Tissue-Nonspecific Alkaline Phosphatase Is Required for MC3T3 Osteoblast–Mediated Protection of Acute Myeloid Leukemia Cells from Apoptosis Free

In recent years, the protective regions of the bone marrow microenvironment, including the endosteal niche, have come to the forefront as targets to enhance the elimination of acute myeloid leukemia (AML) cells. Although many patients typically achieve remission with standard chemotherapeutic treatments, the subsequent relapse experienced by most patients is likely because of those AML cells that survive the treatments within the endosteal niche: the area where the ossified bone and bone marrow interact (1–5). The osteoblast lineage cells that reside in this niche promote the survival of various cell types, including leukemic cells (6–11); however, the molecular mechanisms that mediate osteoblast protection of AML cells are incompletely characterized. Therefore, further characterization of these osteoblast-protective mechanisms is critical to identifying treatments that will target osteoblast protection and thus permit killing of AML cells within the endosteal niche.

We previously demonstrated that differentiating osteoblasts but not earlier precursors of the osteoblast lineage have the ability to protect AML cells in vitro from apoptosis-inducing agents (12, 13). Interestingly, SDF-1, a chemokine abundantly expressed by multiple cells within the bone marrow, induces apoptosis of AML cells expressing high levels of its receptor, CXCR4 (14). Others further confirmed the apoptosis-stimulating properties of CXCR4 in AML cells by showing that a fully humanized anti-CXCR4 Ab directly induces apoptosis in AML by interacting with and subsequently inducing signaling via CXCR4 (15). Nonetheless, AML cells survive in the bone marrow, suggesting that the endosteal niche provides protection from SDF-1–induced apoptosis. Using coculture systems, we showed that differentiating osteoblasts protect AML cells from SDF-1–induced apoptosis (12). In addition, we showed that this protective effect is disrupted by pretreating the osteoblasts with histone deacetylase inhibitors (HDACi) (12, 16). HDACi make up a class of drugs currently being evaluated in numerous malignancies, including AML, and these drugs exert their effects by inhibiting deacetylation of histones as well as other proteins to keep chromatin in a more open state, thus causing alterations in gene expression in both malignant cells and the osteoblast lineage cells of the bone marrow microenvironment (17–27). Indeed, HDACi-mediated gene alterations inhibit differentiation of immature osteoblasts (26, 27).

Another drug capable of inhibiting differentiation of osteoblast lineage cells of the bone marrow microenvironment is cyclosporine A (CSA). CSA has been used to prevent the rejection of organ transplants; however, one of the side effects associated with this medication is osteoporosis (28, 29). CSA has been shown to alter osteoblast differentiation and affect bone matrix synthesis (30–32). Interestingly, CSA has been shown to reduce activity of the key bone matrix mineralization enzyme tissue-nonspecific alkaline phosphatase (TNAP; encoded by Alpl) (30–32). TNAP has been shown to play an essential role in bone mineralization in the skeletal disorder hypophosphatemia (33). TNAP cleaves inorganic pyrophosphate into inorganic phosphate, which combines with calcium and other ions to form hydroxyapatite, the hardened component of bone (34–36). This mineralized matrix then serves as a repository for growth factors, which have been shown to promote the survival and proliferation of metastatic prostate or breast cancer cells that localize to the bone marrow (37).

In this paper, we demonstrate that TNAP expression is critical for osteoblast-mediated protection of AML cells in vitro. We show that either HDACi or CSA treatment of differentiating osteoblasts reduces the expression of TNAP and inhibits the protective properties of differentiating osteoblasts. Therefore, targeting osteoblast differentiation, particularly matrix mineralization, may be an effective strategy to more completely eliminate AML cells from the bone marrow microenvironment and prevent relapse of disease.

