Skip to main content

Identification of novel factors to expand hematopoietic stem cells in vitro

Final Report Summary - NOVEXPAND (Identification of novel factors to expand hematopoietic stem cells in vitro)


Expansion of hematopoietic stem cells (HSCs) will be required to achieve safe and effective transplantation for treatment of malignant blood disorders in adults. Due to regulatory constraints in the bone marrow (BM) stem cell “niche”, HSCs are quiescent under steady state conditions. Identification of regulators with a potential to expand HSC requires model systems with dividing stem cells, for example embryonic stem (ES) cells. To discover novel regulators that can increase the number of HSCs, we designed a screening protocol to identify possible candidates by comparing gene expression in undifferentiated and differentiated ES cells. We developed several key criteria to select candidates from microarray data and identified four extrinsic soluble factor candidates. From this list of genes we chose 12 candidates including genes encoding 4 extrinsic factors and 8 intrinsic factors. Functional analyses finally identified two novel HSC regulators, Cripto and Dppa5.

Cripto is a member of the EGF-CFC soluble protein family and has been identified as an important factor for the proliferation/self-renewal of ES cells and several types of tumor cells. We found that Cripto increases colony formation and maintains reconstitution ability of HSCs after growth in vitro and that this signaling is mediated through association of Cripto with cell surface receptor GRP78 which is expressed on a subset of HSCs. A subpopulation of HSCs expressing GRP78 (GRP78+HSC) is myeloid-biased, more dormant population, whereas GRP78-HSCs are lymphoid-biased, actively dividing cells. Cripto/GRP78 binding up-regulated and/or phosphorylated several glycolytic metabolism-related proteins, e.g. Pyruvate kinase M2 (Pkm2), Phosphoglycerate kinase 1 (Pgk1). It was noteworthy that after in vitro culture with recombinant Cripto, GRP78+HSCs showed significantly lower increase in mitochondrial activity, indirectly indicating higher glycolysis.

Glycolysis is active under low oxygenic condition (hypoxia) and it is known that HSC located in the endosteal area of bones are in a hypoxic environment. We discovered that GRP78+HSCs mainly reside in hypoxic endosteal areas and non-hematopoietic cells in the endosteal area highly express Cripto. Moreover, injection of anti-GRP78 blocking antibody led to displacement of GRP78+HSCs from the endosteal area to the central marrow.

Cripto expression is mainly regulated by the Hypoxia-inducible factor (HIF) signal. HIF-1α null mice have been shown to exhibit failure in the maintenance of functional HSCs even though the immunophenotypic percentage of HSCs is normal. Analysis of the HIF-1α null mice demonstrated a reduced number of GRP78+HSCs within the endosteal areas and decreased number of endosteal niche cells expressing Cripto. Taken together, findings suggests that the Cripto/GRP78 pathway is an important signal, as an intermediary of HIF-1 regulation, to maintain HSCs in the hypoxic environment of the endosteal niche, by inducing glycolytic metabolism in dormant HSCs. (Miharada et al., Cell Stem Cell 2011; Ann N Y Acad Sci 2012)

Dppa5 (developmental pluripotency associated 5) is an RNA binding protein known for its high expression level in ES cells as well as Cripto.

Overexpression of Dppa5 robustly increased the reconstitution level in lethally irradiated mice without loosing multi-lineage potential after 14 days culture in vitro. Conversely, down-regulation of Dppa5 using shRNA resulted in significant severe reduction of reconstitution of the transplanted Dppa5 knockdown HSCs.

Proteomics approach using two-dimensional gel electrophoresis (2D-DIGE) explored that the levels of proteins involved in redox, amino acid/protein processing and mitochondrial activity related proteins were increased. In contrast, decreased proteins included glycolysis related enzymes and ER stress response proteins. Surprisingly, all decreased glycolysis proteins (Pkm2 and Pgk1 etc.) have been observed as increased proteins when HSCs were cultured with Cripto, suggesting alteration in HSC metabolism. In fact, higher mitochondrial activity and reduced lactate production were observed in the Dppa5 over-expressing HSCs.

Moreover, GRP78 (a cell surface receptor for Cripto), GRP94, Calreticulin, Pdia3 and Txndc5 were decreased upon Dppa5 overexpression: These molecules are key components of ER stress chaperones. Importantly, when the ER stress inducer, Tunicamycin, was used to treat HSC in vitro, the competitive advantage of Dppa5 was abolished. We also found that in vitro cell culture rapidly and dramatically increased ER stress levels in HSCs. Remarkably, treatment with an ER stress inhibitor, Tauroursodexycholic acid (TUDCA), maintained functional HSCs for 2 weeks in vitro. These findings strongly indicate that reduced ER stress and improved protein folding in vitro may improve HSC function and activity. (Submitted)

Thus, this project successfully discovered two novel HSC regulators as aimed in the project description. Although both Cripto and Dppa5 enable in vitro maintenance of functional HSCs, their regulation manners are very different as regulating same glycolytic enzymes but toward opposite directions. Regulatory mechanism of Cripto is consisted with current but classic trends, reconstructing the niche in vitro to “maintain” HSCs. In contrast, Dppa5 invoked the completely new aspect for HSC maintenance/expansion, highlighting ER-stress as a critical cause of the impairment in HSC functions. Scientific and technological developments upon the project are not only the identification of the two genes, but also the suggestion for advanced treatments, e.g. the use of TUDCA.

The initial hypothesis of the project was asking what were the specific machineries enabling pluripotent stem cells to increase the cell number without losing their functionality and trying, which can adapt to HSC culture. Many database suggest pluripotent stem cells have large stress-response capacity including ER stress response. The findings of this project therefore are quite reasonable.