Nonredundant Functions for Tumor Protein D52-Like Proteins Support Specific Targeting of TPD52

Purpose:Tumor protein D52 (TPD52 or D52) is frequently overexpressed in breast and other cancers and present at increased gene copy number. It is, however, unclear whether D52 amplification and overexpression target specific functional properties of the encoded protein.

Experimental Design: The expression of D52-like genes and MAL2 was compared in breast tissues using quantitative reverse transcription-PCR. The functions of human D52 and D53 genes were then compared by stable expression in BALB/c 3T3 fibroblasts and transient gene knockdown in breast carcinoma cell lines. In situ D52 and MAL2 protein expression was analyzed in breast tissue samples using tissue microarray sections.

Results: The D52 (8q21.13), D54 (20q13.33), and MAL2 (8q24.12) genes were significantly overexpressed in breast cancer tissue (n = 95) relative to normal breast (n = 7; P ≤ 0.005) unlike the D53 gene (6q22.31; P = 0.884). Subsequently, D52-expressing but not D53-expressing 3T3 cell lines showed increased proliferation and anchorage-independent growth capacity, and reduced D52 but not D53 expression in SK-BR-3 cells significantly increased apoptosis. High D52 but not MAL2 expression was significantly associated with reduced overall survival in breast carcinoma patients (log-rank test, P < 0.001; n = 357) and was an independent predictor of survival (hazard ratio, 2.274; 95% confidence interval, 1.228-4.210; P = 0.009; n = 328).

Conclusion: D52 overexpression in cancer reflects specific targeting and may contribute to a more proliferative, aggressive tumor phenotype in breast cancer.

Chromosome 8q gain is one of the most frequent cytogenetic aberrations in human cancer (1). Increasingly refined amplification mapping studies have indicated numerous target genes along this chromosomal arm, including genes at chromosome 8q21 (2–5). A critical chromosome 8q21.13 gene is indicated to be tumor protein D52 (TPD52 or D52; refs. 6–9), a member of the D52-like family of adaptor proteins (10, 11). Human D52 transcripts or protein are overexpressed in breast, prostate, and ovarian cancer, with this being associated with increased gene copy number in a proportion of cases (6–9). Expression microarray studies also indicate that the D52 gene is amplified and/or overexpressed in multiple myeloma (12, 13), pancreatic cancer (14), and seminoma (15).

The frequency with which D52 is overexpressed in cancer argues strongly for this playing a causal role. However, as extensive chromosome 8q regions are frequently gained (1), and large numbers of 8q genes may be consequentially overexpressed (16), the significance of many overexpressed 8q genes remains unclear. In the case of D52, this is compounded by the fact that D52-like proteins have shared functions, including mutual interactions (17, 18), and binding a common partner MAL2 (19–21), which is itself overexpressed in breast and other cancers (22–24). Most studies reporting phenotypes associated with increased or decreased D52 expression have also not compared D52 functions with those of related proteins (25–29). It is therefore not known whether D52-like proteins beyond D52 itself have similar roles in enhancing cancer-associated phenotypes such as proliferation and anchorage-independent growth (26, 28). There is therefore still uncertainty whether D52 overexpression in cancer passively reflects its gene location or actively reflects nonredundant functional properties that are specifically targeted (10).

To analyze this question, the present study has taken several approaches. Firstly, we have directly compared the expression of three widely expressed D52-like genes and their common partner MAL2 in normal breast and breast carcinoma. This identified that D52, D54, and MAL2 were significantly up-regulated in breast carcinoma and that these genes localize to regions of the genome gained in breast and other cancers. As D52 and D53 represent examples of D52-like genes that are or are not overexpressed in breast cancer, we directly compared the effects of expressing these genes in BALB/c 3T3 fibroblasts, which are highly responsive to D52 expression (28). We also examined the effects of transiently knocking down human D52 or D53 expression in MCF-7 and SK-BR-3 breast carcinoma cell lines, which endogenously express both proteins (18). These approaches identified both shared and nonredundant functions for these proteins, with nonredundant functions indicating cancer-specific roles for D52. As D52 and MAL2 genes were most significantly overexpressed in breast carcinoma relative to normal breast, we then analyzed D52 and MAL2 expression in breast tissue samples using tissue microarray sections. This indicated that high D52 but not MAL2 expression was significantly associated with reduced overall survival in breast cancer patients. The present study has therefore obtained evidence that D52-like proteins have nonredundant cellular functions and that the proliferative functions of D52 may contribute to a more aggressive breast cancer phenotype.

