We employed naïve H4 human neuroglioma (H4) cells and H4 cells stably transfected to express either FL-APP (H4-FL-APP cells) or APP-C99 (H4-APP-C99 cells). Peptide APP-C99 is the product of β-secretase, which therefore contains α- and γ-, but not β-cleavage sites. The H4-APP-C99 cells provide a valid system to assess whether any effects on APP processing are dependent on γ-secretase-mediated APP processing and independent of β-secretase-mediated APP processing. All cell lines were cultured in DMEM (high glucose) containing 9% heat-inactivated fetal calf serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine. Stably transfected H4 cells were additionally supplemented with 200 μg/ml G418.

Small interfering RNA (siRNA) duplex was designed and obtained from Qiagen against human NUMB, the gene encoding Numb (5'- CAGCCTCTTGACCTCGGATAA-3'). Dab siRNA duplex for DAB, the gene encoding Dab, was designed and obtained from Dharmacon research, Inc. (Lafayette, CO 80026) (5'-NNAGGUCAGGAUCGCAGUGAA-3'). Scrambled siRNA (AATTCTCCGAACGTGTCACGT) was obtained from Qiagen and was used as the control siRNA. siRNAs were transfected into cells by using electroporation (AMAXA, Gaithersburg, MD). We mixed 1 million cells, 100 ul AMAXA electroporation transfection solution and 10 ul 20 uM siRNA together, then we employed C-9 program in the AMAXA electroporation device for the cell transfection. We chose 20 uM siRNA of NUMB and DAB in current studies because our previous studies have shown that 20 uM siRNA of other APP adaptor protein genes, including APBA1 17, SHC3 16 and APBB1 16, can affect APP processing and reduce Aβ levels. The transfected cells then were placed in one well of a six-well plate containing 1.5 ml cell culture media. The cells were harvested 48 hours after siRNA treatments.

Cell pellets were detergent-extracted on ice using immunoprecipitation buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40) plus protease inhibitors (1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin A). The lysates were collected, centrifuged at 12,000 rpm for 10 min, and quantified for total proteins by the BCA protein assay kit (Pierce, Iselin, NJ).

Western blot analysis was performed as described by Xie et al. 25. Briefly, 40 μg of total protein of each sample was subjected to SDS-polyacrylamide gel electrophoresis using 4-20% gradient Tris/glycine gels under reducing conditions (Invitrogen, Carlsbad, CA). Next, proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA) using a semi-dry electrotransfer system (Amersham Biosciences, San Francisco, CA). Nonspecific proteins were blocked using 5% non-fat dry milk in TBST for 1.5 h. Blots were then incubated with a primary antibody, followed by a secondary antibody (horseradish peroxidase-conjugated anti-rabbit antibody 1:10,000; Pierce, New York, NY). Blots were washed with TBST for 30 min between steps. Antibody Dab (1:1,000, Novus Biologicals, Littleton, CO) was used to recognize Dab (35 kDa), antibody Numb (1:1,000, Abcam, Cambridge, MA) was used to detect Numb (75 kDa). Antibody A8717 (1:1,000, Sigma, St. Louis, MO) was used to visualize FL-APP (110 kDa), APP-C83 (12 kDa) and APP-C99 (10 kDa) in the Western blot analysis. The intensity of signals was analyzed using an image program (NIH Image 1.62). We first used the levels of β-actin to normalize the levels of Numb, Dab, FL-APP and APP-CTFs (e.g., determining the ratio of Numb amount to β-actin amount) to control for loading differences in total protein amounts. We then presented the changes in the protein levels of Numb, Dab, FL-APP, APP-C99 and APP-C83 in the cells treated with Numb or Dab siRNA as the percentage of those in the cells treated with control siRNA.

Following the treatment with control siRNA, Numb siRNA or Dab siRNA, conditioned media was collected, and secreted Aβ was measured with a Sandwich ELISA assay by the Aβ ELISA Core Facility at the Center for Neurological Diseases, Harvard Institute of Medicine, Harvard Medical School, Boston, Massachusetts, as described by Xie et al. 16. Specifically, 96-well plates were coated with mouse monoclonal antibodies (mAb) specific to Aβ40 (2 G3). Following blocking with Block Ace, wells were incubated overnight at 4°C with test samples of conditioned cell culture media, and then an anti-Aβ (α-Aβ-HR1) conjugated to horseradish peroxidase was added. Plates were then developed with TMB reagent and well absorbance was measured at 450 nm. Aβ levels in test samples were determined by comparison with the signal from unconditioned media spiked with known quantities of Aβ40.

ANOVA with repeated measurements was employed to compare the difference from the control group. P-values less than 0.05 were considered statistically significant.

We first established conditions under which RNAi knock-down of Dab significantly reduced protein levels of Dab in H4-FL-APP cells. The cells were harvested 48 hours after being transfected with either control siRNA or Dab siRNA, and were subjected to Western blot analyses in which antibody Dab was used to visualize Dab levels. Immunoblotting for Dab revealed a visible reduction in Dab levels following Dab siRNA treatment as compared to control siRNA treatment (Figure 1A). Dab siRNA treatment significantly reduced Dab levels by 44% (normalized to β-actin) as compared to control siRNA treatment (Figure 1B, *p < 0.05).

