Hematopoietic Stem Cell Numbers Are Regulated at Both Systemic and Local Levels

Since Schofield proposed the niche model in the 1970s, it has been widely believed that the number of hematopoietic stem cells (HSCs) is determined by the niche size. We and others have identified several components of the HSC niche, including Nestin-GFP⁺ mesenchymal stem cells (MSCs). However, the number of defined niche cells vastly exceeds that of HSCs, raising the question about the actual regulatory mechanisms governing HSC numbers.

To address this question, we initially established a mathematical model of niche availability and occupancy. The model predicted that an increase in available niches initially leads to a corresponding rise in the number of occupying HSCs, which aligns with the concept of local, niche availability-driven restriction on HSC numbers. Interestingly, our model also suggested that this increase in HSC numbers saturates when the number of available niches reaches a critical level. One potential mechanism for this saturation is that HSC numbers may also be regulated at the systemic level, with total HSC numbers in the body being restricted.

To experimentally test this model and hypothesis, we developed a bone transplantation technique that allows us to increase the number of available HSC niches. When a single femur (referred to as a graft) was transplanted per mouse (called a host), we observed the progressive regeneration of bone marrow (BM) stroma and hematopoietic recovery in the grafts. While the number of HSCs (Lin^–Sca-1⁺c-Kit⁺CD150⁺CD48^–CD34^–) in the grafts was still 41.7 ± 7.60% of that in the host femurs at 5 months after transplantation (n = 8, P = 0.0002), they were comparable at 3 months after transplantation when granulocyte colony-stimulating factor (G-CSF) was administered to hosts at 1 month after transplantation (n = 8, P = 0.660). By utilizing Nestin-GFP transgenic mice and a CD45 congenic system, we found that MSCs in grafts are derived from the graft, while hematopoietic cells are derived from the host.

Given that a single femur harbors only 8.3 ± 0.440% of HSCs in the total body (n = 11), 6 femurs were transplanted per mouse, followed by G-CSF administration, to assess the impact of increased HSC niches on total HSC numbers. Strikingly, HSC numbers per host femur and graft femur were lower than those per femur from sham-operated mice (n = 7-48, P = 0.012). Moreover, we found that the sum of HSC numbers in the bone transplantation host and graft bones was equivalent to that in the sham-operated group (n = 7-8, P = 0.364). Total HSC numbers were also maintained when 6 femurs from wild-type (WT) mice were transplanted into Cdh2 (N-Cadherin)-CreER; Cxcl12^fl/fl mice, which have an impaired ability to retain HSCs in the BM (n = 8, P = 0.711). These results indicate that the addition of normal niches does not alter total HSC numbers in the body, regardless of whether the endogenous niche is functional, supporting the conclusion that HSC numbers are indeed restricted at the systemic level (Figure 1A).

To determine whether HSC numbers are also limited at the local level, as suggested by our mathematical model, we transplanted a single WT femur to Cxcl12^fl/fl or Cdh2-CreER; Cxcl12^fl/fl mice. We hypothesized that implanting only one femur would have a minimal impact on the total BM, and therefore HSC numbers in the grafts transplanted to Cxcl12^fl/fl mice would reach the physiological level. We reasoned that if there is no local restriction on HSC numbers, there would be more HSCs in the WT grafts transplanted to Cdh2-CreER; Cxcl12^fl/fl mice compared with those implanted to Cxcl12^fl/fl mice. As expected, HSC numbers per WT graft femur in Cxcl12^fl/fl mice were comparable to those per femur of sham-operated Cxcl12^fl/fl mice (corresponding to the physiological level) (n = 8, P = 0.433). However, CXCL12 deficiency in the host BM did not affect HSC numbers per WT graft femur (n = 8, P = 0.986). Importantly, the sum of HSC numbers in the host body and the graft femur of Cdh2-CreER; Cxcl12^fl/fl recipient mice was lower than that of sham-operated Cxcl12^fl/fl mice (n = 8, P = 0.003), indicating that total HSC numbers in the former mice are below the systemic limit. Collectively, these data suggest that the failure to increase HSC numbers per WT graft in Cdh2-CreER; Cxcl12^fl/fl hosts is, at least in part, due to a local restriction on HSC numbers (Figure 1B).

Our study demonstrates that HSC numbers are restricted at both systemic and local levels, thereby rewriting the long-standing model for the regulation of HSC numbers.