Breast tissue samples. Breast tissue samples analyzed by quantitative reverse transcription-PCR represented 7 normal breast samples, 14 benign breast lesions, and 95 invasive carcinomas. Normal breast samples represented adjacent normal tissue from 4 breast cancer patients and normal tissue from 3 women undergoing cosmetic breast surgery. The tumor group consisted of 11 grade 1 tumors, 12 grade 3 tumors, 12 estrogen receptor (ER)–negative tumors, 12 ER-positive tumors, and 48 additional ER-positive tumors from tamoxifen-treated patients, of which 24 relapsed and 24 remained disease-free (median follow-up, 86 months; range, 18-120 months). Tissue samples were obtained and stored before RNA extraction as described previously (30), and total RNA was extracted from whole frozen specimens. Breast tissue microarray sections were purchased from the Western Australian Research Tissue Network from samples obtained from the Royal Perth Hospital (RPH) Pathology Department under ethics approvals granted by the RPH and Sir Charles Gairdner Hospital Human Research Ethics Committees. Samples were surgically removed at RPH between 1995 and 2001, fixed in formaldehyde, and embedded in paraffin, and sections included samples of normal breast, in situ and invasive carcinomas. Cores containing breast and prostate cancer cell lines were included in all array sections as controls. A total of 357 invasive carcinomas were analyzed for in situ D52 expression, and a subset of 320 cases was analyzed for in situ MAL2 expression. Patients were ages 27 to 91 years at diagnosis (median, 59 years) and had no prior treatment before surgery. Breast carcinomas were analyzed for the expression of established markers including ER by the RPH Pathology Department, with tumors with 10% ER-positive cells being considered positive [ER-positive cases (n = 265), ER-negative cases (n = 85), and unknown status (n = 7)]. Other clinical and histologic data available included lymph node status at diagnosis [node-negative (n = 244) and node-positive (n = 113)], tumor grade [grade 1 (n = 88), grade 2 (n = 159), grade 3 (n = 104), and unknown (n = 6)], and patient survival (median follow-up, 55 months; range, 0.3-123 months). Additional clinical information was provided by the Multidisciplinary Breast Service at RPH.

Total RNA extraction and cDNA synthesis. Total RNA was extracted using the acid-phenol guanidinium method, and RNA quality was assessed using agarose gel electrophoresis and ethidium bromide staining. Reverse transcription reactions contained 1× reverse transcription buffer [500 μmol/L each deoxynucleotide triphosphate, 3 mmol/L MgCl₂, 75 mmol/L KCl, 50 mmol/L Tris-HCl (pH 8.3)], 20 units RNasin RNase inhibitor (Promega), 10 mmol/L DDT, 100 units Superscript II RNase H- reverse transcriptase (Invitrogen), 3 μmol/L random hexamers (Pharmacia), and 1 μg total RNA in 20 μL. Samples were incubated at 20°C for 10 min and 42°C for 30 min, and reverse transcriptase was inactivated by heating at 99°C for 5 min and cooling at 5°C for 5 min.

Real-time reverse transcription-PCR. Real-time reverse transcription-PCR analyses were done as described previously (30). Briefly, reactions were done using an ABI Prism 7700 Sequence Detection System using the SYBR Green PCR Core Reagents kit (Perkin-Elmer Applied Biosystems). The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 min and 50 cycles at 95°C for 15 s and 65°C for 1 min. Quantitative values were obtained from Ct values using the Perkin-Elmer Biosystems analysis software according to the manufacturer's instructions. Results were expressed as fold differences in target gene expression relative to the endogenous RNA control TBP. Gene expression values were then expressed relative to the median value obtained in the 7 normal breast samples, which was set at 1.0 for each gene.

