1

Kim et al.

SupplementalMaterials and Methods

Affymetrix gene chip analysis

Total RNA was isolated from either P21 wildtype or PTEN-deficient RPE cells, and analyzed using Affymetrix Mouse Gene Expression microarrays (Supplemental Table S1).

Differential fixation and extraction

RPE cells were isolated from 2 month-old C57/BL6mice and grown on 24 well Matrigel transwell culture plates as described in Materials and Methods. Cells were washed with PBS-CM three times prior to incubation in 0.5ml 1% Triton X-100 in PBS-CM for 10 min at 4oC with gentle rocking. Cells were then washed twice with PBS-CM, and incubated in 4% paraformaldehyde in PBS-CM for 15 min followed by PBS-CM washing and immunostaining. Alternatively, the cells were extracted with 0.1 ml of 1% Triton X-100/PBS-CM for 10 min at 4oC under gentle rocking, and then the remaining cells were washed with PBS-CM once before being lysed in 0.12 ml SDS sample buffer for the SDS-PAGE and Western blotting.

-gal staining

Anesthetized two week-old fl/fl;BAT-gal or fl/fl;T1Cre;BAT-gal mice were perfused with PBS, and isolated eyes were fixed with 0.5% glutaldehyde/4% paraformaldehydein PBS for 30 minon ice. After incubation in 2% sucrose/PBS for 12 hours at 4oC, eyes were embedded in OTC medium to freeze. Frozen blocks were sectioned to 14 m thickness before X-gal staining as described (Mui et al., 2005).

Supplemental Figure Legends

FigureS1. Retinal degeneration in fl/fl;T1Cre mice.

Sections (10 m) of fl/fl or fl/fl;T1Cremouse eyes from one week-old (P8), one month-old (1mon), or three month-old (3mon) were stained withmouse monoclonal antibody against rhodopsin (A), a marker for rod photoreceptors, or blue opsin (B), a marker for cone photoreceptors.The nuclei of cells were counter-stained withDAPI(blue).Scale bar indicates 100 m.

Figure S2. Normal development of retina in fl/fl;T1Cremice.

Sections (10 m) of P8 fl/fl or fl/fl;T1Cremouse eyes were stained with the indicated primary antibodies, and monitored for the expression of retinal cell type markers (see Supplemental Table S2 for more information). Brn3b, a retinal ganglion cell marker; Pax6, an amacrine cell marker; Calbindin, a marker for horizontal cells and amacrine cell subset; protein kinase C alpha (PKC), a bipolar cell marker; neurofilament 160 (NF160), a horizontal cell marker; recoverin, an early photoreceptor marker; glial fibrillary acidic protein (GFAP), a marker for astrocytes and activated Müller glia; glycogen synthase (GS), a Müller glial cell marker.The nuclei of cells were counter-stained withDAPI(blue).Scale bar indicates 100 m.

Figure S3. Retinal gliosis in fl/fl;T1Cre mice.

Sections (10 m) of 3 month-old fl/fl (A and C) or fl/fl;T1Cre(B and D) mouse retina were stained with glial cell markers, such as GFAP (red in A and B) or GS (green in C and D). The nuclei of cells were stained withDAPI (blue).Scale bar indicates 100 m.

Figure S4. Impaired formation of RPE adherens junctions in PTEN-deficient RPE.

The distribution of markers for adherens junction,-catenin (A and B) or E-cadherin (C and D); a marker for apical microvilli, Ezrin (green in E and F); ormarkers for tight junctions,ZO1 (green in G and H) or Crb3 (green in I and J),in fl/fl (A, C, E, G, and I)orfl/fl;T1Cre(B, D, F, H, and J) mouse RPE cells were assessed by immunostaining.Scale bar is50 m.

Figure S5. Subcellular distribution of PTEN in mouse RPE.

