Breast cancer cell lines MCF7, MDA-MB-468, SKBr3, T47D, ZR75.1 and immortalized mammary epithelial line MCF10A were purchased from American Type Culture Collection (Manassas, VA, USA). All cell lines were cultured in recommended media.

Cultured Zr75.1, MCF10A and MEF cells were incubated with serum-free media for 1 day. They were then treated with PDGF-BB (0.5 ng/ml; Millipore, Billerica, MA, USA) for 0 to 120 minutes before cells were harvested in 1× SDS sample buffer.

NHERF1^+/- mice 31 were inbred to generate littermates of three genotypes. Duplex PCR was used to genotype NHERF1 knockout mice. A common forward primer (5'-ctctgtttattcccagaagga-3') was included in the PCR reaction, together with reverse primers for knockout (5'-caagctcttcagcaatatcac-3') and wild-type (5'-ggttctaccagacggataaac-3') genotypes that were expected to yield 2.4-kilobase and 1.4-kilobase products, respectively. PCR conditions were described previously 15.

To harvest mammary gland of NHERF1 knockout mice, 10-week-old female littermates were killed by carbon dioxide inhalation. Mammary glands were collected and snap frozen at -80°C until use. Tissues were then ground while frozen in liquid nitrogen and lysed in 1× SDS sample buffer. The lysates were then subjected to immunoblotting.

To obtain MEF cells of varied NHERF1 genetic backgrounds, the NHERF1^+/- mice were inbred. At embryonic days 12 to 14, the embryos were dissected to remove heads and internal organs. The remaining tissues were minced, trypsinized and plated in Dulbecco's medified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA). Genomic DNA was also harvested from MEF cells for genotyping. Early-passage MEF cells were then subjected to experimental treatment.

To prepare recombinant retrovirus to knockdown NHERF1 expression, an siRNA sequence that targeted NHERF1 transcript was inserted into pBabe-U6-Puro (a gift from Dr Jinsong Liu, M.D. Anderson Cancer Center, Houston, TX, USA). Two oligonucleotides (5'-ggaaactgacgagttcttcaaagctttgaagaactcgtcagtttccctttttg-3' and 5'-aattcaaaaagggaaactgacgagttcttcaaagctttgaagaactcgtcagtttcc-3') were annealed and subcloned into pBabe-U6, creating pBabe-U6/NHERF-910-Puro. Similarly, oligonuclotides (5'-ggacgaactggtgtaatgatatgaagcttcatatcattacaccagttcgtccctttttg-3' and 5'-aattcaaaaagggacgaactggtgtaatgatatgaagcttcatatcattacaccagttcgtcc-3') were used to knock down PTEN expression (pBabe-U6/PTEN-Puro).

GST-PTEN, pcDNA3.1-PTEN, and pcDNA3.1-PTEN-C124S were gifts from Dr Charles Sawyer (University of California, Los Angeles, CA, USA). To make a PTEN construct with deletion of the last six residues (TQITKV), PCR was conducted to introduce premature termination, using pcDNA3.1-PTEN as a template. This procedure created pcDNA3.1-PTEN-ΔC.

NHERF1 cDNA was obtained by reverse transcription from mRNA of a breast cancer cell line, SKBr3, and amplified by PCR. Reverse transcription PCR products were subcloned into a TA cloning vector (Invitrogen). NHERF1 cDNA was then subcloned into pcDNA3.1(+) (Invitrogen), generating pDNA3.1-NHERF1. NHERF1 cDNA was then enzymatically released from pcDNA3.1-NHERF1 and subcloned into pBabe-Puro, pBabe-Neo (Addgene, Cambridge, MA, USA), and pGEX2TK (GE Healthcare, Piscataway, NJ, USA), yielding pBabe-NHERF1-Puro, pBabe-NHERF1-Neo, and pGEX2TK-NHERF1, respectively. cDNA fragments encoding PDZ-I (residues 1 to 150), PDZ-II (residues 97 to 239), PDZ-I&II (residues 1 to 239), and carboxyl-terminus (residues 231–358) of NHERF1 were PCR-amplified and constructed to pcDNA3.1 or pGEX2TK. All constructs generated by PCR were verified by automated DNA sequencing.