Cell culture and stable transfection of BALB/c 3T3 fibroblasts. Mouse BALB/c 3T3 fibroblasts were grown at 37°C, 5% CO₂ in DMEM supplemented with 10% fetal bovine serum (Invitrogen), 2% l-glutamine (Invitrogen), and 7.5% sodium bicarbonate (Sigma). For the stable expression of human D52, a SalI-BamHI cDNA fragment including the full-length coding sequence was subcloned into the same sites of the PG307 vector (31). The PG307hD52 plasmid, the PG307hD53 plasmid (18), and the PG307 vector were stably transfected into 3T3 fibroblasts by seeding cells at ∼60% confluence in six-well plates and transfecting 24 h later with 5 μg DNA in 250 μL Opti-MEM (Invitrogen) and 6 μL LipofectAMINE 2000 (Invitrogen). After 24 h, cells were passaged into 100 mm dishes and 48 h post-transfection, G418 (Invitrogen) was added (1 mg/mL), and drug selection continued for 2 weeks. Selected G418-resistant clones were screened using Western blot analyses, and 6 cell lines expressing human D52, 5 cell lines expressing human D53, and 3 vector control cell lines were maintained in 1 mg/mL G418 medium for further analyses.

Transient small interfering RNA transfections. Human D52 and D53 small interfering RNA (siRNA) duplexes were synthesized by Dharmacon. The targeted sequences (sense strand; 5′-3′) were si2-1 (GCGGAAACTTGGAATCAAT), si2-2 (GGAGAAGTCTTGAATTCGG) and si2-3 (AAGAAAAGGTCGAAAACTT) for D52 and si3-2 (TCACAAGCCTCAAGACGAA), si3-3 (GCTAGAAGACGAAATTACA), and si3-4 (GCAAGAAGTTCGGAGACAT) for D53. Positive control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) siRNA (TGGTTTACATGTTCCAATA) and nontargeting siControl siRNA (TAGCGACTAAACACATCAA) were also purchased from Dharmacon. The human breast cancer cell lines MCF-7 and SK-BR-3 were cultured in RPMI supplemented with 10% fetal bovine serum (Invitrogen), 3% l-glutamine (Invitrogen), and 10 μg/mL insulin (Sigma). Cells were cultured to 70% confluence, trypsinized, and plated onto glass coverslips, or 2 × 10⁴ MCF-7 and 4 × 10⁴ SK-BR-3 cells were seeded into wells of 24-well plates. After 24 h, cells were transfected with 100 nmol/L siRNA duplexes using TransIT-TKO transfection reagent (Mirus) in complete medium following the manufacturer's instructions and analyzed 48 h later.

Western blot analyses. Cells were washed twice in PBS and lysed in SDS lysis buffer for total protein extracts as described (18). Protein extracts (10 μg/well) were resolved using SDS-PAGE on 12.5% polyacrylamide gels and electrotransferred to nitrocellulose membranes (Millipore). Membranes were blocked overnight at 4°C in 5% skim milk powder in TBS. Membranes were washed twice with TBS and incubated with affinity-purified rabbit polyclonal D52 antisera (1:100),⁸

affinity-purified rabbit polyclonal human D53 antisera (1:100; ref. 18), mouse monoclonal actin (1:2,000; a gift from J. Lessard), or mouse monoclonal GAPDH (1:5,000; Ambion) antibodies in 0.1% Tween 20 in TBS for 1 to 2 h. Membranes were washed three times in 0.1% Tween 20 in TBS and then incubated with a horseradish peroxidase–conjugated donkey anti-rabbit or anti-mouse secondary antibody (1:5,000; GE Biosciences) for 1 h. Membranes were finally washed four times and visualized by Western lightening chemiluminescent reagent (Perkin-Elmer).

