Three different PAI-1 inhibitory compounds (TM5275, TM5509 and TM5614) were used.²⁴ All three selectively and effectively inhibited PAI-1 activity with a half-maximal inhibition (IC50) value <6.95 M in a tPA-dependent hydrolysis assay. The PAI-1 inhibitors (up to 100 M) did not interfere with other serpin/serine protease systems such as 1-antitrypsin/trypsin and 2-antiplasmin/plasmin. The TGF-inhibitor, LY364947, is a potent and selective inhibitor of the TGF- receptor I, inhibiting phosphorylation of Smad3 by TGF-receptor I kinase (IC50=59 nM).

Several different mouse models of CML-like disease were utilized in this study. First, we used a BCR/ABL1 transduction/transplantation- based CML model (BCR/ABL1-CML mice) as previously described.^12,25 Briefly, normal LSK cells (4–5×10⁴ cells per recipient mouse) isolated from mononuclear cells of fetal liver (embryonic day E14.5) or adult BM (8-12 weeks old) were transduced with the human BCR/ABL1-ires green fluorescent protein (GFP) retrovirus and transplanted into irradiated (9 Gy) recipient C57BL/6J mice purchased from CLEA Japan (Tokyo). Another set of experiments, gene-modified 32D cell lines were also transplanted into non-irradiated C3H/HeNCrl mice (CLEA Japan). CMLlike disease developed in these recipient mice by 12–20 days post transplantation.

A tetracycline (tet)-inducible CML mouse model^26,27 was also used in this study. Tal1-tTA mice (JAX, #006209) and TREBCR/ ABL1 transgenic mice (JAX, #006202), both on an FVB/N genetic background, were purchased from the Jackson Laboratory. Tal1-tTA and TRE-BCR/ABL1 transgenic mice were interbred to generate Tal1-tTAxTRE-BCR/ABL1 double-transgenic mice. These animals were maintained in cages supplied with drinking water containing 20 mg/L doxycycline (Sigma-Aldrich). At 8-10 weeks after birth, BM cells of Scl/Tal1-tTAxTRE-BCR/ABL double-transgenic (Ly5.2; CD45.2) mice were mixed with competitor BM cells obtained from congenic (Ly5.1; CD45.1) mice and then were transplanted to lethally irradiated (9 Gy) wild-type (WT) hosts. The recipients bred in the presence of doxycycline, and BCR/ABL expression was induced by doxycycline withdrawal 2 months after transplantation. All induced recipient mice progressively developed CML-like disease associated with a severe myeloid cell expansion in the BM, spleen and peripheral blood (PB), and splenomegaly. The mice became moribund ~3 weeks after induction.

In order to examine the in vivo effects of the combined administration of imatinib (IM) plus PAI-1 inhibitors, BCR/ABL1-CMLaffected mice received vehicle alone, or IM (Gleevec; 150-400 mg/kg/day; Novartis) and/or PAI-1 inhibitor (TM5257, 100 mg/kg/day; TM5009 or TM5614, 10 mg/kg/day) in vehicle, and, at the same time, mice were intraperitoneally injected with 1 mg/kg anti-MT1-MMP antibody (Merck Millipore) or nonimmune species- and isotype-matched control antibody for 5 consecutive days. Treatment was delivered by oral gavage on days 8–30 post transplantation. Three to 4 months after transplantation, PB, spleen, and BM samples were collected from individual recipients and subjected to analyses and secondary transfer experiments.

Protocols concerning animal experiments and recombinant gene experiments were approved by the Animal Care Committee and the Gene Recombination Experiment Safety Committee of Tokai University.