Retroviral stocks were made by transfecting packaging cells (amphotropic Phoenix cells) with retroviral constructs (pBabe-U6-Puro, pBabe-U6-NHERF-910-Puro, pBabe-U6-PTEN-Puro, pBabe-Puro, pBabe-NHERF1-Puro, pBabe-Neo, and pBabe-NHERF1-Neo), using Fugene 6 (Roche Applied Science, Indianapolis, IN, USA).

To prepare MCF10A cells over-expressing NHERF1, MCF10A cells were infected with NHERF1-Puro (or NHERF1-Neo for PTEN knockdown cells) recombinant retrovirus. Infected cells were selected by adding 1 to 5 μg/ml puromycin (or 600 μg/ml G418) to culture media. Surviving cells were assessed for NHERF1 expression by immunoblotting. MCF10A-NHERF1-Neo cells were further infected with PTEN-siRNA retrovirus to knockdown PTEN expression, which was determined by PTEN immunoblotting.

The expression of NHERF1 in Zr75.1 and MDA-MB-468 cells was knocked down by using retrovirus-based siRNA method. Zr75.1 cells were infected with Babe-U6-NHERF-910-Puro retrovirus and subjected to puromycin (2 μg/ml) selection. NHERF1 expression was determined by immunoblotting.

Glutathione S-transferase (GST) pull-down assays were used to assess the interaction of NHERF1 and PTEN. GST-PTEN or GST-NHERF1 and their derivatives were induced by isopropyl-β-D-1-thiogalactopyranoside in pGEX2TK-transformed BL21 strain and purified with glutathione-Sepharose beads (GE Healthcare).

Radio-labeled full-length or defined segments of NHERF1 and full-length or truncated PTEN were synthesized by in vitro transcription from pcDNA3.1 plasmids containing the NHERF1 or PTEN cDNA and translated in the presence of [³⁵S]methionine (T7 Quick TNT kit; Promega, Madison, WI, USA). The translation products were mixed with purified GST-PTEN or GST-NHERF1 immobilized on the beads. Pull-down assays were performed at 4°C for 1 hour in 1× binding buffer (20 mmol/l Tris [pH 7.5], 150 mmol/l NaCl, and 1% NP-40). The beads were then washed thoroughly with 1× binding buffer. Bound proteins were eluted by boiling in 1× SDS sample buffer, separated by SDS-PAGE. Ten per cent of the input TNT lysates was also run on the PAGE to determine relative binding capacity. The PAGE gel was then dried and exposed for autoradiography.

GST pull-down assay was also used to determine whether endogenous PTEN or NHERF1 binds to recombinant NHERF1 or PTEN, respectively. MDA-MB-468, MCF7, and T47D cells were harvested by adding 1× lysis buffer (50 mmol/l Tris-HCl [pH 7.5], 150 mmol/l NaCl, 1 mmol/l EDTA, 1% Triton-X 100, 10 mmol/l NaF, 1 mmol/l Na₃VO₄ and 1× protease inhibitor cocktail [Sigma, St. Louis, MO, USA]). The cell lysates were then incubated with beads coated with GST fusion proteins. The incubation lasted for 2 hours at 4°C. The beads were then washed thoroughly with 1× binding buffer before being boiled in 1× SDS sample buffer. The eluted proteins were then subjected to NHERF1 or PTEN immunoblotting.

To assess the interaction of PTEN with NHERF1 at the endogenous level, cultured MCF7 or Zr75.1 cells were lysed with 1× NETN buffer (20 mmol/l Tris-HCl [pH 7.5], 150 mmol/l NaCl, 1 mmol/l EDTA, 0.5% NP-40, 30 μg/ml aprotinin, 5 mmol/l PMSF, 25 mmol/l NaF, and 2 mmol/l Na₃VO₄). The soluble proteins (about 600 μg) were immunoprecipitated with 2 μg of goat IgG reactive to PTEN (N-19; Santa Cruz Biotechnologies, Santa Cruz, CA, USA) at 4°C overnight, using normal goat IgG as a control. The immunocomplex was collected by addition of 50 μl of agarose-conjugated protein G (Roche Applied Science) and detected with anti-NHERF1 antibody.