Indirect immunofluorescence analyses. BALB/c 3T3 cell lines were cultured to near confluence, trypsinized, and plated onto glass coverslips overnight. Cells were washed twice with PBS, fixed in 4% paraformaldehyde/PBS for 20 min, permeabilized with 0.1% saponin for 10 min, washed twice with PBS, and incubated overnight with affinity-purified human D52 (1:100; ref. 6) or human D53 (1:50) antisera in 0.1% bovine serum album in PBS. Cells were washed twice and incubated with secondary Cy3-conjugated donkey anti-rabbit (1:500; Jackson Immunoresearch) antibody in 0.1% bovine serum album in PBS for 1 h in the dark. For phalloidin staining, cells were washed twice and incubated with FITC-conjugated phalloidin (Sigma) for 15 min in the dark. Cells were washed again and DNA was counterstained with 10 nmol/L 4′,6-diamino-2-phenylindole (Sigma). After washing in PBS, cells were mounted in DABCO (Sigma) prepared according to the manufacturer's instructions. Images were taken using an Olympus BX50 microscope equipped with a SPOT camera (Diagnostic Instruments). Between 50 and 100 cells from each transfected cell line were visually scored according to in situ D52 and D53 expression levels or actin stress fibers, and Image ProPlus (Media Cybernetics) was used to calculate cell area (μm²).

Cell proliferation assays. Parental cells or transfected 3T3 cell lines (1 × 10³), 2 × 10³ MCF-7 cells, or 4 × 10³ SK-BR-3 cells were plated in triplicate in 96-well plates. In all cases, 50 μL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reagent (Sigma) was added to each well either immediately (0 time point) or at the designated number of days post-seeding and incubated at 37°C in the presence of 5% CO₂ for 4 h. To stop each reaction, medium was replaced with 100 μL DMSO (Sigma) per well and mixed thoroughly. Absorbances at 540 nm were taken using a Multiskan Ascent plate reader. Means of triplicate wells from four independent experiments were used to generate data points and SE.

Soft-agar colony formation assays. Stably transfected or parental 3T3 cell lines (5 × 10⁴) were seeded in duplicate into six-well dishes in 3 mL complete medium containing 0.33% agar solution overlaying 0.5% agar (Becton Dickson). Colonies containing >10 cells in 20 random fields were counted at ×100 magnification 14 days after plating. Means of duplicate wells from three independent experiments were used to generate data points and SE.

Crystal violet and apoptosis assays. SK-BR-3 or MCF-7 cells were seeded in duplicate in 24-well plates and transfected with siRNA as described above. Cell density per well was determined after 48 h by crystal violet assay. Briefly, following 15 min of cell fixation with 3% acetic acid/10% methanol, cells were stained with 0.4% crystal violet for 50 min and rinsed twice in water. Cells were air-dried and triplicates were taken using a DC 500 dissecting microscope and camera (Leica Technologies). Cell areas were calculated using Metamorph (version 6.1; Molecular Devices) with manual thresholding. The means of cell areas measured in triplicate from three independent experiments were used to generate data points and SE. DNA fragmentation produced 48 h post-transfection was quantified using the Cell Death ELISAPLUS kit according to the manufacturer's instructions (Roche). Briefly, cells were lysed and centrifuged, and supernatants were transferred to 96-well precoated plates to assay cytoplasmic histone-associated DNA fragments. After incubations and washes, stop substrate was added and color development was measured at 405 nm. Samples were analyzed in duplicate in three independent experiments.

Immunohistochemical analysis of paraffin-embedded breast tissue microarrays. The D52 antisera employed for immunohistochemical analyses have been described previously (6), whereas the derivation of the rabbit polyclonal MAL2 antisera will be described elsewhere.⁹

Tissue microarray slides were washed and rehydrated in xylene and ethanol before rehydrating in water and equilibrating in PBS. Slides were blocked in 10% normal goat serum for 20 min, rinsed in PBS, and incubated for 2 h with affinity-purified D52 (1:100) or MAL2 (1:100) antibodies in 2% normal goat serum in PBS. Primary antibody was omitted in control incubations. Biotinylated goat anti-rabbit secondary (Jackson Immunoresearch) was added (1:500) to the slides and incubated for 1 h. Slides were incubated with hydrogen peroxide for 30 min and stained with tertiary ABC reagent (Pierce) for 1 h. DAB (Sigma) was added for 2 to 5 min and slides counterstained with Nuclear Fast Red (Fronine). Slides were scanned using the Virtual Microscope ScanScope Unit and ScanScope Console program (Aperio Technologies) at ×200 magnification. Tissue arrays were visualized using Image Scope and staining intensity was quantified within tissue cores of fixed and uniform diameter using the Positive Pixel Count algorithm (Aperio Technologies). The number of strong pixels (defined as pixels of 175-220 intensity) was measured per tissue core. Partial tissue cores, those with staining artifacts, or those without epithelial elements (normal or cancerous) were excluded from analyses.

