Haemopoietic spleen colony growth: a versatile, parsimonious, predictive model

Substantial support has been obtained for the stochastic model for stem cell differentiation first proposed by Till, McCulloch & Siminovitch (1964), over 20 years ago. By adding a cell maturation pathway, it is possible to predict (by computer simulation) the total number of cells and consequently the time at which individual colonies appear and disappear. Only a few uncontroversial assumptions are required to predict that cells, uniform with respect to self-renewal, are capable of producing the high proportions of late disappearing and late appearing colonies observed experimentally in the spleens of irradiated mice that have been injected with normal haemopoietic cells. It is shown that differences in stem cell self-renewal only slightly influence the time of appearance of colonies; whereas changes in the kinetics of the maturing cells, by changing the size of colonies, has a marked effect on the time of appearance and disappearance of colonies and on the average doubling-time of colony-forming cells per colony (but not the doubling-time of individual colonies). These results (1) seriously question the prevailing view that spleen colonies scored at 8 days measure a separate population (without the capacity for self-renewal), from those scored at 12 days; (2) argue against the existence of multiple sub-populations of stem cells with differing self-renewal and toxicity to cytotoxic agents; (3) help to identify those experiments for which it is obligatory to postulate heterogeneity, and (4) are consistent with self-renewal being regulated by a feedback control of stem cell differentiation, to which only proliferating stem cells can respond and where the stimulus for differentiation decreases at a time when the bone marrow is known to be depleted.