Ascorbic acid, β-glycerophosphate, DMSO, AMD3100, SDF-1, and the protease inhibitor mixture were acquired from Sigma-Aldrich (St. Louis, MO). SDF-1 was obtained from R&D Systems (Minneapolis, MN). APC-conjugated annexin-V was purchased from BD Biosciences (San Jose, CA). Suberoylanilide hydroxamic acid (SAHA) (vorinostat) was procured from the Cancer Therapy Evaluation Program, National Cancer Institute (Bethesda, MD). LBH-589 (panobinostat) was obtained from Selleck Chemicals (Houston, TX). CSA and TNAP inhibitor MLS-0038949 were purchased from Millipore (Burlington, MA). Live/dead viability assays were obtained from Invitrogen (Carlsbad, CA). The ON-TARGET Plus Control small interfering RNA (siRNA) Pool was purchased from GE Healthcare (Dharmacon, Lafayette, CO). The Alpl Silencer Select siRNA (identification no. s62206) was purchased from Thermo Fisher Scientific (Ambion, Waltham, MA). Alpl (TNAP) mouse reactive polyclonal goat IgG Ab (AF2910) was purchased from R&D Systems, and ERK2 rabbit polyclonal IgG Ab (sc-154) was purchased from Santa Cruz Biotechnology (Dallas, TX).

The KG1a and U937 human AML cell lines (American Type Culture Collection, Manassas, VA) were maintained as previously described (12). KG1a-CXCR4 and U937-CXCR4 AML cells were generated by transfecting a plasmid encoding CXCR4-YFP fluorescent fusion protein (38) into KG1a or U937 cells as described (14). MC3T3 sc4 murine calvarial osteoblasts (American Type Culture Collection) are a robust and well-characterized osteoblast model that were cultured in minimum MC3T3 maintenance medium [α-MEM without ascorbic acid (Invitrogen, Carlsbad, CA), 10% FCS (v/v), and 1% penicillin/streptomycin (v/v)] (39). For use in assays, MC3T3 cells were plated in 12-well plates. Upon reaching confluence, MC3T3 cells were treated (defined as day 0) with osteogenic differentiation medium [α-MEM, 10% FCS (v/v), 1% penicillin/streptomycin (v/v), 50 μg/ml ascorbic acid, and 4 mM β-glycerophosphate].

On day −1, MC3T3 cells were plated in 12-well plates in maintenance medium. Where indicated, prior to plating, MC3T3 cells were transfected with 0.8 nmol of either control or Alpl (TNAP) siRNA via electroporation as described (16). On day 0, osteogenic differentiation medium was added to MC3T3 cells [with or without either 0.1% DMSO, 0.025 mg/ml CSA, or 10 μM TNAP inhibitor (MLS-0038949) as indicated]. The CSA dose was selected to inhibit TNAP activity and mineralization of MC3T3 cells (31). The TNAP inhibitor dose used inhibits TNAP activity (40). On day 1, either vehicle (0.1% DMSO), 10 μM SAHA, or 1 μM LBH-589 was added where indicated. Because of the short t_1/2 of SAHA (41, 42), the 10-μM SAHA dose was selected to ensure persistent histone H3 acetylation (a marker of SAHA activity) within SAHA-treated MC3T3 cells for the duration of the 30-h pretreatment course (13, 16, 25). The 1-μM LBH-589 dose showed persistent histone H3 acetylation in LBH-589–treated MC3T3 cells for the duration of the 30-h pretreatment course as well (13, 16). On day 2, the cells were rinsed with PBS, fresh medium consisting of RPMI 1640 and 10% FCS (v/v) was added to the cells, and 1 × 10⁶ KG1a-CXCR4 AML cells per well were added to the differentiating MC3T3 cell cultures. Where indicated, some of these KG1a-CXCR4 AML cells were pretreated with 30 μM AMD3100 to inhibit SDF-1/CXCR4 signaling. After 1 h of coculture, cells in selected wells were challenged with 1.3 × 10⁻⁸ M SDF-1. On day 3, the apoptosis of KG1a-CXCR4 AML cells was assayed via use of APC-conjugated annexin-V staining and flow cytometry to detect surface phosphatidylserine.