Next, we assessed effects of RNAi-mediated knock-down of Dab on APP processing in H4-FL-APP cells by measuring protein levels of FL-APP, APP-C99 and APP-C83 following Dab or control siRNA treatments. 48 hours after transfection of Dab siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses in which antibody A8717 was used to detect FL-APP, APP-C99 and APP-C83. Immunoblot analysis of APP-CTFs revealed increases in levels of APP-C99 and APP-C83 in the cells treated with Dab siRNA as compared to the cells treated with control siRNA (Figure 1A). Meanwhile, no significant differences in FL-APP levels were observed between Dab siRNA- and control siRNA-treated cells. Quantification of FL-APP, APP-C99 and APP-C83 (normalized to β-actin) revealed that Dab siRNA treatment led to a 207% increase in ratio of APP-C99 to FL-APP (Figure 1C, *p < 0.05), and a similar (176%) increase in ratio of APP-C83 to FL-APP (Figure 1D, *p < 0.05), as compared to control siRNA treatment. Note the different parts of Western blot representing Dab were obtained from same samples of the experiments.

Next, we assessed effects of Dab siRNA on Aβ levels in conditioned media. 48 hours after treatment with control siRNA or Dab siRNA, we measured secreted Aβ40 levels in conditioned cell culture media. Dab siRNA treatment significantly decreased Aβ40 levels as compared to control siRNA treatment (Figure 1E, *p < 0.05): 34 pg/ml (Dab siRNA) versus 57 pg/ml (control siRNA). Collectively, these data indicate that RNAi knock-down of Dab increases levels of APP-C99, APP-C83, and decreases secreted Aβ in H4-FL-APP cells, in a manner similar to that of γ-secretase inhibitor treatment.

We next assessed effects of RNAi knock-down of Numb, another APP adapter molecule, on APP processing and Aβ levels in H4-FL-APP cells. We first established conditions under which Numb siRNA treatment reduced protein levels of Numb in H4-FL-APP cells. The cells were harvested 48 hours after being transfected with either control siRNA or Numb siRNA, and were subjected to Western blot analyses with Numb antibody to measure protein levels of Numb. Numb immunoblotting revealed a 54% reduction in protein levels of Numb following Numb siRNA treatment as compared to control siRNA treatment (Figure 2A, B, *p < 0.05).

We then assessed effects of RNAi-mediated knock-down of Numb on APP processing in H4-FL-APP cells by measuring levels of FL-APP, APP-C99 and APP-C83 following Numb siRNA or control siRNA treatments. 48 hours after transfection of Numb siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses with antibody A8717 to detect FL-APP, APP-C99, and APP-C83. Levels of APP-C99 and APP-C83 were increased in the cells treated with Numb siRNA versus control siRNA (Figure 2A). Meanwhile, no significant differences in FL-APP levels were observed for Numb siRNA versus control siRNA-treated cells (Figure 2A). Numb siRNA treatment led to a 202% increase in ratio of APP-C99 to FL-APP (Figure 2C, *p < 0.05), but only a mild (122%) increase in ratio of APP-C83 to FL-APP (Figure 2D, N.S.), normalized to β-actin, as compared to control siRNA treatment.

Next, we measured secreted Aβ levels in conditioned cell culture media. 48 hours after treatment with control siRNA or Numb siRNA in H4-FL-APP cells, Numb siRNA decreased Aβ40 levels: 34 pg/ml for Numb siRNA versus 48 pg/ml for control siRNA (Figure 2E, *p < 0.05). Collectively, these data indicate that RNAi knock-down of Numb also affects APP processing and Aβ production in a manner similar to that of γ-secretase inhibitor treatment in H4-FL-APP cells.

As discussed in prior section, β-secretase cleaves FL-APP to produce APP-C99, which can be cleaved by γ-secretase to produce Aβ. Therefore, changes in APP processing and Aβ production following treatments of Dab siRNA and Numb siRNA could be due to alterations in either β-secretase or γ-secretase activities. In the following experiments, we set out to determine the extent to which the alterations in APP processing and Aβ production following RNAi knock-down of Dab or Numb were independent of β-secretase-mediated APP processing and dependent on γ-secretase-mediated APP processing, employing H4 cells over-expressing APP-C99 (H4-APP-C99 cells).

We employed H4-APP-C99 cells in order to determine whether RNAi knock-down of Dab-induced alterations in APP processing and Aβ levels were independent of β-secretase-mediated APP processing. As described in the prior section, APP-C99 is the product of β-secretase and harbors α- and γ-cleavage, but not β-cleavage sites, therefore H4-APP-C99 cells provide a valid system to assess whether any effects on APP processing are dependent on γ-secretase-mediated APP processing and independent of β-secretase-mediated APP processing.