LSC are analogous to HSPC in so many aspects; both cell types are characteristically slow cell-cycling and have a high tendency to localize in particular sites, so called niche dependency.^{14,15,28–30} In addition, the surface markers of CML-LSC and normal HSPC are shown to be almost identical both in murine models and in human samples. For example, transplantation studies of BCR/ABL-induced murine CML models demonstrated that BCR/ABLexpressing LSC activity is confined to the fraction of Lineage (Lin)^–Sca1⁺c-Kit⁺ (LSK) cells that contains murine HSPC population.^12,31 We have recently shown that TGF-signaling regulates the motility of HSPC in the niche through selective upregulation of iPAI-1 expression in HSPC.^17,20 Therefore, we hypothesized that the TGF-−iPAI-1 axis is similarly operational in CML-LSC. In order to investigate the activation of TGF-−signaling in CML cells in vivo, we generated a CML-like myeloproliferative disease mouse model. Normal immature LSK cells obtained from fetal liver (FL) or adult BM were transduced with retrovirus carrying a human BCR/ABL-ires-GFP vector and transplanted into irradiated (9 Gy) recipient mice (Figure 1A). Consistent with previous reports,^12,13 we found that BCR/ABL-transduced LSK cells efficiently induced CML-like disease in recipient mice by 12-20-day post transplantation (Figure 1B). In line with our previous reports, the highest levels of TGF-signal activation was confined to LSK cells, the majority of which are considered to be CML-LSC, as demonstrated by flow cytometric analysis for the phosphorylation of Smad-3 protein, a downstream signaling molecule of TGF-which forms a complex with Smad4 in the PAI-1 promoter and enhances transcriptional activation of PAI-1 gene (Figure 1C). A gradual increase in TGF activity in the CML-LSC fraction correlated with an increase in iPAI-1 expression within the same fraction (Figure 1C). The higher expression of iPAI-1 in CML-LSC was abolished by administration of TGF-inhibitor (LY364947, 10 mg/kg) (Figure 1D), indicating that the TGF-−iPAI-1 axis is indeed activated in CML-LSC.

HSPC are known to be remarkably resilient (in part) because they are kept in cell cycle dormancy through activation of TGF-signaling where iPAI-1 plays a pivotal role in retaining HSPC in their protective niche.^17,20 Since we found that the TGF-−iPAI-1 axis is similarly activated in CML-LSC (Figure 1), we hypothesized that functional activation of the TGF-−iPAI-1 axis contributes to TKI resistance of CML-LSC by facilitating the supportive interaction within the niche. In order to test this hypothesis, we systemically investigated the influence of iPAI-1 on sensitivity of CML cells to TKI by using established CML-like cell lines (32D cells transduced with retrovirus carrying the human BCR/ABL-ires-GFP vector) that are genetically modified to enhance (PAI-1 overexpression [OE]) or nullify (PAI-1 knockout [KD]) the expression of iPAI-1 (Figure 2A). Twenty hours after being treated with IM in vitro, where extracellular fibrinolytic factors, such as tissue plasminogen activator, a PAI-1 target, did not exist, the apoptotic response of these cells was assessed by annexin V/PI staining. We found that a significantly higher percentage of PAI-1 OE cells survived IM treatment than parental 32D cells while a majority of PAI-1 KD cells underwent apoptosis (Figure 2B). This in vitro experiment clearly ruled out the involvement of extracellular anti-fibrinolytic function of PAI-1 in TKI resistance and suggested direct involvement of iPAI-1 in TKI resistance.

Next, we examined the effect of iPAI-1 overexpression on the sensitivity to TKI in vivo (Figure 2C). Seven days after transplantation, the presence of BCR/ABLGFP⁺ CML cells were confirmed in mice that received any of the genetically modified CML cells. IM was orally administered to these recipient mice for 7 consecutive days, and the percentage of GFP⁺CML cells in the BM was analyzed on the day after final IM administration. At day 7 of post transplantation, initial engraftment of PAI-1 KD CML cells in the BM was found to be slightly lower than that of WT CML cells. In addition, consistent with the in vitro findings described above, downregulation of iPAI-1 expression resulted in a significant reduction of CML cells in the BM of TKI-treated mice (Figure 2D). On the contrary, although PAI-1 OE CML cells were significantly less successful in initial engraftment compared to WT CML cells and PAI-1 KD CML cells, PAI-1 OE CML cells were resistant to TKI treatment and overgrew in the BM. These results confirmed that the intensity of iPAI-1 expression governs the susceptibility of CML cells to TKI treatment.