Subcellular locations of PTEN in frozen sections (10 m) of post-natal C57/BL6 mouse eyes were examined by immunostaining with anti-PTEN antibody (green in A – F) with nuclear DAPI counter staining (blue in B, D, and F). PTEN was distributed evenly in neonatal RPE cells (A and B), but startedto concentrate at the apical region about P8 (C and D). The apically localized PTEN wasmaintained in P28 RPE cells (E and F) and afterward. (G and H) The 3-dimenstional localizationof PTEN in the primary cultured mouse RPE cellswas examined by immunostaining. Confocal microscopic image of PTEN at theapical (G) or the basalside of RPE cells (H) provided differential immunostaining pattern of PTEN at each focal plane. The basolaterally concentrated PTEN appeared as puncta in yz axis (arrow heads in I). The nuclei of cells in the z section were co-stained with DAPI(blue).The basolateral localization of PTEN was further confirmed by the differential detergent extraction procedure (J – L). Cellular PTEN was diffusely expressed in primary cultured RPE cells (J), but was more concentrated at the basolateral area if the cells were pre-extracted by Triton-X100 first, and then fixed for immunostaining (K). The relative amount of PTEN protein at the basolateral detergent-resistant fraction (P in L) was much lower than PTEN in the detergent-extractable apical and cytosolic fraction (S in L). The successful separation of detergent extraction was confirmed by the exclusive presence of E-cadherin in the P fraction in contrast to S fraction-enriched Akt. The amount of each protein in total cell lysates (L in L) were also examined. See Supplemental Materials and Methods for detailed protocols.Arrow heads indicate PTEN locating at intercellular junctional areas. Asterisks (*) indicate microvilli of RPE cells. Scale bars in A and G are 20 m.

Figure S6. Recovery of junctional -catenin in PTEN-deficient RPE by ectopic expression of PTEN variants.

The number of PTEN-deficient cells with junctional -catenin (red in Figure 4), among the cells infected with retroviruses expressing human PTEN variants (green in Figure 4), wasdetermined. Results were obtained from the combined numbers of six different samples for each PTEN variant infection.

Figure S7. Metastatic tumorigenesis in fl/fl;T1Cremice.

Lymphnodes(LN; A) or spleens (C) from 6 month-old fl/fl orfl/fl;T1Cre miceare shown in A and C. B and D show H&E staining of LN tumor (indicated with an arrowhead in A) and pigmented spleen (indicated with an arrowhead in C), respectively. Pigmented cells are indicated by white arrowheads in D. (E) Development of LN tumors in fl/fl (open bar) orfl/fl;T1Cre (closed bar) mice. (F) The Cre-mediated deletion between two loxP sites in floxed alleles, which can be detected as the 849bp PCR band (4-5; upper panel), were only seen in a LN tumor from fl/fl;T1Cre mice (LN* in lane 2), but not in the LNsfromT1Cre mice (LN in lane 1) or the normal LNs from fl/fl;T1Cre littermate (LN in lane 3). The 849bp PCR bandin Lane 4 was obtained from genomic PCR of P20 fl/fl;T1Cremouse RPE cell chromosomal DNA. The bottom panel is the genotyping PCR results of corresponding tissues.

Figure S8. Elevated -catenin/TCF transcription activity in PTEN-deficient RPE cells.

Frozen sections (14 m) from two week-old fl/fl;BAT-gal or fl/fl;T1Cre;BAT-gal mouse eyes were stained with X-gal as described in the Supplemental Materials and Methods to detect nuclear lacZ (BAT-gal) expression. (A) BAT-gal-positive nuclei were barely detected in fl/fl;BAT-gal mice, while blue nuclear lacZ signals (arrow heads) were evident in the fl/fl;T1Cre;BAT-gal mouse RPE layer. The number of BAT-gal-positive cells among 200 fully differentiated RPE cells was also counted, and shown as a graph in the bottom panel. (B) BAT-gal-positive cells in the ciliary body (CB) were clearly detected in both mouse eyes (arrow heads), and the number of these cells was not significantly different. Numbers shown in the graphs are results from four different samples.

Figure S9. The perturbed interaction between E-cadherin and tyrosine phosphorylated -catenin in PTEN-deficient RPE.

(A) The level of tyrosine-phosphorylated -catenin in the immunecomplexes, isolated by immunoprecipitation with rabbit polyclonal antibodies against -catenin (-catenin pAb), was examined by Western blotting with 4G10 mouse anti-phospho-tyrosine antibodies (upper panel). The same immunoblot was reprobed with -catenin pAb to confirm the relative amount of immunoprecipitated -catenin (bottom panel). (B) Affinity of -catenin to E-cadherin was examined by incubating RPE cell lyates from fl/fl or fl/fl;T1Cre mice with GSH-agarose attached with GST-fused E-cadherin cytoplasmic domain (734-884) recombinant protein. The level of -catenin pulleddown together with GST-E-cadherin was assessed by Western blotting with -catenin pAb. The 10% of total -catenin in each cell lysateused for the experiments (input) is shown in the bottom panel.

Table S1. Gene expression comparison between wildtype versus PTEN-deficient RPE cells.

RPE cells were isolated from 3 week-old (P21) wildtype or fl/fl;T1Cremice (see Supplemental Materials and Methods), and total RNA was isolated to produce cDNA to hybridize to Affymatrixtm Mouse Gene Expression Chip. The results represent the mean value of the selected results from two independent experiments.

Table S2. Antibodies used in this study

1