Protein concentrations were measured by using BCA reagent. Lysates were then subjected to immunodetection of phospho-Akt (p-Akt) and total Akt. Immunoblottings were carried out essentially as described previously 32. Antibodies used were human NHERF1 (EXBIO Praha, Vestec, Czech Republic), β-actin (Santa Cruz Biotechnology), mouse NHERF1 (Affinity Bioreagents, Golden, CO, USA), PTEN (Millipore), caspase-3, phospho-Akt-Ser473, total Akt, and phospho-p70 S6 kinase-Thr421/Ser424 (Cell Signaling Technology, Danvers, MA, USA).

MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) assays were used to measure cell viability. Cells were seeded in 96-well cluster dishes at 5,000 cells/well with 100 μl complete medium. After overnight incubation, medium was replaced to include various concentrations of STI-571 (Novartis Pharmaceuticals). After 2 days of treatment, cells were fed 100 μl fresh medium that contained 1 mg/ml MTT (Sigma). The incubation lasted for 2 hours before the medium was removed and cells dissolved in 150 μl dimethyl sulfoxide. Absorbance was measured using a multiSkan plate reader (LabSystems, Waltham, MA, USA) at a wavelength of 570 nm. Each sample was processed in triplicate. Experiments were repeated at least three times.

To verify that there is an interaction between NHERF1 and PTEN, we first used GST-PTEN (Figure 1a) to pull down NHERF1 from lysates of MCF7 and MDA-MB-468 cells. We found GST-PTEN, but not GST alone, to be associated with NHERF1 (Figure 1b). Of note, the quantity of full-length GST-PTEN bound to the beads was much lower than that of the GST control (Figure 1a). As a result, much less GST-PTEN was used in the experiments. To determine whether the interaction between NHERF1 and PTEN was direct, we conducted a GST pull-down assay using in vitro synthesized full-length NHERF1. GST-PTEN but not GST interacted directly with the full-length NHERF1 and the PDZ-I&II domains (Figure 1c). Thus, the PDZ-I&II domains of NHERF1 are likely to mediate this direct interaction. In reverse experiments, we used GST fusion of PDZ-I and -II domains (residues 1 to 239) or carboxyl-terminal half (CT; residues 231 to 358) of NHERF1 to pull down PTEN (Figure 1d). As shown in Figure 1e, GST-PDZ-I&II bound to PTEN from MCF7 and T47D cell lysates. In contrast, CT did not bind to PTEN, confirming that NHERF1 interacted with PTEN through its PDZ domains. To further assess whether the PDZ-I or the PDZ-II domain was responsible for PTEN association, we conducted a similar GST pull-down assay using GST-PDZ-I (residues 1 to 150) or GST-PDZ-II (residues 97 to 239) beads (Figure 1f). Although PDZ-II did not interact with PTEN, PDZ-I associated with PTEN at a level similar to that of PDZ-I&II, indicating that PDZ-I is responsible for the interaction of NHERF1 with PTEN (Figure 1g).

The direct interaction between PTEN and NHERF1 was also confirmed by pull-down assays using GST-PDZ and [³⁵S]-labeled PTEN products. Consistent with earlier findings, the wild-type PTEN bound strongly to PDZ-I but very weakly to PDZ-II (Figure 1h). The interaction was not affected by a tumor-derived mutation (C124S). To determine whether the PDZ-binding motif in PTEN is required for its interaction with NHERF1, we made PTEN with a deletion of the last six amino acids (ΔC). Pull-down assays, showing this PTEN mutant to be incapable of interacting with NHERF1 (Figure 1h), indicated that the PDZ-binding motif of PTEN is responsible for its binding to NHERF1. We further examined whether NHERF1 bound to PTEN at the endogenous expression level by using MCF7 and Zr75.1 cells that expressed high levels of NHERF1 and PTEN. Co-immunoprecipitation identified a significant interaction between the two proteins, demonstrating that NHERF1 associates with PTEN in vivo when they are expressed at the endogenous level (Figure 1i).

