As hemopoietic stem cells differentiate, the length of time the cells can proliferate shortens by unknown mechanisms. Homeobox cluster (Hox) genes have been implicated by their expression in undifferentiated hemopoietic precursor cells and by enhancement of self-renewal when they are transduced into hemopoietic cells. However, it has been difficult to demonstrate their requirement for self-renewal by gene deletion because of their functional redundancy. Here we enforced HOXB4 expression in purified precursor stages, and compared responses of early stages expressing the endogenous genes with later stages that did not. Contrary to the prevalent view that transduced Hox genes enhance the self-renewal of hemopoietic stem cells in such experiments, we found immortalization, extensive self-renewal and acquisition of reconstituting potential to occur in committed erythroid and myeloid progenitors where the endogenous genes were shutting down. Strikingly, Hox-immortalized clones regressed and disappeared upon Cre-induced deletion of the Hox transgene. The results change our understanding of the stages affected by exogenous HOX proteins and point to shutdown of the endogenous genes as a key determinant of the shortened clonal lifespans of committed progenitor cells.
The preprint is available for view in BioRxiv, while further work continues on defining the stages immortalized by transduced HOXB4.
The study arose through detailed discussions with the Keller group and developed with continued input during the course of Bhaskar Chanda's doctoral project. The paper was the first to show retinoic acid signaling to be required for the generation of definitive hematopoietic stem cells in the paraaortic mesenchyme during embryonic development.
This study defined the first robust deletion phenotype for GATA3 in long-term hemopoietic stem cells, demonstrating shuttling of GATA3 protein from cytoplasm to nucleus contingent on activation by p38 MAP kinase, and the restraining influence of activated GATA3 on self-renewal of the stem cells.
This report built on findings initiated in the lab by Harry Atkins in the early '90s, and further pursued by Christine Ichim in her Masters thesis work in the late 90's. Their work was the first to demonstrate a regulated and graded hierarchy of in vitro proliferative potential in early passage freeze-downs of human AML samples. Among cells from exponentially growing cultures, most individual cells were unable to divide further, many were able to make 2 - 3 divisions, only 3% could make clones of 50 or more cells, while only 1 in 2500 were able to grow extensively to regenerate passageable lines. The subcloned lines then showed the same karyotypic features and hierarchical composition as the parent population. Christine Ichim demonstrated that sister cells in 4 - 8 cell AML clones generated daughter clones of similar sizes to each other, confirming the pre-programmed nature of their proliferative lifespans. She also captured amplified cDNA from single cells whose sibling cells were examined for clonogenic potential, and identified on microarrays genes whose expression correlated with extent of proliferative potential in the sibling cells. In Richard Wells' lab she went on to identify EAR-2 among such genes and demonstrate its biological role in proliferative control in gain and loss experiments.
Long-term multipotent stem cells sustain self-renewal and systemic blood cell production indefinitely. We reported here the identification, purification and characterization of a previously unrecognized intermediate in the murine stem cell hierarchy. Like the long-term HSC, these cells were shown to be capable at single cell level of regenerating all hematopoietic lineages in a mouse, and to be quiescent in normal marrow. Unlike the long-term HSC, they sustained multilineage blood cell production for only 12 weeks. The intermediate-term HSC were separable from the long-term HSC by virtue of expression of alpha2 integrin and dominated conventionally sorted HSC populations numerically. In culture they entered cell cycle 12 hours earlier than long-term stem cells, indicating a first-reported cycle-primed status.
This analysis used the public EST databases to identify valid 3' termination sites for all murine and human transcripts, revealing that 25 - 50% of standard "reference" transcript sequences did not end at correct termini. Identification of true 3' ends allowed construction of genuine 3' microarrays and supported more robust interrogation of cDNA samples amplified globally from single cells or small cell numbers.
This was the first achievement in the field of absolute purification of functionally defined multipotent hemopoietic stem cells able to repopulate mice from a single injected cell. Two independent strategies were used to demonstrate that every intravenously injected HSC effectively reconstituted the marrow, establishing that earlier reports of inefficient engraftment were likely reflective of incomplete HSC purification.