Statistical analyses. The SPSS for Windows package (version 13; SPSS) was used in most analyses. Distributions of continuous variables were often skewed and summarized using medians and interquartile ranges. Mann-Whitney U tests were used to test for differences in cell size according to human D52 or D53 expression categories. Categorical variables were summarized using percentages within each group. Spearman rank correlation and Fisher's exact test were used to compare protein expression and other variables. In patient samples, strong pixel counts were compared in normal breast, in situ carcinoma, or breast carcinoma samples from the same patients using the Wilcoxon signed rank test. Survival distributions were estimated by the Kaplan-Meier method, and the significance of differences between overall survival rates was ascertained using the log-rank test. Best-fitting multiple Cox proportional hazards models with backward stepwise selection were used to identify independent predictors of survival from potential risk factors. Results of all cell proliferation, soft agar, crystal violet, and apoptosis assays are expressed as mean ± SE of three to four independent experiments. Comparisons between groups were made using two-tailed, unequal variance Student's t tests calculated using Excel (Microsoft).

D52-like and MAL2 gene transcript levels in breast tissues. Quantitative reverse transcription-PCR was employed to study the relative expression of D52-like genes and MAL2 in a cohort of 7 normal breast samples, 14 benign breast lesions, and 95 invasive carcinomas. Gene expression values were expressed relative to median values obtained in normal breast samples, which were set at 1.0 for each gene. The results of comparing median relative gene expression values in different tissue types using Mann-Whitney tests are summarized in Table 1. Of the genes examined, D52, D54, and MAL2 transcripts were all detected at significantly increased levels in breast carcinoma samples relative to normal breast (Table 1) and also in those 24 ER-positive tumors from patients who subsequently relapsed after tamoxifen therapy compared with 24 tumors from patients who remained disease-free (Mann-Whitney U test, P < 0.001 for D52, P = 0.001 for D54, and P = 0.001 for MAL2; n = 48). In addition, D54 transcripts were detected at significantly increased levels in benign lesions relative to normal breast, and MAL2 transcripts were detected at significantly increased levels in breast carcinoma samples relative to benign lesions (Table 1). In contrast, D53 was not differentially expressed between the sample categories compared (Table 1).

D52 and D53 expression is associated with reduced cell area and actin stress fibers in 3T3 cells. As human D52 and D53 genes exhibited different expression patterns in normal breast relative to carcinoma, we compared their functions by stably expressing these in BALB/c 3T3 fibroblastic cells, which were shown previously to be highly responsive to increased mouse D52 expression (28). Six human D52-expressing cell lines (2-1, 2-2, 2-3, 2-4, 2-5, and 2-6), 5 human D53-expressing cell lines (3-1, 3-2, 3-3, 3-4, and 3-5), and 3 vector control cell lines (V1, V2, and V3) were employed in further studies (Fig. 1A and B). Cells were immunofluorescently labeled for human D52 or D53 and phalloidin to examine the intracellular distribution of these proteins and overall cell morphology. Whereas vector control cells not expressing D52 or D53 showed the expected fibroblastic morphology with prominent actin stress fibers (Fig. 1C, top row; data not shown), cells expressing moderate to high levels of D52 or D53 were smaller, with dense phalloidin staining and few actin stress fibers (Fig. 1C, middle and lower rows, respectively). We consistently observed in situ heterogeneity of D52 and D53 levels within transfected cell lines, with individual cells showing low/undetectable or moderate/high levels of D52 or D53 (Fig. 1C, middle and bottom left, respectively). Transfected cells expressing low/undetectable D52 or D53 levels showed fewer alterations to actin stress fibers and were comparable with vector controls (Fig. 1C).