Expression of NHERF1 reportedly accelerates the decay of p-Akt induced by PDGF stimulation. We first verified this conclusion in NHERF1^+/+ and NHERF1^-/- MEFs isolated from 14-day embryos that resulted from crossing of NHERF1^+/- mice. Their NHERF1 genetic status was determined by genotyping, and their NHERF1 protein expression was verified by immunoblotting (Figure 2a). NHERF1^-/- and NHERF1^+/+ MEFs were then subjected to treatment with PDGF ligands and harvested after 0 to120 minutes. PDGF initially induced p-Akt to similar levels in the two groups of MEFs (10 minutes). p-Akt in NHERF1^-/- cells remained at high levels at 90 and 120 minutes, but the p-Akt level in NHERF1^+/+ cells deminished markedly (Figure 2b). We repeated these experiments in two cell lines of breast origin. MCF10A cells over-expressing NHERF1 cDNA by retroviral infection exhibited a faster turnover of p-Akt than did control cells (Figure 3a). Likewise, knockdown of NHERF1 expression in Zr75.1 cells by retroviral infection with NHERF-910 siRNA decelerated the decay of p-Akt signals induced by PDGF (Figure 3b). In agreement with these findings, the phosphorylation status of an Akt downsream target, namely p70 S6 kinase, was consistent with p-Akt level in both MCF10A and Zr75.1 cells. These results agree with the working model that NHERF1 attenuates PDGF-initiated downstream survival signals by forming protein complex with PTEN and PDGFR.

Knockout or knockdown of NHERF1 resulted in slower dephosphorylation of Akt in vitro (Figures 2 and 3). If this occurs in vivo as well, then an increase in p-Akt level would be expected in the NHERF1 knockout mice. To clarify this, we measured the p-Akt level in mammary gland specimens taken from the three groups of mice (NHERF1^+/+, NHERF1^+/-, and NHERF1^-/-), whose NHERF1 genetic background was verified (Figure 4a). NHERF1 protein was also measured. Despite expressional variability among mice in the same group, NHERF1^+/- mice had an overall lower NHERF1 protein level in mammary glands than did wild-type mice. No NHERF1 protein was detectable in the NHERF1^-/- mice (Figure 4b). This finding was consistent with the NHERF1 expression pattern in kidney extracts among the three genotypes 31. When p-Akt was measured, the NHERF1^-/- mammary gland was shown to maintain a markedly higher p-Akt level in comparison with that of the NHERF1^+/+ specimens, whereas the total Akt level remained little changed (Figure 4b). This result suggests that normal NHERF1 function involves suppression of Akt survival signaling by affecting the balance between PI3K and PTEN. Interestingly, compared with the wild-type mice, the NHERF1^+/- mice exhibited a moderate but consistent increase in p-Akt level (Figure 4b), indicating that decreased NHERF1 expression resulting from deletion of a single NHERF1 allele may be sufficient to promote cell survival in the mammary gland.

Because NHERF1 markedly affected p-Akt turnover in response to PDGF stimulation, we surmised that NHERF1 expression status might also affect cell response to PDGFR inhibition. Zr75.1 cells with or without NHERF1 knockdown were treated with STI-571 for 2 days before MTT assay for surviving cells. STI-571 is a small molecule inhibitor of tyrosine kinases, including BCR-Abl, PDGFR and c-kit, and as a result it has been used effectively in patients with chronic myelogenous leukemia, gastrointestinal stromal tumor, and other types of tumors 333435. We first compared the Zr75.1 cells (base control and NHERF1 knockdown cells) in tems of their sensitivity to STI-571. While untreated, NHERF1 knockdown cells exhibited a modest increase in cell proliferation after 2 days of culture, which is in agreement with our earlier report 25. When cells were treated with STI-571, we found that Zr75.1 cells with NHERF1 knockdown were much more resistant to STI-571 than the control (Figure 5a). Similarly, over-expression of NHERF1 in MCF10A cells rendered them more sensitive to STI-571 treatment than their corresponding Babe control (Figure 5b). These findings suggest that changes in cell survival signaling resulting from NHERF1 loss affect cell sensitivity to PDGFR blocker.

