Human breast tumour cell lines BT20, BT474, BT483, BT549, CAMA1, DU4475, HCC1428, HCC1954, MB157, MCF7, MDA-MB-134, MDA-MB-231, MDA-MB-330, MDA-MB-361, MDA-MB-415, MDA-MB-435S, MDA-MB-453, MDA-MB-468, SKBr3, T47D, and ZR75-1 were purchased from American Type Culture Collection (Manassas, VA, USA). The SUM149-PT line was a gift from Dr. Stephan Ethier (University of Michigan, Ann Arbor, MI, USA). All cell lines were cultured in recommended media supplemented with 10% foetal bovine serum (FBS) (Invitrogen, Carlsbad, CA, USA).

A vector-based short hairpin RNA (shRNA) method was used to generate MCF7 and T47D cells with inhibited NHERF1 expression. A two-step ligation method 26 was used to insert the interfering sequences into pBS/U6 (a gift from Dr. Yang Shi, Harvard Medical School, Boston, MA, USA). Two NHERF1 mRNA sequences corresponding to cDNA positions 786 and 910 were targeted. Oligonucleotide sequences for NHERF-786 were 5'-GGAGATACAGAAGGAGAACAGA-3' (oligo 1, forward), 5'-AGCTTCTGTTCTCCTTCTGTATCTCC-3' (oligo 1, reverse), 5'-AGCTTCTGTTCTCCTTCTGTATCTCCCTTTTTG-3' (oligo 2, forward), and 5'-AATTCAAAAAGGGAGATACAGAAGGAGAACAGA-3' (oligo 2, reverse). Oligonucleotide sequences for NHERF-910 were 5'-GGAAACTGACGAGTTCTTCAA-3' (oligo 1, forward), 5'-AGCTTTGAAGAACTCGTCAGTTTCC-3' (oligo 1, reverse), 5'-AGCTTTGAAGAACTCGTCAGTTTCCCTTTTTG-3' (oligo 2, forward), and 5'-AATTCAAAAAGGGAAACTGACGAGTTCTTCAA-3' (oligo 2, reverse). Interference sequences were verified by automated DNA sequencing. The hairpin loop sequences were then released by digesting with BamHI and EcoRI and subcloned into a retroviral vector pBabe-U6 (a gift from Dr. Jinsong Liu, M. D. Anderson Cancer Center, Houston, TX, USA) 27, yielding pBabe-U6/NHERF-786 and pBabe-U6/NHERF-910. Retroviruses were produced by transfecting packaging cells (amphotropic Phoenix) with pBabe-U6/NHERF-786, pBabe-U6/NHERF-910, or parental pBabe-U6, using Fugene 6 (Roche Applied Science, Indianapolis, IN, USA). The medium was collected 2 days after transfection. After centrifugation, the supernatant was then passed through a 0.45-μm filter. The retrovirus stock was stored at -80°C until use. Cultured MCF7 and T47D cells were infected with a virus cocktail (1 ml of retroviral stock, 2 ml of medium, and 4 μg of polybrene). The next day, the virus was removed and replaced with fresh medium that contained 0.5 μg/ml puromycin. Surviving cells were assessed for NHERF1 expression by immunoblotting.

Thymidine incorporation assay was used to measure the DNA synthesis rate as described previously 26. MCF7 and T47D cells cultured in 24-well plates were pulsed with 1 mCi [³H]-thymidine (3,000 Ci/mmol; PerkinElmer Life and Analytical Sciences, Inc., Shelton, CT, USA). After 5-hour labeling, non-incorporated tritium was removed by trichloroacetic acid washes. Acid-insoluble tritium was assessed by scintillation counting (microBeta Trilux 1450; Wallac, now PerkinElmer Life and Analytical Sciences, Inc.). The relative cell proliferation rate was obtained by dividing the counts from cells in which NHERF1 was downregulated by the ones from control cells. Experiments were repeated three times.