The live/dead cell assays were imaged with a LSM 780 laser scanning confocal microscope (Carl Zeiss, Oberkochen, Germany) with ZEISS Efficient Navigation software (Carl Zeiss), an EC Plan Neofluar 10×/0.3 DICI objective, a total magnification of 100×, and excitation/emission wavelengths of 561 nm/626 nm (dead, red) and 488 nm/522 nm (live, green). The percentage of dead (red) cells was calculated by dividing the total number of dead (red) cells by an average total number of live (green) cells plus the dead (red) cells.

On day 0, MC3T3 osteoblasts were treated with osteogenic differentiation medium. On day 1, the MC3T3 cells were treated with HDACi (10 μM SAHA or 1 μM LBH-589) or vehicle control (0.1% DMSO), and they were harvested on day 2 for quantitative real-time PCR (qRT-PCR). The RNeasy Plus Kit (Qiagen, Hilden, Germany) was used to isolate RNA, and SuperScript III First-Strand Synthesis System was used to reverse transcribe the RNA into cDNA (Invitrogen, Carlsbad, CA). The qRT-PCR reactions to measure gene expression used 25 ng of cDNA per 20 μl with QuantiTect SYBR Green PCR Kit (Qiagen) and the CFX384 Real-Time System (Bio-Rad, Hercules, CA). The housekeeping gene Gapdh was used to normalize transcript levels. The 2^ΔΔCt method was used to quantify gene expression levels. Osteogenic genes assayed were Runx2, Sp7, Satb2, Ogn, Omd, Sparc, Ibsp, Bglap, Alpl (TNAP), and Col1a1. Reference genes used were Hprt, Akt1, Atf1, and Sp1. Ezh1 and Hist2h4 were used as controls to show that not all genes are decreased in expression upon HDACi treatment. The qRT-PCR primer sets used are described in Table I.

Two-tailed t tests were used for statistical analysis (Microsoft Excel) with the means of the two distributions being considered significantly different if p < 0.05, unless otherwise specified. For qRT-PCR, transcripts of the indicated genes were normalized to Gapdh (set to 100), and the mean ± SD for expression of each gene are shown for three independent experiments.

We previously demonstrated that HDACi treatment during the early stages of MC3T3 osteoblast differentiation inhibits the osteoblast-mediated protection of cocultured AML cells from SDF-1–induced apoptosis (16). Additionally, we showed that HDACi treatment of primary bone marrow–derived mesenchymal stromal/stem cells during the early stages of osteoblast differentiation led to downregulation of Col1a1, osteocalcin, and TNAP, which are key osteogenic genes (26). We therefore hypothesized that downregulation of one or more of these osteoblast-specific genes required for differentiation might also inhibit osteoblast-mediated protection of AML cells. We began by determining which osteogenic genes displayed altered expression at the time when the AML cells were added to the cocultures with the osteoblasts (Table I). The MC3T3 osteoblast cells were treated with osteogenic differentiation medium on day 0, treated with either HDACi (SAHA or LBH-589) or vehicle on day 1, and then cells were harvested on day 2 for qRT-PCR (Fig. 1A). The 2-d time point was selected because this was the point in differentiation that robust protection of AML cells from SDF-1 began to be observed (12). As expected, compared with undifferentiated MC3T3 cells, day 2 differentiating MC3T3 cells began to show increases in osteoblast differentiation genes, including Alpl, Bglap, and Ibsp (Fig. 1B). In contrast, both SAHA- and LBH-589–treated differentiating MC3T3 cells displayed significantly decreased expression of these and other genes required for osteoblast differentiation: Runx2, Sp7, Satb2, Ogn, Omd, Sparc, Ibsp, Bglap, Alpl (TNAP), and Col1a1 (Fig. 1B). Nonetheless, the expression levels of several reference genes, including Hprt and Akt1 were not significantly altered by HDACi treatment, whereas <2-fold differences were seen in the expression levels of reference genes Atf1 and Sp1 (Fig. 1B). In contrast to the osteogenic genes, Ezh1 and Hist2h4 were upregulated with the addition of SAHA and LBH-589, demonstrating that all genes are not decreased upon HDACi treatment (Fig. 1B). These results identify several key osteoblast differentiation genes that are downregulated by HDACi, including Alpl (which encodes TNAP).