The findings described above prompted us to investigate whether PAI-1 inhibitor administration can increase the susceptibility of CML cells to TKI treatment. CMLbearing mice created by transplantation of the parental BCR/ABL-ires-GFP-transduced 32D cells to non-irradiated mice were treated with either IM alone, a PAI-1 inhibitor alone or IM in combination with a PAI-1 inhibitor for 7 consecutive days and were observed for the following 40 days (Figure 3A). In this study, three different orallyactive, selective PAI-1 inhibitors, namely TM5275, TM5509 and TM5614 were used. The effect of the PAI-1 inhibitors was evaluated by assessing the percentage of BCR/ABL-GFP⁺CML cells in the BM and the overall survival of CML-bearing mice. Seven days after transplantation, all recipient mice developed a CML-like disease characterized by myeloid cell expansion in the BM, spleen and PB, accompanied by splenomegaly (Figure 3BC). Treatment with any one of the PAI-1 inhibitors alone did not markedly extend the survival of CML-bearing mice compared to the vehicle-treated CML mice (Figure 3D). Although treatment with IM by itself delayed disease onset, all IM-treated CML-bearing mice experienced relapse and died before the end of the observation period (Figure 3D). Interestingly, the combined treatment of IM plus any one of the PAI-1 inhibitors significantly inhibited the persistence of CML cells in the BM, reduced spleen size, and prolonged the survival of CML-bearing mice (Figure 3B-D). Furthermore, PAI-1 blockade combined with TKI was found to be effective in overcoming the TKI-resistance in CML-mice transplanted with PAI-1 OE CML cells (Figure 3E-F).

In order to further evaluate the potential therapeutic benefit of PAI-1 inhibitor treatment, we examined the effects of the PAI-1 inhibitor using a CML-induction mouse model, a model that closely resembles CML-development in human. In this previously described tetracycline (tet)–inducible CML mouse model,^26,27 BM cells of Scl/Tal1-tTAxTRE-BCR/ABL doubletransgenic (Ly5.2; CD45.2) mice were transplanted to lethally irradiated (9 Gy) WT hosts, and the recipients were maintained in the presence of doxycycline at least for 2 months (Figure 4A). Within 3 weeks upon the withdrawal of doxycycline and induction of BCR/ABL expression, mice progressively developed CML-like disease. At the time of BCR/ABL induction, mice were treated with either IM alone, a PAI-1 inhibitor alone, or the combination of IM and a PAI-1 inhibitor for 14 consecutive days. On day 15, the expression of BCR/ABL in CML cells in the BM was determined by quantitative real-time PCR. IM treatment alone reduced the level of BCR/ABL–expression in the BM LSK fraction of CML cells to approximately 1/4 of saline treated mice, whereas combined treatment of IM plus the PAI-1 inhibitor reduced the expression of BCR/ABL in CML cells to barely detectable levels (Figure 4B). In addition, the overall survival of this CML-affected mice treated with IM plus the PAI-1 inhibitor was markedly extended (Figure 4C). These results are in accordance with our earlier experiments using genetically modified CML cell lines and BCR/ABL transduced LSK cells and indicate that inhibition of iPAI-1 activity increases the susceptibility of CML cells to TKI treatment.