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays were conducted to measure the relative number of viable cells. Cells were seeded in 96-well cluster dishes at 2,500 cells per well with 100 μl of complete medium. At indicated time points, medium was replaced with 100 μl of fresh medium supplemented with 20 μl of 5 mg/ml MTT (Sigma-Aldrich, St. Louis, MO, USA). The incubation lasted for 2 hours before the medium was removed and cells dissolved in 100 μl of lysis buffer. Absorbance was measured using a multiSkan plate reader (Thermo Scientific, Waltham, MA, USA) at a wavelength of 570 nm. Experiments were repeated at least three times.

Cells (1 × 10⁴) were suspended in 1 ml of 1× culture medium that contained 0.35% agarose. The suspension was added on top of 4 ml of solidified 0.7% agarose. After the cells were set in agarose, 1 ml of fresh medium was added to cover the agarose. Assays were performed in triplicate. Plated cells were incubated for 20 days at 37°C before formed colonies larger than 50 μm in diameter were counted. Experiments were repeated three times.

Cultured cells (approximately 2 × 10⁶) were trypsinised and washed twice with 1× phosphate-buffered saline (PBS). Cells were then fixed by being added drop-wise to 5 ml of ice-cold 80% ethanol while vortexing. After fixing for at least 1 hour at room temperature, the cells were stored at -20°C. Before being stained, the cells were washed with 1× PBS and incubated at 37°C for 30 minutes with propidium iodide (50 μg/ml; Sigma-Aldrich) in the presence of 10 μl of RNase A (10 mg/ml; Sigma-Aldrich). Cell cycle analysis was performed with an FACS station equipped with CellQuest (Becton Dickinson, Franklin Lakes, NJ, USA). At each cell cycle phase, the population was determined by computer model fitting (Verity Software House, Topsham, ME, USA).

Serum starvation was used to synchronise MCF7 at the G₀/G₁ phase. MCF7 cells were seeded at 8 × 10⁵ per 60-mm dish. After being cultured in complete medium overnight, cells were incubated with serum-free medium for 1 day. The cells were then re-fed with medium supplemented with 10% FBS for various time periods before being harvested for fluorescence-activated cell sorting (FACS) analyses.

Four- to five-week-old female athymic nude mice (Harlan, Indianapolis, IN, USA) were used for experimental tumourigenicity assays. To facilitate the establishment of xenografts of oestrogen-dependent cells, each mouse was inoculated subcutaneously with an oestrogen pellet (0.7 mg 17β-estradiol per pellet; 60-day slow-release; Innovative Research of America, Sarasota, FL, USA). Two days after pellet implantation, equivalent amounts of T47D cells (1.5 × 10⁶; Babe control or NHERF-910) resuspended in 100 μl of mixture (1:1 with un-supplemented media) of Matrigel (BD Biosciences, San Jose, CA, USA) were injected into each side of second-pair breast mammary fat pads (3 × 10⁶ cells in total). Six weeks after injection, mice were euthanised by carbon dioxide, and the established tumours on both sides of mammary glands were dissected, pooled, and weighed. All procedures were performed according to the recommendations of the Institutional Animal Care and Use Committee.

Immunoblottings were carried out essentially as described previously 28. Antibodies used were NHERF1 (EXBIO Praha, Bestec, Czech Republic), Rb and p27 (BD Biosciences), cdk2 (Calbiochem, San Diego, CA, USA), cdk4 and cyclin D1 (Cell Signaling Technology, Inc., Danvers, MA, USA), cyclin E and β-actin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), and α-tubulin (Sigma-Aldrich).