We next asked whether other agents that inhibit MC3T3 osteoblast differentiation could also block the protective properties of the differentiating osteoblasts similar to HDACi treatment. It was previously shown that high doses of CSA inhibit both TNAP activity and matrix mineralization in differentiating MC3T3 osteoblasts (30–32). To assess whether CSA could inhibit osteoblast protection, we used our previously defined coculture model that combines MC3T3 osteoblasts with the KG1a AML cell line transiently transfected with CXCR4-YFP (KG1a-CXCR4 cells). KG1a-CXCR4 cells mimic primary AML patient isolates that express cell-surface CXCR4 and respond to SDF-1 by undergoing apoptosis (14). Additionally, the YFP tag permits unambiguous identification of the KG1a-CXCR4 AML cells following coculture. As shown in Fig. 2A, on day 0, confluent MC3T3 cells received both differentiation medium and either CSA or vehicle. On day 2, the MC3T3 cells were washed with PBS to remove the CSA, and then KG1a-CXCR4 AML cells were added to the differentiating MC3T3 cell cultures and challenged with SDF-1; where indicated, SDF-1/CXCR4 antagonist AMD3100 was also added to cultures as a control (Fig. 2A). On day 3, the apoptosis of KG1a-CXCR4 AML cells was assessed via annexin-V staining and flow cytometry. As expected, SDF-1 induced apoptosis of KG1a-CXCR4 cells, but coculturing KG1a-CXCR4 AML cells together with differentiating MC3T3 cells protected the AML cells from SDF-1–induced apoptosis (Fig. 2B, 2C). Interestingly, CSA pretreatment of the differentiating MC3T3 osteoblasts significantly inhibited their ability to protect KG1a-CXCR4 AML cells from SDF-1–induced apoptosis (Fig. 2B, 2C). To confirm that CSA was not directly inducing apoptosis of the AML cells, we used AMD3100, a small molecule inhibitor of SDF-1/CXCR4 interactions that prevents SDF-1–induced apoptosis of AML cells (14). As shown in Fig. 2B and 2C, AMD3100 significantly inhibited SDF-1–induced apoptosis of KG1a-CXCR4 AML cells in both the presence and absence of differentiating osteoblasts as well as in the presence and absence of CSA-pretreated osteoblasts (Fig. 2B, 2C). Thus, CSA does not directly induce apoptosis of AML cells. These experiments were repeated as in Fig. 2A with a second human AML cell line, U937 (Fig. 2D). CSA pretreatment of the differentiating MC3T3 osteoblasts significantly inhibited their ability to protect U937-CXCR4 AML cells from SDF-1–induced apoptosis. Similar to KG1a-CXCR4 cells, AMD3100 significantly inhibited SDF-1–induced apoptosis of U937-CXCR4 AML cells, even in the presence of CSA-pretreated osteoblasts (Fig. 2D). Together, the results in Fig. 2 demonstrate that pretreatment of differentiating osteoblasts with CSA, a known inhibitor of osteoblast differentiation, significantly inhibits the osteoblasts’ ability to protect AML cells from SDF-1–induced apoptosis (Fig. 2B, 2C).

To characterize the molecular mechanism of this response, we explored the role of TNAP. TNAP is an osteoblast-produced alkaline phosphatase enzyme encoded by the Alpl gene and is required for calcification of the bone matrix (34–36). CSA treatment of osteoblasts has been shown to inhibit TNAP activity (31). Furthermore, in Fig. 1B, we showed that Alpl mRNA levels were decreased by HDACi treatment of differentiating MC3T3 osteoblasts coincident, as was previously shown, with the loss of the ability of these osteoblasts to protect AML cells from apoptosis (16). We therefore hypothesized that TNAP expression is also decreased by CSA treatment of differentiating MC3T3 osteoblasts. Indeed, CSA treatment of differentiating MC3T3 cells resulted in decreased expression of TNAP protein as compared with vehicle control (Fig. 3A). To confirm that CSA treatment did not simply kill the majority of the MC3T3 cells, live/dead assays were conducted to assess MC3T3 viability after CSA treatment. These results showed that although there was a slight increase in dead (red) MC3T3 cells with CSA treatment (∼8.5% dead cells) as compared with vehicle control treatment (∼0.3%), the majority of CSA-treated MC3T3 cells in these experiments still formed a confluent monolayer of predominantly live (green) cells (Fig. 3B, 3C). Together, the results in Figs. 2 and 3 indicate that CSA treatment of differentiating MC3T3 osteoblasts significantly reduces both TNAP expression and the ability of the differentiating MC3T3 osteoblasts to protect cocultured AML cells from SDF-1–induced apoptosis.

