(A) Chemical structure of XMD8-92.
(B) HeLa cells were serum starved overnight followed by treatment with 1 or 5 μM XMD8-92 or 1 μM PD184352 as indicated for one hour. Cells were then stimulated with EGF for 17 min and BMK1 activation was detected by mobility retardation (Abe et al., 1996). ERK1/2 activation was detected by an anti-phospho-ERK1/2 (T202/Y204) antibody.
(C) HEK293 cells were co-transfected with expression plasmids of MEK5D and BMK1. After 48 hr, BMK1 was immunoprecipitated and in vitro kinase assays were performed in the presence of the indicated amount of XMD8-92. The ATP concentration was measured by the Kinase-Glo® Luminescent Kinase Assay Platform. Kinase activity is expressed relative to the kinase activity in cells without XMD8-92 treatment, which was taken as 1. n = 3, ± SEM.
(D) Expression plasmids of MEK5D and BMK1 were transfected into HEK293 as indicated. After 36 hr, these cells were co-transfected with vectors for pCMV-β and the reporter plasmid pG5ElbLuc along with a GAL4 fusion expression vector containing MEF2C for 3 hr and then treated with 5 μM XMD8-92 for a further 16 hr. The luciferase activities were normalized against cells transfected with pG5ElbLuc and pGAL4 reporter plasmid alone, whose value was taken as 1. n = 3, ± SEM, *p < 0.01.
(E) HeLa cells were infected with Ad-EV or Ad-BMK1(AEF), as indicated, 24 hr prior to treatment with or without XMD8-92 (5 μM) for 48 hr, followed by MTT assays. % growth inhibition = (1- MTT value of cells in each experimental group/MTT value of cells treated with Ad-EV only) × 100. n = 3, ± SEM.
(F) Immunofluorescent analysis of CD31 (red) and TUNEL (green) in heart sections from control and XMD8-92 treated. DNase treatments were used as positive controls for TUNEL staining. Scale bar, 100 μm. n = 6 mice, n = 10 slices. (see also Figure S1, Table S1, S2, S3, S4 and S5)

(A) Cellular BMK1 was immunoprecipitated from cell lysates from A549 and HeLa cells with anti-BMK1 antibody or rabbit IgG, as control. The resultant immunoprecipitates and total cell lysates were then tested for the presence of PML using anti-PML antibody.
(B) Endogenous PML was immunoprecipitated from cell lysates of A549 and HeLa cells with anti-PML antibody or rabbit IgG, as control. The existence of BMK1 in these immunoprecipitates was analyzed using western blots as described in (A).
(C) HeLa and A549 cells were transfected with control or BMK1-specific siRNA. Cell lysates were analyzed by western blot using anti-p21 or anti-GAPDH antibodies as noted. n = 3, ± SEM, *p < 0.01.
(D) Pml^+/+ MEF and Pml^−/− MEF cells were treated with or without XMD8-92 (5 μM) for 16 hours as indicated. Cell lysates were analyzed by western blot using anti-p21 antibody. The levels of p21 protein expression were quantified by densitometry and normalized against GAPDH. The amount of the p21 protein in the Pml^+/+ MEF without XMD8-92 treatment was taken as 1. n = 5, ± SEM, *p < 0.01. (see also Figure S2 and Table S6)

(A) A schematic representation of wild-type and mutant PMLs. The amino acid sequences of potential BMK1 phosphorylation sites in PML are indicated.
(B) Mutation of potential BMK1 sites in full-length PML reduces phosphorylation by BMK1. Equal amounts of full-length wild-type or mutant recombinant PMLs were subjected to in vitro BMK1-mediated phosphorylation assay of PML in the presence of activated recombinant BMK1. The levels of relative [γ³²P]-incorporation in PML or PML mutants were quantified by densitometry and normalized against the loading amount of PML or PML mutants in each reaction mix, respectively. The value of relative [γ³²P]-incorporation of the wild-type PML by BMK1 was taken as 100%. n = 3, ± SEM, *p < 0.01.
(C) Pml^−/− MEF cells were infected with Ad-EV, Ad-PML or Ad-PML2D as indicated. Three days later, these cells were treated with or without XMD8-92 (5 μM) for 16 hours as noted. The expression levels of p21 were quantified as described in Figure 2D except that the amount of the p21 protein in the Ad-EV-infected Pml^−/− cells without XMD8-92 treatment was taken as 1. n = 3, ± SEM, *p < 0.01. (See also Figure S3)