We used immunoblotting to analyse the expression of NHERF1 in 22 breast cancer cell lines (Figure 1a), among which the origin of MDA-MB-435S was under debate 293031. MDA-MB-231 and SUM149-PT, two lines harbouring both NHERF1 mutation and LOH 1, had low NHERF1 protein levels. High expression of NHERF1 was found in some oestrogen receptor (ER)-α-positive cells, such as T47D, Zr75.1, and MCF7, consistent with an earlier report that correlated NHERF1 expression with ER-α positivity 32. However, NHERF1 expression does not necessarily follow the ER-α status. For example, BT20 cells expressing constitutively active ER-α had low NHERF1 expression, whereas NHERF1 was maintained at a high level in ER-α-negative MDA-MB-453, MDA-MB-468, and SKBr3 cells, suggesting that mechanisms other than oestrogen induction must exert an effect on NHERF1 expression.

One way to determine whether NHERF1 functions as a tumour suppressor gene in human breast cancer is to assess the resultant phenotypic responses by knocking down endogenous NHERF1 expression in NHERF1-high expressors. We used a retrovirus-based system to deliver shRNA to knock down NHERF1 expression in T47D cells. Introduction of NHERF1 shRNA-786 markedly lowered the NHERF1 level (Figure 1b, left panel). By contrast, NHERF1 expression was unaffected by empty retrovirus (Babe). ShRNA can sometimes create non-specific outcomes that are not related to the targeted gene of interest 33. To ascertain the specificity of phenotypic changes resulting from NHERF1 loss, we also prepared an NHERF1 knockdown line from MCF7 cells by introducing a different targeting sequence, NHERF1 shRNA-910. Introducing shRNA-910 led to virtual elimination of NHERF1 expression in MCF7 cells, in comparison with that of parental cells and Babe control (Figure 1b, right panel).

One frequent response of a given tumour suppressor gene is to inhibit cell proliferation. To determine whether NHERF1 affects cell proliferation, we sought to compare the growth curve of T47D/NHERF-786 and MCF7/NHERF-910 cells with that of their corresponding Babe controls, as determined by MTT assay in 96-well plates. As shown in Figure 2a and 2b, although infection of Babe retrovirus did not affect cell proliferation (parental T47D versus T47D/Babe cells), T47D/NHERF-786 cells grew consistently faster than T47D/Babe control cells, most obviously at days 6 and 7 after plating (by approximately 35%). Similarly, the growth rate of MCF7/NHERF-910 was found to be higher (by 30% to 40%) than that of MCF7/Babe or parental MCF7 cells (Figure 2c). These results suggested an inhibitory effect of NHERF1 on cell proliferation.

The DNA synthesis of these cells was also compared using [³H]-thymidine incorporation assays. Knockdown of NHERF1 in T47D cells resulted in a significantly higher DNA synthesis rate than that of the Babe control cells (P = 0.022; Figure 3a). MCF7 cells with NHERF1 knockdown consistently showed accelerated proliferation (P = 4 × 10^-6; Figure 3b). The regulatory effect of NHERF1 on cell proliferation was also reflected in anchorage-independent growth. T47D/NHERF-786 cells showed a more efficient colony outgrowth in soft agar than did either parental T47D (P = 0.0043) or T47D/Babe (P = 0.020) cells (Figure 3c), indicating that NHERF1 exhibits growth suppression activity in breast cancer cells.

The growth-promotion response as result of NHERF1 knockdown was also examined in a mouse xenograft model. We compared the in vivo growth of T47D cells infected with NHERF1 shRNA-786 retrovirus or vector control. To facilitate tumour formation from T47D cells, which are oestrogen-dependent, we pre-implanted each mouse subcutaneously with a slow-release oestrogen pellet. Two days after pellet implantation, T47D/NHERF-786 or T47D/Babe cells were mixed with Matrigel and injected into the mouse mammary pads. Tumours were visibly established in mice 42 days after injection, when all mice were sacrificed to compare tumour size. As shown in Figure 4a and 4b, tumours formed from T47D/NHERF-786 (72.4 ± 13.9 mg, n = 10) were significantly larger than those from T47D/Babe cells (37.7 ± 8.8 mg, n = 8) (P = 0.043), suggesting that lowered NHERF1 expression promotes tumour growth in vivo. None of the tumours was found to metastasise to lung or liver.