To quantitate these phenomena, 50 to 100 cells from each D52- or D53-expressing cell line were visually scored for both D52 or D53 expression (categorized as undetectable, low, moderate, or high) and actin stress fibers (categorized as none/poor, moderate, or prominent). Cell area (μm²) was calculated using Image ProPlus software, and measurements from individual D52 or D53 cells were grouped. Comparing median cell areas according to D52 or D53 expression status (undetectable/low versus moderate/high) indicated that moderate/high D52 or D53 expressing cells were significantly smaller than cells with low/undetectable D52 or D53 levels (Mann-Whitney U test, P < 0.001, n = 665 for D52 and P < 0.001, n = 502 for D53; Fig. 2A and B). Cell area was also significantly inversely correlated with D52 or D53 expression category (Spearman's rank correlation test r_s = −0.455, P < 0.001, n = 665 for D52 and r_s = −0.326, P < 0.001, n = 502 for D53), but the overall median cell areas of D52- versus D53-transfected cells were not significantly different (Mann-Whitney U test, P = 0.126, n = 1,167). Similarly, actin stress fibers were significantly inversely correlated with D52 or D53 expression category (Spearman's rank correlation test r_s = −0.246, P < 0.001, n = 665 for D52 and r_s = −0.308, P < 0.001, n = 502 for D53; Fig. 2C and D). Future studies will be required to determine whether D52 and D53 expression reduce fibroblast cell area by altering actin stress fibers or whether this occurs through another mechanism.

D52 expression increases proliferation and anchorage-independent growth in 3T3 cells. As previous analyses have identified D52 as a positive regulator of cell proliferation (26, 28), D52, D53, and vector control cell lines were compared using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays over 4 days. Assays were carried out in triplicate for each cell line on four independent occasions, with the results for individual D52, D53, and vector cell lines being grouped for analyses. These analyses indicated significantly increased proliferation in D52-expressing but not D53-expressing cell lines at 3 and 4 days, relative to vector controls (Fig. 3A and B). Colony formation assays were also carried out for all D52, D53, and vector control cell lines in duplicate on three separate occasions. After 14 days of growth, all visible colonies greater than 10 cells were counted in 20 random fields, and the cumulative results from all experiments using D52, D53, and vector control cell lines were grouped for analyses. This indicated that D52- but not D53-expressing cell lines produced significantly more colonies than vector controls (P < 0.001; Fig. 3C and D). Scratch wound-healing assays were subsequently carried out in all cell lines, but these did not reveal significant differences in migration at 12 h post-wounding in four independent experiments (data not shown).

Reduced D52 levels were associated with reduced adherent MCF-7 and SK-BR-3 cell numbers and increased apoptosis in SK-BR-3 cells. Endogenous D52 or D53 levels were transiently reduced in MCF-7 and SK-BR-3 breast cancer cells, which detectably express both proteins (18). The SK-BR-3 cell line is also amplified at the D52 locus (4, 6) and expresses high D52 levels (4, 6, 18). Three different D52 and D53 siRNAs were employed as well as a siRNA targeting GAPDH as a positive control and a non-targeting siRNA, and transfection reagent-treated cells were employed as negative controls (Fig. 4). At 48 to 96 h after transfection, D52 or D53 protein levels were reduced by treatment with relevant siRNAs but were unaffected by GAPDH or nontargeting siRNA or TKO treatment only. Reduced D52 levels were not associated with reduced D53 levels or vice versa (Fig. 4A; data not shown).

When MCF-7 and SK-BR-3 cells transfected with D52 and D53 siRNAs, crystal violet assays revealed reduced adherent cell numbers 48 h post-transfection (Fig. 4B). Areas of crystal violet staining were quantified using Metamorph, and data from the three D52 and D53 siRNAs were pooled and compared with that obtained from nontargeting siControl transfections. There were significantly fewer adherent cells in D52 siRNA-treated MCF-7 and SK-BR-3 cells compared with nontargeting siControl-transfected cells (Student's t test, P < 0.001; Fig. 4C). In contrast, reduced D53 levels were associated with reduced adherent cell numbers in MCF-7 cells only (Student's t test, P < 0.05; Fig. 4C). As reduced cell numbers may reflect increased cell death, we determined whether decreased D52 or D53 expression in MCF-7 and/or SK-BR-3 cells was associated with increased apoptosis using the Cell Death Detection ELISAPLUS kit. Reduced D52 expression in SK-BR-3 cells was associated with significantly increased apoptosis at 48 h post-transfection relative to siControl-transfected cells (Student's t test, P < 0.01; Fig. 4D). Although there was a similar trend in MCF-7 cells, this was not significant (Fig. 4D). Significant increases in apoptosis were not measured in D53 siRNA-transfected MCF-7 or SK-BR-3 cells (Fig. 4D).

