We have identified surprising overlap between genes expressed in HSCs and genes that are expressed in the brain and encode proteins in protein interaction networks for neurological diseases such as ataxia and autism. We have shown that this relationship is not recapitulated by genes expressed in differentiated hematopoietic cells, but is specific to genes expressed in HSCs. Thus, these data reveal a previously underappreciated functional relationship and raise the possibility that additional genes critical for normal brain function might be candidates for regulating HSCs, and vice versa.
This finding is surprising given the generally low overlap between genes annotated to neuronal and hematologic phenotypes (). The fact that there is little correlation between mice with a nervous system phenotype and those with a reported hematopoietic stem cell phenotype as described in MGI, most likely reflects the limited set of phenotypes most investigators consider when studying their genes of interest. Our finding that Atxn1L, a gene identified from the Ataxia interactome, has a hematopoietic phenotype when ablated supports this concept. These data argue that this approach of highly focused phenotyping may obscure unexpected correlations that may have functional relevance. Cross-system analyses, particularly when functions might be predicted from computational approaches derived by mining available biological and in silico data, may be of significant value.
Our data clearly establish Atxn1L as a negative regulator of HSC function. By employing multiple functional assays we show that Atxn1L−/−
HSCs are super-HSCs. They regenerate the blood of recipient mice to higher levels than WT HSCs, they recover more quickly from myeloablative treatment, and they exhibit better engraftment even after secondary transplantation, a rigorous measure of HSC self-renewal capacity. Although, there are now a number of genes that when ablated result in decreased stem cell function, there are relatively few that result in enhanced HSC activity 
. Genes that act normally to restrain HSC activity, resulting in higher performance after KO, include Cbl
. Importantly, none of these genes showed significant down-regulation in Atxn1L−/−
HSCs, suggesting that the mechanism of enhanced stem cell function in the Atxn1L−/−
mice is distinct.
Many of the genes that affect HSC function impact the proliferation rate of HSCs 
. Paradoxically, higher proliferation of HSCs is usually linked to lower HSC activity. For example, HSCs from Irgm
KO mice show excessive proliferation and poor engraftment properties, owing to hyper interferon signaling 
, and Gfi1
mutant HSCs are also hyperproliferative and similarly defective 
. While not well understood, this link between high HSC proliferation and poor engraftment probably relates to differentiation-associated HSC proliferation that ultimately depletes the stem cell pool. Consistent with this, some mutants that decrease HSC proliferation, for example Gli1
, augment HSC function 
. On the other hand, increased HSC proliferation can also be associated with enhanced HSC function: KO of Slug
, both putative transcriptional repressors, results in improved HSC bone marrow engraftment activity along with slightly increased HSC proliferation 
, similar to our observations in Atxn1L−/−
mice. Again, while not fully understood, moderately higher proliferation may enable more rapid engraftment after transplantation (similar to accelerated recovery of blood counts after 5FU) that, if not excessive, may also preserve stem cell function. These findings underscore the critical balance that is maintained to optimize the competing roles of stem cells in self-renewal and differentiation.
The molecular mechanism of the Atxn1L−/−
HSC phenotype is not easy to establish at this time as no major pathways were altered in the gene expression analysis to suggest particular avenues for further study. Atxn1 and Atxn1L have both been shown to interact with the transcriptional repressor Capicua (Cic), which mediates a number of their downstream effects. In the lung, loss of the Atxn1/Atxn1L destabilizes Cic complexes leading to de-repression of activators of matrix metalloproteinases that in turn contribute to the lung alveolarization defects 
. We have not examined protein levels of Cic in HSCs, but we do detect high expression of Cic
in HSCs 
, which leaves open the possibility of a role for Cic in the hematopoietic phenotype as well. Ultimately, better understanding of the mechanisms that lead to enhanced stem cell function could lead to strategies to expand HSCs for bone marrow transplantation which, despite much effort, has still not been achieved.
More broadly, our work suggests the existence of molecular networks that are utilized in both brain and hematopoietic stem cells, but not their differentiated counterparts. Whether these networks are also used in other adult stem cells, as was recently suggested for imprinted genes 
is an open question. Our work also suggests a paradigm for using cross-tissue bioinformatic analyses to identify new key regulators in blood or brain. While other genes are anecdotally linked in both systems, we expect many others could be probed. With the advent of the large-scale mouse phenotyping efforts stimulated by the knock-out mouse consortia, these types of analyses offer a parsimonious use of resources to efficiently identify important phenotypes and cross-tissue phenotype comparisons.
It is interesting to consider why this apparent relationship exists. The fact that neither HSCs nor most brain cells actively divide does not seem to be the cause, as our quiescence signature genes do not show the same enrichment as the HSC fingerprint (). We speculate that there is either a relationship rooted in ontology or evolution that has not been previously noted, or that there is some underlying functional origin. For example, HSCs have a close relationship with other cells in their niche- perhaps they utilize a “synapse” to communicate with other key bone marrow components. Along these lines, a link has previously been noted between some genes with an impact on endothelial cell function and those involved in brain function. For example, classical axon-guidance cues also help guide blood vessel formation 
. HSCs and endothelial cells have a close relationship that originates in their development. HSCs arise from specialized endothelial cells 
and co-express a number of key genes such as Runx1
, and Scl/Tal1
. Thus, it is possible that underlying relationship between HSCs and the brain is also linked to their commonalities with endothelial cells. Systematic analyses with endothelial-specific genes of the type we have performed here would be required to probe this possibility further.
It is also possible that our observations of common brain-HSC networks may hold for humans. Some well known genetic syndromes have been recognized to exhibit both neurologic and hematologic components. For example, Alpha-Thalassemia mental Retardation X-linked syndrome (ATRX) is named for its involvement in both alpha thalassemia and mental retardation (OMIM 301040). Similarly, Ataxia-Telangiectasia (OMIM 208900) and Nijmegen Breakage syndrome, (OMIM 251260) have both hematologic and neurological manifestations. Furthermore, Autism patients may have higher frequencies of infections 
, which could suggest shared genetic etiology. Further studies to explore this intriguing link between the neurologic and hematopoietic system defects are clearly warranted.