To examine whether the growth-inhibitory activity of NHERF1 can be attributed to its effect on cell cycle progression, we first compared the cell cycle distribution of asynchronised NHERF1-negative and -positive T47D and MCF7 cells. As shown in Figure 5a, T47D/NHERF-786 cells had a lower G₁/G₀ and a higher S-phase population (G₁/G₀ 50.4% ± 0.69%, S 33.3% ± 0.95%) than Babe control cells (G₁/G₀ 58.6% ± 1.49%, S 26.0% ± 1.03%). No significant difference was observed in G₂/M-phase populations between the two groups (16.3% ± 0.64% versus 15.4% ± 0.47%). Similar results were shown for MCF7 cells with NHERF1 knockdown (G₁/G₀ 37.2% ± 1.69%, S 42.1% ± 0.98%, and G₂/M 20.4% ± 0.60% for NHERF-910 versus G₁/G₀ 43.9% ± 1.75%, S 33.8% ± 1.39%, and G₂/M 22.3% ± 0.39% for Babe; Figure 5b), which suggested that NHERF1 affects cell cycle progression, probably by stalling cells at G₁ phase.

To analyse this possibility in more detail, we serum-starved MCF7 cells to arrest the cell cycle at the G₁/G₀ phase. One day of starvation led to accumulation of G₁ phase from approximately 40% with normal serum condition to more than 70% with serum starvation. The cells were then re-fed with medium containing 10% serum and were allowed to grow for 0 to 29 hours before being harvested for FACS analyses (Figure 6a). At 14 hours after serum feeding, G₁/G₀ phase cells began advancing to the S phase. Between 14 and 19 hours, a remarkable increase occurred in the S-phase population. Comparing MCF7/Babe and MCF7/NHERF-910 cells, we found that NHERF-910 cells proceeded to S phase significantly faster than their corresponding control. At the 19-hour point, 43.2% and 50.0% of the Babe and NHERF-910 cells, respectively, were in S phase. This trend lasted through 24 hours after serum feeding. At 24 hours, S phase comprised 53.9% and 60.3% of the Babe and NHERF-910 cells, respectively (Figure 6a). In contrast, more cells remained at G₁ phase in the Babe (47.8% and 30.8%) than in the NHERF-910 group (42.4% and 23.2%) at the 19- and 24-hour time points, respectively. These results indicated that NHERF1 prohibited G₁-to-S cell cycle transition, reflecting the growth-inhibitory activity of NHERF1.

Phosphorylation of Rb protein is one of the most critical processes in enforcing the G₁-to-S transition 34. To further verify whether NHERF1 plays an inhibitory role in G₁-to-S cell cycle progression, we compared the Rb protein expression in MCF7/Babe and MCF7/NHERF-910 cells. As shown in Figure 6b, the Rb protein in growth-arrested cells (at 0 hours) was mainly the hypo-phosphorylated form. Corresponding to the beginning of S-phase entry at 14 hours after serum feeding, we detected a high level of hyper-phosphorylated Rb, which gradually receded during the observation period. In consonance with an accelerated G₁-to-S transition as a result of NHERF1 loss, MCF7/NHERF-910 cells contained a significantly higher level of hyper-phosphorylated Rb than did the Babe control, most prominently at 14 hours and 19 hours after serum feeding. To explore the mechanism responsible for the difference in Rb phosphorylation and G₁-to-S transition, we compared the two groups of cells in their expression of some of the most prominent cell cycle regulatory proteins (Figure 6b). No difference was shown in the level of p27, cdk2, cdk4, or cyclin D1 between the MCF7/NHERF-910 and MCF7/Babe cells. However, we found that the cyclin E level in MCF7/NHERF-910 cells was significantly higher than that in MCF7/Babe at 14 hours and 19 hours after serum stimulation (Figure 6b). A similar result was obtained from another independent experiment. Taken together, these results indicated that NHERF1 knockdown leads to accelerated G₁-to-S transition that may involve increased cyclin E content and elevated Rb phosphorylation status.