D52 and MAL2 expression using immunohistochemical analysis of breast tissue microarrays. As D52 and MAL2 overexpression in cancer has been reported previously (6, 22–24), and these genes were most significantly overexpressed in breast carcinoma in the present study (Table 1), we carried out immunohistochemical analyses of D52 and MAL2 expression in a large cohort of normal breast, in situ, and invasive carcinomas in tissue microarray format. Both D52 and MAL2 immunoreactivity were detected predominantly in the cytoplasm as reported previously for in situ D52 expression in breast cancer samples (6). As predicted, compared with strong pixel counts measured in normal breast tissue cores, pixel counts were significantly elevated in invasive breast tumors (Wilcoxon signed rank test, P < 0.001, n = 176 for D52 and n = 148 for MAL2) and in situ carcinomas (Wilcoxon signed rank test, P < 0.001, n = 95 for D52 and n = 82 for MAL2) from the same patients.

Cut-point determination using expression data distributions and decile analyses indicated that it was most relevant to compare the top 20% of tumors according to D52 or MAL2 strong pixel counts to all other tumors (data not shown). We approximated this cut point to 70,000 pixels (top 20.4% of tumors) for D52 and to 100,000 pixels (top 20.6% of tumors) for MAL2. Use of these cut points indicated that high in situ D52 expression (Fig. 5A), but not high in situ MAL2 expression, was strongly associated with reduced overall patient survival (log-rank test, P < 0.001, n = 357; Fig. 5B). Significant associations between D52 expression and reduced overall survival were also measured when cut points of 100,000 and 50,000 pixels were employed (data not shown), indicating a possible dose-response relationship between in situ D52 expression and patient survival. High D52 expression was significantly associated with reduced overall survival in patients with ER-negative tumors (log-rank test, P < 0.001, n = 85 ER-negative patients; P = 0.284, n = 265 ER-positive patients; Fig. 5C and D) in node-positive patients (log-rank test, P = 0.005, n = 113 node-positive patients; P = 0.393, n = 244 node-negative patients) and in patients with high-grade tumors (histologic grade 3; log-rank test, P = 0.008, n = 104 grade 3 tumors; P = 0.191 n = 247 grade 1/2 tumors). The proportions of tumors with high D52 expression also significantly differed according to tumor grade, with 8 of 88 (9%) grade 1 tumors, 29 of 159 (18%) grade 2 tumors, and 36 of 104 (35%) grade 3 tumors expressing high D52 levels (Pearson's χ² test, P < 0.001, n = 351). A significant positive correlation was also measured between strong D52 pixel counts and tumor grade (Spearman rank correlation coefficient r_s = 0.244, P < 0.001, n = 351). Multivariate analyses identified high D52 expression as an independent predictor of survival (hazard ratio, 2.274; 95% confidence interval, 1.228-4.210; P = 0.009; n = 328) after adjustment for age at diagnosis, node and ER status, and tumor grade.

This study was carried out to investigate whether D52 overexpression in breast cancer passively reflects its gene location at chromosome 8q21.13 or specifically reflects nonredundant functional properties of the D52 protein. Direct comparison of D52-like gene expression in breast tissue samples confirmed that genes overexpressed in breast cancer map to amplified genomic regions. We therefore directly compared the cellular functions of D52 and D53, representing genes that are and are not overexpressed in breast cancer, respectively.

