The analysis of the transcriptome of

The analysis of the transcriptome of OXM-treated HSPCs was revealing. First, the data showed very clearly that the mRNA levels of mTERT were not different between treated and untreated animals. In vitro studies by others have shown that OXM can enhance telomerase activity through estrogen receptor-mediated signaling in short-term assays (Calado et al., 2009). However, this effect was not observed in the long-term in vivo studies described herein. Furthermore, enhanced expression of telomerase, even if present, would not explain the observed effect on stem cell quiescence. We conclude that increased telomerase activity is not the primary mechanism by which OXM acts in vivo. Second, the RNA expression data confirmed that the cell cycle was changed by OXM at the transcriptional level. For example, the mRNA encoding the KI67 antigen was clearly induced. More importantly, however, the data demonstrated a profound (∼10-fold) decrease in osteopontin mRNA. This effect of OXM on osteopontin transcription was AR dependent and likely mediated directly by an AR-target site in Spp1 gene. This finding has clinical implications because it indicates that the use of AR antagonists (Gao, 2010) is not a viable strategy to ameliorate the masculinizing side effects of androgens in bone marrow failure patients. Although osteopontin is best known as a protein found in bone (Nilsson et al., 2005; Stier et al., 2005), it is expressed at robust levels in the stem cells themselves, as evidenced by our RNA-Seq analysis of KSL cells and previously published gene expression database on SPKSL cells (Chambers et al., 2007) (Figure S6). Its precise function in HSPCs is unclear, but it has been shown that Spp1 stem cells have an accelerated Anisomycin (Nilsson et al., 2005). It has generally been thought that osteopontin expressed by bone cells in the HSC niche acts on stem cells in a paracrine mode (Nilsson et al., 2005; Stier et al., 2005). Our data indicate that this protein may also have a cell autonomous effect in stem cells. Future studies with cell-type specific knockouts will be required to address this hypothesis. A second gene, Oasl2, was also suppressed by OXM. Oasl2 (2′-5′ oligoadenylate synthetase-like 2), has not been well studied in HSPCs, despite its known high expression in hematopoietic tissues (Hartmann et al., 1998; Tiefenthaler et al., 1999). Our RNA-Seq data further showed that Oasl2 expression was 11-fold higher in KSL cells than that in whole bone marrow cells. Several publications have identified its proapoptotic and antiproliferative roles in other cell types (Ghosh et al., 2001; Kumar and Mendelsohn, 1989), but its high expression level in KSL cells suggests that it might have a direct role in HSPC function. As a cytokine, osteopontin upregulates the expression of interferons and interleukins. Conversely, Oasl2 is known to be induced by interferons (Hovanessian and Justesen, 2007), promote apoptosis, and suppress proliferation (Ghosh et al., 2001; Kumar and Mendelsohn, 1989). Collectively, it is tempting to speculate that osteopontin and 2′-5′ oligoadenylate synthetase-like 2 function in the same pathway to inhibit HSPC proliferation. OXM’s primary mode of action would be to transcriptionally repress this growth inhibitory pathway. Many publications have suggested that overexpression of osteopontin may play a role in the biology of some cancers, including acute myelogenous leukemia (Bandopadhyay et al., 2014; Liersch et al., 2012), and that its suppression could be therapeutically beneficial. Our data suggest that OXM or other androgens could readily be used for this purpose. OXM should be tested in preclinical animal models of relevant tumors to determine whether it impairs tumor growth. Our results also shed light on another hypothesis regarding the mechanism of action of androgens in anemia. Early work suggested that androgens stimulate erythropoiesis via the activation of EPO pathway. However, subsequent studies found no correlation between serum EPO and androgen levels (Chute et al., 2010). Similarly, we also observed no difference in serum EPO levels between OXM- and placebo-treated mice, despite finding a substantial increase in renal mass in OXM-treated mice, a phenomenon well known to be associated with chronic androgen administration (Shukla et al., 1992). Consistent with this, RNA-Seq transcriptome analysis of early erythroid progenitors did not show any induction of critical EPO-inducible genes or EPO target genes after OXM treatment. Moreover, under our experimental conditions, OXM reduced the MCV level whereas EPO causes macrocytosis, indicating a clear divergence between the action of the two. Our data therefore argue strongly against an EPO-mediated mechanism of action for androgen therapy.