CD150⁻ cells are transiently reconstituting multipotent progenitors with little or no stem cell activity

To the editor:

Goodell and colleagues recently published a paper in Blood (Weksberg et al¹ ) that confirmed many of our conclusions from prior studies demonstrating the utility of signaling lymphocytic activation molecule (SLAM) family markers to enhance the purification of mouse hematopoietic stem cells (HSCs).^2-5  However, they took issue with 2 conclusions. First, they argued that SLAM family members are not sufficient for the identification of HSCs and second that “a substantial fraction” of HSCs are CD150⁻, in contrast to our published data. However, the data in the Goodell paper do not support these conclusions.

Weksberg et al¹  argued that in contrast to our published data,^2-5  only a minority of cells expressing SLAM family markers (CD150⁺CD48⁻CD41⁻) expressed the HSC markers c-kit and Sca-1. However, Figure 1B in the Goodell paper shows that they used an inappropriately loose gate that included mainly CD150⁻ cells in the CD150⁺CD48⁻CD41⁻ population. Because the gate included 1% of cells in the isotype control and 0.4% of cells in the experimental sample, the authors did not attempt to sort the CD150⁺CD48⁻CD41⁻ population characterized in prior studies, which represents only 0.007% of cells.³  The gate used by Weksberg et al¹  is different from that illustrated in our papers (compare Figure 2A from Kiel et al²  to Figure 1B in Weksberg et al¹ ). Obviously, SLAM markers will render poorer purity than reported in prior studies if gates are set to include mainly CD150⁻ cells.

The only experimental support offered by Weksberg et al¹  for the existence of CD150⁻ HSCs was an experiment in which 100 CD150⁻ c-Kit⁺Lineage⁻Sca-1⁺ (KLS) side population (SP^KLS) cells were injected into irradiated mice.¹  They observed significantly less reconstitution than that observed from CD150⁺ SP^KLS cells but nonetheless concluded that CD150⁻ cells represent “a substantial fraction” of HSCs. However, to test this conclusion it would be necessary to perform a more careful clonal analysis. What percentage of CD150⁻ cells actually give long-term multilineage reconstitution: 1%, 10%, 100%? Without such data it is impossible to know what percentage of HSCs are CD150⁻, whether such cells represent a significant fraction of HSCs, or whether the HSC activity could be explained by rare contamination from CD150⁺ cells. Weksberg et al did not even disclose what fraction of mice that received transplants of 100 CD150⁻ SP^KLS cells showed long-term multilineage reconstitution. If some did not, this would prove that less than 1% of CD150⁻ SP^KLS cells are HSCs, disproving the conclusions of this paper.

We have performed clonal analyses of the CD150⁻ fraction of c-kit⁺Sca-1⁺CD48⁻ (KS48) or c-kit⁺Thy-1^lowLineage⁻Sca-1⁺ (KTLS) cells. We found less than 5% of HSCs in the CD150⁻ fraction of whole bone marrow, KS48, or KTLS cells (Table 1). CD150⁻ cells give transient multilineage reconstitution of irradiated mice. These results have been independently replicated by other labs in papers^6,7  not cited by Weksberg et al.¹  Our conclusion that CD150⁻ KS48 cells are mainly transiently reconstituting multipotent progenitors is consistent with data in the Goodell paper that showed more rapid proliferation and reduced myeloid reconstitution from CD150⁻ SP^KLS cells compared with CD150⁺ SP^KLS cells. Goodell's conclusion that CD150⁻ cells represent a substantial population of rapidly dividing HSCs that can give rise to CD150⁺ HSCs but that have less reconstituting potential than CD150⁺ HSCs is internally inconsistent, inconsistent with more rigorous data from multiple labs, and inconsistent with decades of work demonstrating that HSCs are slowly dividing.