This highly cited report helped to end the stem cell "plasticity" frenzy that seized the field for several years by documenting a failure to reproduce a prominently published claim and setting an example of the rigour required to support or rule out the phenomenon.
The 1995 and 2001 papers captured cDNA from single hemopoietic precursor cells whose identity was established by a novel strategy. Sibling cells in nascent 4-8-cell hematopoietic colonies were first shown generally to possess identical potentials when their growth and differentiation was separately tracked in culture. For analysis of gene expression, single cells were withdrawn from nascent colonies and processed through global RT-PCR. In parallel, the remaining sibling cells were isolated and followed separately in culture. Identical outcomes of the sibling cells provided a probable identification of the biological potentials of the cells processed through global RT PCR. The resulting set of cDNA samples represented various stages and lineages in the precursor hierarchy. Lineage assignments were validated by hybridization of known lineage-specific probes. cDNA subtractions yielded libraries consisting mainly of transcripts expressed at one stage but not another. HPK1, described in the 1996 paper, was discovered in this way.
In the 2001 paper, the cDNA sample archive was expanded to include examples of single cells multipotent for erythroid and myeloid differentiation. Hybridization to a matrix of cytokine receptor probes provided further confirmation of the stage and lineage assignments of the individual cDNA samples. The approach provided the first description of gene expression resolved to discrete precursor stages at single cell level in a differentiation hierarchy.
The 1990 paper described the first approach to global amplification of mRNA quantities in single cells that preserved information about abundance. A key feature was the limiting of the size of the first cDNA strands, permitting efficient amplification that was independent of the size of the original mRNA transcripts.
The 2002 study, performed in the microarray era, was first in the field to establish exponential PCR as a valid approach to amplification of small cDNA samples. PCR amplification was shown to be superior in efficiency and to achieve comparable preservation of representation in direct comparison to linear isothermal amplification, considered before our study to be the standard for single cell work. The findings provided a foundation for the eventual dominance of PCR strategies for transcriptomic analysis of small samples including RNAseq platforms.
Marrow cells were transplanted serially in mouse hosts over the course of 4 years. Long-term reconstituting stem cells, quantitated functionally, were shown for the first time to be capable in vivo of extended, and potentially limitless, expansion in number. The work also showed that numbers of stably engrafted long-term stem cells could be further expanded within the host mice by treatment with specific cytokines.
Hemopoietic stem cell research had reached a pivotal juncture by 1990. Up to then, the spleen colony assay had been widely used as a convenient surrogate measure requiring only 10 - 12 days for a readout, and spleen colony-forming cells (CFU-S) had been isolated in labs in Holland and in California to near purity. However, evidence was accumulating elsewhere to suggest that the cells responsible for permanent reconstitution and maintenance of marrow might be distinct from the CFU-S detected by spleen colony formation. In 1990 Jones, Sharkis et al (Nature 347:188-9) reported that reconstituting cells could be physically purified into fractions distinct from those containing CFU-S, thus seemingly settling the issue definitively. The Nature News & Views piece listed above set forth the context of that study, and concluded that further progress on the reconstituting stem cells would necessitate adoption of longer term in vivo assays. The related 1991 correspondence, also linked above, illustrated the dismay experienced in the field at the prospect of forsaking a convenient and economic short-term assay. The 1995 study from my group was first in the field to describe the phenotype of the cells in mouse marrow capable of long-term reconstitution of marrow measured in vivo over long timespans.
The ensuing 1996 paper went on to confirm the differing phenotypes and separability of long-term HSC from CFU-S by FACS parameters. The study contained other novelties. The kitW41 mouse was first introduced to the field here as an advantageous recipient for study of the reconstituting potential of single transplanted HSC. Another nugget was demonstration of the utility of unlabelled thymidine for operational detection of cycling status and reversal of the lethal effect by addition of unlabelled cytidine.