Expressing D52 and D53 proteins in 3T3 cells produced both shared and distinct phenotypes. Common phenotypes included smaller, rounder cells with reduced actin stress fibers, and we also noted significant positive correlations between in situ D52 or D53 and cyclin B1 expression (data not shown), as reported for endogenous D53 expression in MCF-7 cells, and exogenous D53 expression in MDA-MB-231-derived D53-H1 cells (32). Distinct phenotypes represented increased proliferation and anchorage-independent growth in D52-expressing cell lines only, in agreement with the findings of Lewis et al. (28), where mouse D52 was expressed in 3T3 cells, and of previous studies where the D52 isoform PC-1/PrLZ was expressed in different cell types (26, 29). A common limitation of the present and previous studies is the use of fibroblast cell lines, and analyses of D52 overexpression in a breast cell line such as MCF10A may be more informative in terms of the role of D52 role in breast cancer. However, the present study extends previous findings by showing that D53 expression does not similarly increase proliferation and anchorage-independent growth. Increased proliferation in response to D52 expression is therefore likely to be mechanistically independent of the morphologic phenotype induced by both proteins.

Transient knockdown of D52 and D53 protein expression in MCF-7 and SK-BR-3 cells reinforced the existence of nonredundant functions for these proteins in breast cancer cells. Notably, reduced D52 levels were associated with significantly increased apoptosis in SK-BR-3 cells only, which highly overexpress D52 in response to gene amplification (4, 6). This finding supports those of previous studies where reduced expression of other oncogenes increased apoptosis in amplified or overexpressing cell lines only (33–35).

Of the genes analyzed in the present study, D52 and MAL2 were maximally overexpressed in breast cancer and also showed significantly positively correlated gene or protein expression values (data not shown). We therefore examined and compared the clinical significance of D52 and MAL2 expression in a large cohort of human breast cancers, applying digital image analysis to identify tumor subpopulations with high D52 and MAL2 expression. These analyses showed that high D52 but not MAL2 expression was associated with reduced overall survival in the breast cancer cohort examined, and particularly in ER-negative, node-positive, and high-grade tumor groups, and was an independent predictor of patient survival. This indicates that high D52 expression may significantly promote further tumor progression in patients with ER-negative tumors and/or more aggressive disease at diagnosis. Overall survival analyses in patients grouped according to adjuvant treatment status did not identify high D52 expression as predicting response to radiotherapy, chemotherapy, or hormone therapy (data not shown). We therefore hypothesize that high D52 expression contributes to a more aggressive tumor phenotype, which is also supported by the fact that high D52 expression was significantly more frequent in grade 3 tumors. High D52 expression in breast tumors may therefore represent a clinically useful marker following development of a reliable semiquantitative immunohistochemical assay (36).

Several studies have now linked poor outcome or adverse breast cancer histology with the gain of chromosome 8q21 (37–39), and D52 has been included in signatures associated with adverse prognosis in breast (40, 41) and prostate cancer (42) and reduced survival in mantle cell lymphoma (43). Proliferation genes are emerging as a common driving force in prognostic cancer gene signatures (44). As D52 positively regulates cell proliferation and anchorage-independent growth (26, 28, 29), the reduced survival associated with high D52 expression in breast cancer likely reflects this contributing to a more proliferative and aggressive cancer phenotype. The identification of nonredundant cancer-promoting properties for D52 also support the specific targeting of D52 expression in cancer, as do the results of recent mapping studies indicating narrow gene amplification peaks at 81 Mb, corresponding to the position of the D52 locus (2–5). Previous associations between chromosome 8q21 gain in breast cancer and poor patient outcome (37–39) may therefore reflect high D52 expression in tumors. These combined data suggest that reducing D52 expression or inhibiting D52 function may improve outcomes in patients with D52-amplified or D52-overexpressing breast cancer and possibly in other cancer types where D52 is overexpressed.

No potential conflicts of interest were disclosed.

We thank Prof. Peter Gunning (The Children's Hospital at Westmead) for support and for providing the PG307 expression vector, Drs. Daniel Catchpoole, Geraldine O'Neill, and Rosemary Balleine (Westmead Millennium Institute) for scientific discussions, Angela Bailey and Jayne Hardy for excellent technical assistance, and Jill Tinning (RPH) for providing clinical data associated with breast tissue arrays.