To determine whether apoptosis is involved in the cell density difference between NHERF1-positive and NHERF1-negative cells, we harvested the cells during STI-571 treatment and assayed cell lysates for cleaved forms of caspase-3, a marker of active apoptotic process. Immunoblottings showed that before STI-571 treatment, no cleaved caspase-3 isoforms were present in either the NHERF1 knock-down or the control Zr75.1 (Babe) cells. One day of STI-571 treatment resulted in significant increases in caspase-3 cleavage in Zr75.1 Babe control cells, suggesting that apoptosis is at least partially responsible for STI-571-induced growth inhibition. Interestingly, the same treatment led to markedly lower caspase-3 cleavage in NHERF1-knockdown Zr75.1 cells (Figure 6a), which is consistent with a decreased response to STI-571-induced cell death in comparison with that of NHERF1-positive Zr75.1 cells (Figure 5a). Similar experiments were carried out on MCF10A cells, in which we showed that over-expression of NHERF1 markedly enhanced STI-571-induced cleavage of caspase-3 (Figure 6b). All of these findings indicated that loss of NHERF1 may affect cell response to growth factor inhibition as a result of increases in intrinsic cell survival 36.

To assess whether PTEN is required for the increased sensitivity to STI-571, we first determined whether altered NHERF1 expression level affected cell responses to STI-571 in a PTEN-null background. MDA-MB-468 cells are NHERF1 positive but lack PTEN as a result of homozygous deletion. We knocked down NHERF1 expression in MDA-MB-468 cells using siRNA retrovirus, and these cells were tested for STI-571 sensitivity. We found that in the absence of PTEN the NHERF1-knockdown cells responded to STI-571 at a level similar to that in Babe control cells (Figure 7a). We then determined whether PTEN knockdown was able to influence NHERF1-inducible sensitivity to STI-571 in MCF10A cells. NHERF1 expression (or Neo control) in MCF10A cells was retrovirally transduced and the resultant cells were subjected to PTEN knockdown through retrovirally delivered siRNA (or Babe control). These steps yielded four cell lines with varying PTEN and NHERF1 expression. These cells were then tested for STI-571 sensitivity. PTEN knockdown resulted in a modest increase in proliferation as compared with the control. As expected, over-expression of NHERF1 in parental cells elevated STI-571-inducible killing. In contrast, NHERF1 expression in cells with PTEN knockdown remained highly resistant to STI-571. These findings indicated that increased sensitivity to PDGFR inhibition by NHERF1 was dependent on PTEN (Figure 7b).

To explore whether LOH is responsible for a decrease in NHERF1 expression, we correlated the two parameters in 22 breast cancer cell lines that had been tested 1525. Among the six cell lines with a high NHERF1 expression level, five (83%) were found to be LOH negative. In contrast, of the 16 cell lines with low and no NHERF1 expression, only two (12.5%) were LOH negative. Thus, LOH at the NHERF1 locus is highly associated with lowered NHERF1 expression (P = 6.8 × 10^-6, by Fisher's exact test).

The physical and functional relation between NHERF1 and PTEN indicates they may exist in a common tumor suppressor pathway, which would predict segregation between NHERF1 and PTEN (or PI3KCA) alterations. We thus evaluated whether NHERF1 LOH is correlated with PTEN/PI3KCA mutational status. A total of 39 breast cancer cell lines with available genetic status were used for the correlative analyses 37. Among the 23 NHERF1 LOH-positive cell lines, 15 (65%) had an unaltered PTEN or PI3KCA gene. In contrast, out of the 16 LOH-negative cell lines, only 4 (25%) showed wild-type PTEN/PI3KCA status. The genetic alteration in either PTEN or PI3KCA gene was strongly associated with intact NHERF1 alleles (P = 0.022, by Fisher's exact test).