(A) Fluorescent microscopy images of HeLa cells transfected with expression vectors encoding BMK1, PML, MEK5A or MEK5D, as indicated. These cells were stained with anti-BMK1 (green, first-column panels) or with anti-PML (PG-M3) antibody (red, second-column panels) as noted. Nuclei were shown by DAPI staining (blue, third-column panels). Merged images (yellow, fourth-column panels). n = 5. Scale bar, 10 μm.
(B) Fluorescent microscopy images of HeLa cells transfected with expression vectors encoding the N-terminal kinase domain of BMK1 (BMK1-KD), the C-terminal Non-Kinase domain of BMK1 (BMK1-NKD) and/or PML, as indicated. These cells were stained with anti-BMK1 (green, first-column panels) or with anti-PML (PG-M3) antibody (red, second-column panels) as noted. Nuclei were shown by DAPI staining (blue, third-column panels). Merged images (yellow, fourth-column panels). n = 5. Scale bar, 10 μm. (see also Figure S4)

(A) A549 cells were injected subcutaneously into the flanks of Nod/Scid mice. After three weeks, these tumor-bearing mice were injected intraperitoneally with XMD8-92 or vehicle for two days. The XMD8-92 concentration in these tumors were analyzed and shown in the supplemental information (See also Figure S5). The tumor homogenates were used to detect both BMK1 activity and p21 expression using western blot as described in Figure 2D.
(B) Mouse xenograft models were established as described in Experimental Procedures. (Top panels) Starting one day or at the indicated time after injection of the human (HeLa) or mouse (LL/2) tumor cells, XMD8-92 was administered twice daily and the tumor growth in these mice was measured every two or three days. (Lower panels) Representative mice from each group (control mice on the left) showing the growth of xenograft tumors at the end of an experiment. n = 6 mice, ± SEM, *p < 0.01.
(C) Fluorescence microscopy images of HeLa or LL/2 tumor sections from XMD8-92 treated or control mice as indicated. The sections were stained with anti-BrdU antibody or DAPI as noted. n = 6 mice, n = 9 slices. Scale bar, 10 μm.
(D) Mice were implanted with Matrigel containing 400 ng/mL bFGF. The next day, mice were treated with XMD8-92 or vehicle for 6 days. The Matrigel plugs were then harvested, sectioned and stained with Masson’s trichrome (n = 6 mice, n = 9 slices.). Hemoglobin content in the Matrigel plugs was assessed as described (Hayashi et al., 2005). n = 6 mice, ± SEM, *p < 0.01. Scale bar, 100 μm.

(A) Mouse tumor model with syngeneic LL/2 carcinoma was established as described in Experimental Procedures. Starting seven days after injection of LL/2 tumor cells into mice, the mice were treated with intratumoral injection of recombinant adenovirus encoding BMK1(AEF) [Ad-BMK1(AEF)] or empty vector control (Ad-EV) as indicated. One day after starting the injection of recombinant adenoviruses, XMD8-92 was administered to the indicated mice. n = 6 mice, ± SEM, *p < 0.01.
(B) Mouse tumor models with PML-knockdown (PML-KD) and control LL/2 carcinoma (Ctrl) were established (middle panel) similar to syngeneic LL/2 tumor model. Starting seven days after injection of the LL/2 tumor cells, the mice were treated with XMD8-92 or vehicle for another nine days as described in Figure 7A. (Bottom panel) Mice bearing PML-KD LL/2 tumors for seven days were injected with recombinant adenovirus encoding wild type PML [Ad-PML], mutant PML [Ad-PML2D] or empty vector control [Ad-EV] intratumorally. One day after starting the injection of recombinant adenoviruses, XMD8-92 or vehicle was administered to the indicated mice. n = 6 mice, ± SEM, *p < 0.01.