The main finding of this work is that exposure to hypoxia-a common characteristic of many lung disorders, recruits pluripotent VSELs from BM to the peripheral blood and induces a distinct gene expression signature in these stem cells. Using a computationally-intensive approach, we mapped the functional and network properties of activated transcriptional programs and identified hubs that may serve as key regulators of the VSEL response to hypoxia.
To our knowledge, this is the first report on the global transcriptional effects of hypoxia on stem cells using in vivo exposure at physiologically relevant oxygen partial pressures. Utilizing a similar strategy, we recently used a murine model to investigate the consequences of cyclical hypoxia during sleep on BM-derived VSELs . The intermittent hypoxic exposure was intended to model obstructive sleep apnea and that project was therefore more narrowly defined in its focus and clinical relevance. Nevertheless, several of our current findings were also observed in the intermittent hypoxia system, including mobilization of VSELs from BM to PB, elevation of stem cell chemoattractant chemokines, and enrichment of transcriptional programs involved in development and organogenesis. Network analysis of the differentially expressed genes in VSELs also revealed that some of the key hubs were shared between the hypoxia models (intermittent, sustained), including hypoxia inducible factor 2 alpha (Hif2α), proliferator-activated receptor gamma (Pparγ), integrin beta 3 (Itgb3), and insulin growth factor binding protein 5 (Igfbp5). This observation implies that global exposure to reduced oxygen, whether continuous or intermittent, activates a number of common pathways in pluripotent stem cells. However, there were also significant differences in the transcriptional programs enriched in VSELs under sustained vs. intermittent hypoxic exposure. Firstly, a larger number of genes were differentially expressed during continuous hypoxia suggesting a more profound perturbation. Secondly, a much broader spectrum of pathways was enriched in sustained compared to intermittent hypoxia that extended beyond developmental programs and included processes involved in defense and stress response, wound healing, chemokine activity, cell adhesion and structure (Figure 4). Thirdly, many of the network hubs identified during exposure to continuous hypoxia were not differentially expressed during cyclical hypoxia (Figure 5, Table 1), including the most densely connected node, Fos (FBJ osteosarcoma oncogene), and several other key transcriptional regulators such as early growth response 1 and 2 (Egr1, Egr2), and TATA box binding protein (Tbp).
Fos and Egr1 are prototypic master regulators known as immediate early-genes (IEG) that orchestrate widespread activation of cellular gene expression in response to specific perturbations, including hypoxia, ischemia, atherosclerosis, angiogenesis, and neuronal survival [35–37]. Egr2 has been shown to govern the development and segmentation of the mouse hindbrain , and mutations in this gene have been associated with congenital hypomyelinating neuropathies in humans . Intriguingly, a recent study demonstrated that a population of Egr2-expressing precursor cells develop into the first embryonic respiratory rhythm-generating neuronal circuit , and Egr2
transgenic mice die from breathing irregularities resulting in severe hypoxia .
Two members of the heat shock protein 70 group, Hspa8 and Hspa5, were also densely connected network nodes whose expression increased significantly during hypoxia. Products of these genes play crucial roles in protecting cells from environmental stressors including high temperature, oxidative stress, ischemia and hypoxia . More specifically, targeting Hspa8 and Hspa5 has been demonstrated to protect cardiomyocytes  and mesenchymal stem cells from hypoxia-induced apoptosis , whereas delivering these heat shock proteins into transplanted mesenchymal stem cells rescues heart function after myocardial infarction in rats .
Another differentially upregulated hub within the interactome, Hif2α, is a member of the HIF family of basic helix-loop-helix transcription factors that control the global cellular response to oxygen deprivation. Hif2α overexpression is seen many solid tumors characterized with significant hypoxic burden including renal cell carcinoma, non small cell lung cancer and meduloblastoma, and is associated with poor outcome in these cancers . Recent reports suggest that Hif2α is a critical activator of a subpopulation of cancer cells with stem cell-like properties [46, 47], and pharmacologic and genetic targeting of this gene affects differentiation of progenitor stem cells [48, 49]. Interestingly, a single nucleotide polymorphism in Hif2α was recently demonstrated to be highly associated with adaptability to high altitude hypoxia in ethnic Tibetans .
The mobilization of VSELs from BM to peripheral circulation in response to oxygen deprivation implies the activation of organ-specific transcriptional programs under regulatory controls. Our network analysis provided an overview of the transcriptional landscape of these pluripotent stem cells during hypoxia. Since the functional integrity of such scale-free biologic networks is dependent on densely connected nodes , we postulate that these hubs comprise key regulators of hypoxia-induced gene expression in VSELs. As depicted in Figure 5, the hubs not only form the structural foundation of the interactome, but are themselves highly interconnected via direct interactions, implying a high degree of functional coordination in orchestrating the hypoxia-induced transcriptional response of stem cells.
To explore whether this hypoxia-induced transcriptional signature is unique to VSELs or common among leukocyte subpopulations, we compared our results to expression profiles of mononuclear cells (MNCs) and mesenchymal stem cells (MSCs) under hypoxia as reported by Onishi et al . While there are substantial differences between the two studies in terms of experimental protocols, hypoxic exposures, and microarray platforms, less than 1% of the differentially expressed genes in VSELs were significantly altered in MSCs or MNCs, suggesting that many of the hypoxia-induced genes in VSELs are exclusive to this subpopulation of cells. However, from a functional standpoint, differentially expressed genes in VSELs and MSCs were both enriched in several developmental processes, whereas MNCs were highly enriched in defense/immune responses and inflammation. Therefore, VSELs appear to possess transcriptional programs in common with both MSCs and MNCs-a finding that is consistent with their pluripotent properties.
Several limitations in the current study deserve mention. We have only explored the effects of hypoxia on gene expression and not assessed the post-transcriptional modifications of gene products activated during hypoxic exposure. The duration of hypoxia was short-the long term consequences of chronic oxygen deprivation on VSELs were not investigated in this study and represent a future research goal. Although bone marrow is the primary depot of VSELs, our results do not preclude the possibility that some of these stem cells were also mobilized from other tissue niches during exposure to hypoxia. Nevertheless, our expression profiling experiments were focused on BM-derived VSELs. We have not identified the tissue source of hypoxia-induced elevation in stem cells chemokines, but given the extremely low number of circulating VSELs, we believe it is unlikely that VSELs are the primary source. Although the bone marrow is a known source of these chemoattractants, hypoxic injury to various tissues can lead to the release of the chemokines into peripheral blood [5, 14, 16]. Finally, we did not follow the fate of mobilized VSELs to identify their target end organs, although given the global effects of hypoxia and the diversity of enriched developmental programs, we expect that these stem cells will be recruited to multiple sites and undergo tissue-specific differentiation.