. (1990) Primary polycythaemia: positive diagnosis using the differential response of primitive and mature erythroid progenitors to erythropoietin, interleukin 3 and alpha-interferon. Br J Haematol 75:188.