Supplemental Figure S1. Bcl2L12 expression in normal and neoplastic brain tissue.

TMA analysis ofBcl2L12 protein expression, as detected by anti-L12-2 antiserum, in formalin-fixed paraffin-embedded tissue derived from infiltrating low-grade glioma with normal entrapped neurons (a), normal brain (b) adjacent to neoplastic tissue and primary GBM (c)-(h). White arrowheads indicate neurons staining positive for Bcl2L12. Normal glia and low-grade glioma cells are negative. Blue arrowheads indicate weak Bcl2L12 positivity in blood vessels of normal brain. Red arrowheads in (c) and (d) indicate Bcl2L12-positive vascular endothelial cells. Pictures for (b) and (e) were taken from the same case. Framed areas in (a) to (d) were enlarged in middle row as indicated.

Supplementary Figure S2. Domain structure of the Bcl2L12 polypeptide.

(A) Schematic representation of the domain structure of Bcl2L12. BH2, Bcl-2 homology domain 2; P, proline-rich region. The positions of PxxP tetrapeptide motifs are indicated.

(B) BH2 sequence alignment of Bcl2L12 with members of the Bcl-2 family using the Lipman-Pearson algorithm. Identical amino acids are shown in black, conserved amino acid substitutions are highlighted in grey boxes. Using three different scoring matrixes (PAM30, PAM70 and BLOSUM80) in blastp searches, alignments between Bcl2L12 and Bcl2L14, its closest homologue, gave E-values of 0.008, 0.009 and 0.05, respectively. We also used the Pfam HMM protein family database that unlike pair wise alignment methods deals more sensibly with multidomain proteins, and again, statistically significant alignment was evident between Bcl2L12 and Bcl-2 with an E-value of 0.0014.

(C) Phylogenetic analysis of Bcl2L12 and BH2-containing Bcl-2 family members. Pro-apoptotic proteins are depicted in blue.

Supplementary Figure S3.Knockdown of Bcl2L12 increases intratumoral apoptosis with predominant activation of caspase-7.

Histological analyses of intracerebral tumors derived from U87MG cells stably expressing Bcl2L12-specific shRNAs. Tumor sections were immunohistochemically stained for TUNEL and active caspases 3 and 7 according to standard protocols (see Materials and Methods). Pictures were taken at a 40x magnification and framed areas were enlarged to document details of the immunostainings.

Supplementary Figure S4. Subcellular localization of Bcl2L12.

(A) Endogenous Bcl2L12 localizes to nuclear and granular cytosolic structures. U87MG shNT- and shL12-1-infected cells were subjected to deconvolution immunofluorescence microscopy using the polyclonal anti-L12-2 serum with AlexaFluor 594-labeled goat-anti-rabbit secondary antibodies. Nuclei were visualized using DAPI-staining (left column) and overlayed with the AlexaFluor 594 stainings (merge, right column). Bar = 15 m.

(B)Ink4a/Arf-deficient astrocytes expressing Bcl2L12V5 or pBabe were subjected to deconvolution immunofluorescence microscopy using a monoclonal anti-V5 (Invitrogen) antibody with AlexaFluor 594-labeled goat-anti-mouse secondary antibodies. Nuclei were visualized using DAPI-staining (left column) and overlayed with the AlexaFluor 488 stainings (merge, right column). Bar = 15 m.

(C) LNZ308 cells were co-stained for Bcl2L12 (left column) and mitochondria (center column) using an anti-cytochrome c antibody (cyt c; Bar 15 m). Lower panel: area framed by a white box in upper panel was enlarged.

(D) LNZ308 cells were stained for Bcl2L12 as above and co-stained for Golgi apparatus (anti-Golgin 97 antibody), early endosomes (anti-EEA-1 antibody) and endoplasmic reticulum (anti-cytochrome P450 reductase antibody). Bcl2L12, organelle and DAPI stainings were overlaid (Merge/DAPI, right column). Bar = 15 m. Co-staining experiments were also performed using a protein disulfide isomerase (PDI) monoclonal antibody as an additional ER-specific marker, which similarly revealed a lack of co-localization between Bcl2L12 and ER structures (data not shown).

(E) LNZ308 cells were treated with STS [1 M] for the indicated periods of time and subjected to deconvolution microscopy using the anti-L12-2 antiserum (left columns) and the anti-cytochrome P450 reductase antibody (center column). The overlay with DAPI stainings is presented in the right column. Bar = 15 m. Note orange color in merge/DAPI images, indicating close proximity of green and red signals without overlap.

Supplemental Figure S5. The caspase-7 knockdown phenotype.

(A) Ink4a/Arf-/- astrocytes were transfected with non-targeting control siRNA oligonucleotide (siNT) and caspase-7-specific (siCasp-7) siRNA oligonucleotides (SmartPool) and subjected to Western blot analysis. The migration positions of pro-caspase-7 and Hsp70 as a loading control are indicated.

(B) siNT and siCasp-7-transfected Ink4a/Arf-/- astrocytes were stimulated with STS [1 M] for 0, 8 and 16 hrs and subjected to transmission electron microscopy. Control cells stimulated with STS for 16 hrs showed chromatin condensation (red arrowheads) and preserved plasma membrane integrity (white arrowheads), whereas siCasp-7-transfected cultures exhibited profound plasma membrane disintegration (yellow arrowheads) and were characterized by the absence of chromatin condensation and swollen subcellular organelles (blue arrowheads). Bar, 2 m.

(C) Knockdown of caspase-7 results in a decreased apoptotic and an increased necrotic index. ~50 cells were counted per grid (total 3 grids) to assess the fraction of apoptotic (cells with chromatin condensation) and necrotic cells (cells with disintegrated plasma membrane). Error bars represent standard deviations and two-tailed p values were calculated using Student’s t test.

1