Either HDACi or CSA treatment of differentiating MC3T3 cells decreased expression of TNAP and inhibited osteoblast-mediated protection of AML cells. Therefore, we next determined whether TNAP expression is required for osteoblasts to mediate protection of cocultured AML cells. MC3T3 cells were transfected with either control or Alpl (TNAP) siRNA on day −1 (Fig. 4A). Subsequently, differentiation was initiated on day 0. On day 2, KG1a-CXCR4 cells were added with or without SDF-1, and then, on day 3, the KG1a-CXCR4 cells were harvested and assayed for apoptosis (Fig. 4A). Alpl (TNAP) siRNA inhibited upregulation of TNAP in day 2, differentiating MC3T3 cells (Fig. 4B). Interestingly, transfection of differentiating MC3T3 cells with Alpl (TNAP) siRNA significantly inhibited their ability to protect KG1a-CXCR4 AML cells from SDF-1–induced apoptosis compared with control siRNA–transfected MC3T3 cells (Fig. 4C, 4D). These experiments were repeated as in Fig. 4A with a second human AML cell line, U937 (Fig. 4E). Alpl (TNAP) siRNA transfection of the differentiating MC3T3 osteoblasts significantly inhibited their ability to protect U937-CXCR4 AML cells from SDF-1–induced apoptosis (Fig. 4E). To confirm that transfection with Alpl (TNAP) siRNA did not simply kill the majority of the MC3T3 cells, live/dead assays were conducted to assess MC3T3 viability. These results showed that dead (red) MC3T3 cell levels were similar between transfection with control siRNA (∼1.7%) and Alpl siRNA (∼1.5%), and there was a confluent monolayer of predominantly live (green) cells (Fig. 4F, 4G). Together, these results demonstrate that reduction of TNAP expression reduces MC3T3 osteoblast–mediated protection of AML from SDF-1–induced apoptosis.

To further confirm that TNAP was necessary for MC3T3 osteoblast–mediated protection of AML cells from SDF-1–induced apoptosis, MC3T3 osteoblasts were pretreated with a TNAP inhibitor. TNAP inhibitor MLS-0038949 specifically inhibits TNAP activity (40). As shown in Fig. 5A, on day 0, confluent MC3T3 cells received both differentiation medium and either TNAP inhibitor or vehicle. On day 2, the MC3T3 cells were washed with PBS to remove the TNAP inhibitor, and then KG1a-CXCR4 AML cells were added to the differentiating MC3T3 cell cultures and challenged with SDF-1 (Fig. 5A). On day 3, the apoptosis of KG1a-CXCR4 AML cells was assessed via annexin-V staining and flow cytometry. The results show that pretreating differentiating osteoblasts with the TNAP inhibitor significantly impaired their ability to protect KG1a-CXCR4 AML cells from SDF-1–induced apoptosis (Fig. 5B, 5C). Similar results were obtained with U937-CXCR4 cells (Fig. 5D). To confirm that pretreatment of differentiating osteoblasts with the TNAP inhibitor did not simply kill the majority of the MC3T3 cells, live/dead assays were conducted to assess MC3T3 viability. These results showed that dead (red) MC3T3 cell levels were similar between vehicle treatment (∼0.7%) and TNAP inhibitor treatment (∼0.7%), and there was a confluent monolayer of predominantly live (green) cells (Fig. 5E, 5F). Together, the results in this paper indicate that both HDACi and CSA effectively downregulate TNAP expression and that decreased expression or enzymatic activity of TNAP in differentiating osteoblasts is sufficient to reduce osteoblast-mediated protection of AML cells (Fig. 6).