We then sought to identify how iPAI-1 signaling influences the sensitivity of CML cells to TKI. Since the blockade of iPAI-1 causes detachment of HSPC from the niche,^17,20 we hypothesized that iPAI-1 is similarly involved in the motility of CML-LSC in the BM. In order to test this hypothesis, we examined the expression of MT1-MMP, which has been shown to control the motility of HSPC,^17–19 in CML-LSC derived from PAI-1 inhibitor treated mice. Mice were transplanted with normal FL LSK cells that were transduced with retrovirus carrying the human BCR/ABL-ires-GFP and were treated with a PAI-1 inhibitor for 7 consecutive days starting at day 7 post transplantation (Figure 5A). As expected, treatment with a PAI-1 inhibitor reduced the expression of iPAI-1 in the LSK fraction of BCR/ABL-GFP⁺ CML cells, which represent CML-LSC, and at the same time, increased the expression of MT1-MMP in the same fraction of cells (Figure 5B). Furthermore, iPAI-1 blockade altered the expression levels of CD44 and VLA-4, key adhesion/de-adhesion molecules known to determine the localization of hematopoietic cells within the BM^{14,30,32–34} (Figure 5B). Consistent with this in vivo observation, expression of CD44 and VLA-4 were altered in iPAI-1 OE and KO CML cells, confirming the link between the expression of adhesion/de-adhesion molecules and the iPAI-1 activity (Online Supplementary Figure S1). In order to confirm that iPAI-1 regulates the motility of CML-LSC through the action of MT1-MMP, we examined the directional motility of CML-LSC toward a chemokine gradient using a trans-reconstituted basement membrane (Matrigel^TM) migration assay system. Lin–c-Kit+ immature CML cells isolated from PAI-1 inhibitor treated CML mice exhibited significantly higher trans-Matrigel migration activity compared to those of saline-treated mice (Figure 5C). Importantly, addition of anti-MT1-MMP neutralizing antibody to the culture abo-lished the enhanced migration activity of immature CML cells (Figure 5C), confirming the critical role of MT1-MMP in regulating the motility of CML-LSC. We therefore examined the effect of PAI-1 blockage in the localization of CML-LCS in vivo. BM sections were stained with antibodies against hematopoietic lineage markers (CD3, B220, Mac-1, Gr-1, Ter119, CD41, and CD48) and c-kit to identify CML-LSC and were simultaneously stained with antibody against TGF-to identify niche cells (Figure 5D, the fluorescence images in lower magnification are shown in the Online Supplementary Figure S2), and the sections were evaluated to determine the positional relationships between CML-LSC and niche cells. In PAI-1 inhibitor-treated mice, BCR/ABL-GFP⁺Lin^–ckit⁺ CML-LSC were frequently found further away from TGF--expressing niche cells (Figure 5D), in contrast to vehicle-treated mice in which CML-LSC were often in contact with, or within close proximity to, TGF- -expressing niche cells (Figure 5D). Importantly, the administration of an anti-MT1-MMP neutralizing antibody counteracted the observed detachment of CML-LSC from niche cells in PAI-1 inhibitor-treated mice (Figure 5D). These results suggest that iPAI-1 activity is an important determinant of CML-LSC’ sensitivity to TKI, whereby controlling the MT1-MMP dependent retention of CML-LCS in the BM protective environment.

As shown earlier in this study that CML-LSC exhibited higher iPAI-1 expression (Figure 1), we focused on the effectiveness of iPAI-1 targeting therapy on primitive CML-LSC. Recipient mice that received BCR/ABL-GFP⁺ cells obtained from the BM of mice transplanted with human BCR/ABL-ires-GFP retrovirus transduced FL LSK cells were treated with IM alone or with IM plus a PAI-1 inhibitor for 14 consecutive days (Figure 6A). IM treatment alone reduced the percentage of BCR/ABL-GFP+ LSK cells in the BM approximately by half and mildly decreased the spleen size (Figure 6B-C). Notably, the combined treatment of IM plus a PAI-1 inhibitor diminished BCR/ABL-GFP⁺ LSK cell numbers in the BM and the size of spleen (Figure 6B-C).

Next, we attempted to validate the rationale of PAI-1 inhibitor therapy for overcoming the IM resistance of CML-LSC using a serial BM transplantation assay system. Spleen and BM cells obtained from mice undergoing the drug treatment described above were transplanted to recipient mice. As expected, mice that received cells of saline-treated donor mice developed CML and died within 20 days post transplantation, indicating the persistence of CML-LSC. Even though IM treatment reduced BCR/ABL-GFP⁺ LSK cell numbers in the BM of donor mice by half, it did not protect recipient mice from relapse and all recipient mice died within 30 days. In contrast, more than 70% of mice that received cells from the donor mice that received the combined treatment survived until the end of the observation period (Figure 6D). In line with our in vitro findings, the effect of PAI-1 inhibitor on CML-LCS was completely abolished by the administration of a neutralizing antibody specific for MT1-MMP, indicating a critical role of this molecule in TKI sensitivity of CMLLSC (Figure 6B-D). These data confirmed that iPAI-1 blockade in combination with TKI effectively eliminates CML-LSC.