To further understand how chronic

To further understand how chronic hypoxia affects BMSCs we identified down-regulation of genes involved in Wnt (Fzd-1, Bbs1, GSK3β, Smo) and Hh (Ihh, Smo) signaling (Table 1) (Barakat et al., 2013). Interestingly, we observed up-regulation of Vangl2, which is involved in the non-canonical Wnt pathway and migration (Ross et al., 2005). As previously reported, these changes in gene expression likely reflect a change in the differentiation potential of BMSCs grown under hypoxic conditions (Drela et al., 2014; Fehrer et al., 2007). In addition to the hypoxic down-regulation of Smo transcripts, we observed that hypoxia mediated the re-localization of Smo proteins to the cilia (Fig. 3C). Furthermore, hypoxia induced Gli2, a target of many of the down-regulated genes, to re-localize to the cilia (Fig. 3D). Additionally, Gli2 and Smo ciliary re-localization has been observed in response to the de-regulation of ciliary calcium levels (Delling et al., 2013). Supporting our findings, it was observed that alterations in Smo and Gli2 levels accompanied changes in cilia length (Nathwani et al., 2014). Future work will be required to determine if long-term exposure to hypoxia alters expression of Hh targets, such as Ptch and Gli1. Ihh and Smo have previously been implicated in signal transduction in chondrocytes (Shao et al., 2012), suggesting that cilia related signaling and hypoxia could influence BMSC differentiation decisions. Indeed previous work has indicated that the long-term culture of BMSCs under hypoxic conditions supports BMSCs retaining an undifferentiated state (Basciano et al., 2011; Berniakovich and Giorgio, 2013). Many of the genes we identified as down-regulated are linked to cell cytoskeleton re-organization, which is required for differentiation (Müller et al., 2013) and motility/migration (Antoniades et al., 2014; Döppler et al., 2013; Fong et al., 2009; Ishizuka et al., 2011; Kulaga et al., 2004; Laeremans et al., 2010; Lin et al., 2012; Lock et al., 2012; Shinozaki et al., 2008). Consistent with this, it has been reported that under normoxic conditions ROCK inhibition increases ciliation and cilia length (Hernandez-Hernandez et al., 2013; McMurray et al., 2013) as does inhibition of F-actin (Yan and Zhu, 2012), both of which mimic the effects we observed under hypoxia. Together with our data this suggests that the differentiation potential of BMSCs will be, at least partially, dependent upon the oxygen tension in which the Ezetimibe are cultured and that cilia may play an important role in this decision. Investigating the effects of chemotactic-mediated migration, under different oxygen tensions, on BMSC cilia we found that the cilia of sub-confluent primary BMSCs did not align to the direction of chemotactic migration (Fig. 5C). Cilia can align to the direction of migration, however this was reported using confluent cancer cells migrating into space, via a scratch wound assay (Jones et al., 2012; Schneider et al., 2010). We found that oxygen tension did markedly effect chemotactic-mediated BMSC migration, however this did not reach statistical significance (Fig. 4A). The reduction in migration observed under hypoxic conditions, in conjunction with our gene expression data and the re-localization of Smo and Gli, supports a recent report which indicated that Smo localization and down-regulation of the Gli transcription factors can affect migration (Chen et al., 2014; Lin et al., 2012). Notably, supporting our gene expression data (Fig. 3A), we did find that the combination of hypoxia and migration significantly affected the length of cilia (Fig. 4B). The longer cilia observed in migrating BMSCs is likely linked to the process of actin polymerization required for the movement of migrating of cells, however further work will be required to assess this. Together, this suggests that the mechanisms governing cilia length and migration are linked and can be influenced by oxygen tension. Exploring this, our network modeling elucidated links between the ciliation and migration pathways, identifying key proteins potentially linking the two processes (Fig. 6B). We validated our modeling using a key protein identified, HSP90, which has previously been localized to cilia and linked to heat shock induced cilia shortening (Prodromou et al., 2012). Inhibition of HSP90 (by AUY922) under normoxia and normal growth conditions significantly increased the number of ciliated cells (Fig. 6C). Supporting this, it has previously been shown that the inhibition of HSP90 leads to a G1 arrest (Georgakis et al., 2006), which would increase ciliation. Interestingly, we found that AUY922 treatment had inverse, oxygen tension-dependent effects on cilia length and we speculate that this is due to different, oxygen tension-dependent, targets of HSP90. Supporting this, oxygen tension-dependent effects of AUY922 on the actin cytoskeleton, cytoskeleton dependent signaling and stress signaling pathways have previously been reported (Djuzenova et al., 2012; Hartmann et al., 2013; Schilling et al., 2012). Specifically we found that primary BMSCs respond to HSP90 inhibition, while previous reports indicate that chondrocytes do not (Wann et al., 2013). This suggests that differentiation may affect cilia related signaling. Interestingly, chronic HSP90 inhibition may promote BMSC differentiation (van der Kraan et al., 2013). However further work is needed to determine if long-term hypoxia exposure, which supports an undifferentiated state (Basciano et al., 2011; Berniakovich and Giorgio, 2013), influences the differentiation potential of BMSCs.