The endosteal niche within the protective bone marrow microenvironment provides a safe harbor for AML cells during traditional chemotherapy (1, 2, 4, 5); however, the molecular mechanisms provided by this niche are incompletely understood. We previously showed that a chemokine found within the endosteal niche, SDF-1, induces apoptosis of AML cells that express high levels of its receptor, CXCR4 (12). However, differentiating osteoblasts protected AML cells from SDF-1–induced apoptosis, thus likely permitting the AML cells to thrive in the bone marrow (12, 14). Interestingly, HDACi treatment of the differentiating osteoblasts was sufficient to inhibit osteoblast-mediated protection of AML cells from SDF-1 (16). Therefore, characterizing the mechanisms behind HDACi inhibition of osteoblast-mediated protection could lead to the ability to design more targeted combination therapies for AML.

To identify HDACi-mediated mechanisms that regulate osteoblast protection, we first performed mRNA expression analysis via qRT-PCR, which demonstrated that expression of several key differentiation genes was inhibited by HDACi treatment of differentiating osteoblasts, including Alpl, which encodes TNAP. These results combined with our previous findings that osteoblast precursors as well as HDACi-treated osteoblasts fail to protect AML cells suggested that osteoblast differentiation is necessary for protection of AML cells (12, 16). Similarly, pretreatment of differentiating osteoblasts with a different inhibitor of osteoblast differentiation, CSA, also inhibited the ability of osteoblasts to protect AML cells. Because, like HDACi, CSA reduced the expression of TNAP in differentiating MC3T3 osteoblasts, we specifically reduced the expression of TNAP in differentiating osteoblasts via siRNA and demonstrated that downregulation of TNAP was sufficient to inhibit their protection of AML cells from SDF-1. In addition, TNAP inhibitor treatment of differentiating MC3T3 osteoblasts reduced protection of AML cells from SDF-1. Together, our results suggest that manipulating osteoblast differentiation, specifically TNAP-mediated matrix mineralization, could inhibit the protective properties of differentiating osteoblasts within the endosteal niche of the bone marrow.

The ability of HDACi and CSA to inhibit both osteoblast differentiation and osteoblast-mediated protection of AML cells suggests that osteoblast differentiation is necessary to protect AML cells from apoptosis-inducing agents. This result is consistent with the clinical observations indicating that an increase in osteoblast-secreted factors correlates with both worse prognosis and the development of chemoresistance in AML (43, 44). Interestingly, mesenchymal stromal cells from AML patients are primed for osteogenic differentiation and express higher levels of TNAP compared with those of healthy controls (45). Furthermore, leukemic induction of osteogenic differentiation of these mesenchymal stromal cells promoted AML cell growth (45). Our results are also consistent with the previous observation that HDACi-mediated gene alterations inhibit differentiation of immature osteoblasts (26, 27). Supporting the role of HDAC involvement in bone development, HDAC3-deficient endochondral bone exhibits reduced expression of extracellular matrix production, bone development, and ossification genes (46). HDACi also directly alter gene expression in malignancies, which makes them attractive options to target malignancies (17–23). However, although HDACi treatment does not eradicate CXCR4 expression in AML cells, it can decrease CXCR4 expression levels, which theoretically may partially impair SDF-1/CXCR4–induced apoptosis of AML cells (47). Thus, we explored alternative strategies to target TNAP expression/activity with drugs that work via different mechanisms of action, including CSA or a TNAP inhibitor. Interestingly, changes in mineralization are observed in malignancy within the bone (48). Sclerostin is a negative regulator of Wnt signaling and bone formation, is produced by osteocytes and malignancies within the bone, and can inhibit osteoblast mineralization and the expression of osteoblast-specific proteins (48). It is unknown whether sclerostin contributes as an attempted protective countermeasure or actively contributes to disease pathogenesis (48).