48 hours after transfection of H4-APP-C99 cells with Dab siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses in which Dab antibody was used to detect Dab. Dab immunoblotting revealed a significant reduction in protein levels of Dab in the cells treated with Dab siRNA as compared to the cells treated with control siRNA (Figure 3A). There was no significant difference in amount of β-actin in control siRNA- and Dab siRNA-treated cells. Quantification of Dab in the Western blot, normalized to β-actin, showed that Dab siRNA treatment caused a 65% reduction in protein levels of Dab (Figure 3B, *p < 0.05).

We next assessed effects of RNAi-mediated knock-down of Dab on APP processing. 48 hours after transfection with Dab siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses with antibody A8717. Dab siRNA treatment did not alter endogenous levels of FL-APP as compared to control siRNA treatment (data not shown). APP-CTFs immunoblotting revealed visible increases in protein levels of both APP-C99 and APP-C83 in the H4-APP-C99 cells treated with Dab siRNA, compared to control siRNA (Figure 3A). There was no significant difference in amount of β-actin in the control siRNA- or Dab siRNA-treated H4-APP-C99 cells. Quantification of FL-APP, APP-C99 and APP-C83 revealed that Dab siRNA treatment led to a 188% increase in ratio of APP-C99 to FL-APP (Figure 3C, *p < 0.05) and a 199% increase in ratio of APP-C83 to FL-APP (Figure 3D, *p < 0.05), as compared to control siRNA treatment.

We then measured secreted Aβ levels in conditioned cell culture media 48 hours after treatment with either control siRNA or Dab siRNA in H4-APP-C99 cells. Dab siRNA decreased Aβ levels as compared to control siRNA treatment: 63.5 pg/ml (Dab siRNA treatment), 144.5 pg/ml (control siRNA treatment) (Figure 3E, *p < 0.05). These findings suggest that effects of RNAi knock-down of Dab on APP processing and Aβ generation are dependent on γ-secretase- and independent of β-secretase-mediated cleavage of APP.

Next, we asked whether alterations in APP processing and Aβ levels induced by RNAi knock-down of Numb were also independent of β-secretase-mediated APP processing. For this purpose, we set out to determine effects of RNAi knock-down of Numb on APP processing and Aβ levels in H4-APP-C99 cells. We first established RNAi knock-down of Numb in H4-APP-C99 cells by showing that RNAi knock-down of Numb reduced protein levels of Numb in H4-APP-C99 cells (Figure 4A, B, a 46% reduction, *p < 0.05).

We then assessed effects of RNAi knock-down of Numb on levels of APP-C83, APP-C99 and Aβ in H4-APP-C99 cells. 48 hours after transfection with Numb siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses with antibody A8717. Numb siRNA treatment increased protein levels of both APP-C99 and APP-C83 in H4-APP-C99 cells, without altering endogenous FL-APP levels (data not shown), as compared to control siRNA treatment (Figure 4A). Quantification of the Western blot revealed that Numb siRNA treatment led to a 209% increase in the ratio of APP-C99 to FL-APP (Figure 4C, *p < 0.05) and only a 130% increase in ratio of APP-C83 to FL-APP (Figure 3D, N.S.), as compared to control siRNA treatment. Numb siRNA treatment decreased secreted Aβ levels as compared to control siRNA treatment in H4-APP-C99 cells: 44 pg/ml (Numb siRNA treatment) versus 68 pg/ml (control siRNA treatment) (Figure 4E, *p < 0.05). These findings suggest that RNAi knock-down of Numb may affect APP processing and Aβ production at least partially by inhibiting γ-secretase-, but not β-secretase-, mediated cleavage of APP, a manner similar to that of γ-secretase inhibitors.

A recent study by Chapman et al. 24 showed that high levels of Notch can decrease Numb and Numblike. Both APP and Notch are substrates of γ-secretase, we therefore assessed effects of APP on Numb levels in H4-APP cells. 48 hours after transfection with APP siRNA or control siRNA, the cells were harvested and subjected to Western blot analyses with antibodies anti-Numb and A8717. We first showed that APP siRNA treatment decreased protein levels of both FL-APP and APP-CTFs (Figure 5A), suggesting that the APP siRNA treatment can reduce levels of APP (both FL-APP and APP-CTFs) in H4-APP cells. Then we were able to show that the same APP siRNA treatment also reduced levels of Numb in H4-APP cells (Figure 5A). Quantification of the Western blot revealed that APP siRNA treatment led to a 55% (Figure 5B, **p < 0.01), 41% (Figure 5C, **p < 0.01) and 25% (Figure 5D, **p < 0.01) reduction in levels of FL-APP, APP-CTFs and Numb, respectively, as compared to control siRNA treatment. These findings suggest that RNAi knock-down of APP may affect Numb metabolism as well.

Finally, We found that control siRNA did not affect APP processing or alter Aβ levels as compared to saline treatment in H4-FL-APP cells or H4-APP-C99 cells (data not shown). These results confirmed that effects of RNAi knock-down of Dab or Numb on APP processing and Aβ levels in our experiments were not due to control siRNA (scrambled siRNA) effects, but owing to reductions in protein levels of Dab and Numb.