To our knowledge, this is the first report demonstrating that the key matrix mineralization enzyme TNAP is required for osteoblast-mediated protection of AML cells. Osteoblasts form the mineralized matrix that serves as a reservoir for growth factors that are secreted by many cell types within the bone marrow and that are released upon bone resorption by osteoclasts (37). Malignant cells can influence the behavior of osteoblasts and osteoclasts, and, in turn, the growth factors of the mineralized matrix influence the behavior of the malignant cells (37). A vicious cycle between osteoblasts, osteoclasts, and malignant cells can aid in tumor cell survival and proliferation (37). Decreasing the expression of TNAP could alter matrix structure and thereby disrupt accumulation of key growth factor(s) required for the survival of AML cells. Numerous growth factors have already been found in the mineralized bone matrix, including insulin-like growth factor–1 and –2 (IGF-1 or IGF-2) and TGF-β, which have been reported to support AML cell survival (37, 49, 50). Because inhibiting TNAP reduces mineralization, prolonged TNAP inhibition could cause osteopenia. As such, TNAP inhibition would need to be used at strategic time points in treatment, such as with induction therapy, to maximize the effects on the bone marrow microenvironment during the peak time of malignancy eradication (51). In addition, TNAP inhibition (be it through HDACi, CSA, or potentially, new TNAP inhibitor treatment) could be combined with bisphosphonate treatment to prevent osteoclast resorption of bone and further halt the vicious cycle (51).

Our results reported in this paper add to the accumulating evidence indicating the complex roles of SDF-1 and its receptor CXCR4 in regulating the leukemic bone marrow microenvironment. The SDF-1/CXCR4 axis has been shown to impact multiple pathways of leukemic and other cells of the bone marrow that regulate cell growth, adhesion, migration, gene expression, and survival (52). Using an in vitro approach, we previously established that SDF-1/CXCR4 signaling in AML cells potently induces an apoptosis signaling pathway via decreased expression of Bcl-X_L, upregulation of Noxa, and upregulation of Bak (14). In addition, we found that conditions within the bone marrow microenvironment such as hypoxia and differentiating osteoblasts protect AML cells from SDF-1/CXCR4–induced apoptosis (12, 14). Others subsequently confirmed the apoptosis-stimulating properties of CXCR4 in AML cells, for example, showing that a fully humanized anti-CXCR4 Ab directly induces apoptosis in hematological malignancies, including AML, by interacting with and subsequently inducing signaling via CXCR4 (15, 53). In addition to causing apoptosis in AML cells, CXCR4 signaling has been shown to be capable of inducing apoptosis in T cells, colorectal carcinoma cells, and several other cell types (54–60). In contrast, other studies have revealed a protective effect of SDF-1/CXCR4 interactions on human myeloid tumor cells and have additionally shown that CXCR4 antagonists can impair leukemic cell survival and promote differentiation (47, 61–64). Moreover, CXCR4 antagonists have been shown to mobilize AML cells from the bone marrow (65–69). Nevertheless, CXCR4 antagonists in disease models are not as effective as hoped, possibly because of the leukemic stem cells resisting mobilization (65–69) and/or the inhibition of SDF-1/CXCR4–stimulated apoptosis signaling in these leukemic cells (14, 15, 53). Our results reported in this paper suggest that therapies that target the protective osteoblast-derived mechanisms of the bone marrow microenvironment such as TNAP may be an effective strategy during treatment of AML.

In summary, our results shown in this study demonstrate that inhibiting osteoblast differentiation and specifically matrix mineralization via targeting TNAP could transform the bone marrow microenvironment into a less hospitable environment for AML cells. Additionally, TNAP may also play a role in promoting the survival of metastatic breast or prostate cancer cells that localize to the bone marrow (37). Therefore, combination therapies that involve the manipulation of the osteoblast mineralized matrix could be more efficacious than standard therapies alone in preventing relapse.

The authors have no financial conflicts of interest.