Archival specimens collected between 2005 and 2012, and 50 patients with invasive ductal carcinoma of no special type (IDC-NST) (Fig. 1A), 45 patients with ductal carcinoma in situ (DCIS) (Fig. 1B) and 10 patients with intraductal papilloma of the breast (Fig. 1D) were enrolled in the study. Non-neoplastic duct tissue was obtained from each specimen to serve as a control (Fig. 1C). All patients were surgically treated with either breast-conserving lumpectomy or modified radical mastectomy and axillary lymph node dissection at Asahi General Hospital (Chiba, Japan). All samples were collected with approval from the ethics committee of Asahi General Hospital. The clinicopathological characteristics of each patient are shown in Tables I and II.

The primary tumor specimens were fixed in 10% buffered formalin and embedded in paraffin using standard tissue processing methods. Pathological tumor staging was determined using the current tumor-node-metastasis system (UICC). Diagnoses and histology were confirmed by two pathologists (Noriyuki Tanaka and Kenichi Harigaya) who reviewed the hematoxylin-eosin (H&E)-stained slides prepared from the paraffin blocks.

Ten micrometer-thick membrane sections (Leica Microsystems) were prepared from formalin-fixed paraffin-embedded (FFPE) tissue using standard protocols. The slides were air-dried and stained with toluidine blue. Laser dissection was performed using a laser capture microdissection microscope (Leica AS LMD; Leica Microsystems, Wetzlar, Germany) with a pulsed 337 nm UV laser according to the manufacturer’s protocols. The size of each dissected specimen was ≥2 mm². Total RNA was purified from dissected tissue using an RNeasy FFPE kit (Qiagen, Venlo, The Netherlands) according to the manufacturer’s protocols. cDNA was synthesized from the extracted total RNA with Primescript reverse transcript reagent (Takara) according to the manufacturer’s protocols.

Semi-quantitative real-time polymerase chain reaction (PCR) was then performed using the ABI PRISM 7000 (Perkin-Elmer Applied Biosystems, Foster City, CA, USA) thermal cycler with PreMix ExTaq reagent (Takara), as previously described (14). The primers were as follows: MENA (sense, 5′-GCGCAGAATATCAAGTGCTG-3′; antisense, 5′-TCCCAGCACAGAGTTTAGAGG-3′), MENA^11a (sense, 5′-TTTTGACAACAGGTCCTATGATTC-3′; antisense, 5′-CTTCCGTCTGGACTCCATTG-3′), MENA^INV (sense, 5′-CCATGATGCATGCCTTAGAA-3′; antisense, 5′-TCCTGGGTAGCAGTAACCTTG-3′), human GAPDH (sense, 5′-AGCCACATCGCTCAGACAC-3′; antisense, 5′-GCCCAATACGACCAAATCC-3′). The TaqMan Probes (UPL, Roche) used for each variant were as follows: hMena (TaqMan no. 34), hMena^11a (TaqMan no. 26), hMena^INV (TaqMan no. 26), and GAPDH (TaqMan no. 60). The PCR protocol consisted of 30 sec at 95°C followed by 60 cycles of 5 sec at 95°C and 31 sec at 60°C. The relative expression level of hMena variants was calculated by the comparative threshold cycle (Ct) method using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the internal control (15). All experiments were performed in triplicate and the results were expressed as mean ± SD.

Human breast cancer cell lines MDA-MB231, BT549, MDA-MB468, MCF-7, T47D, ZR75-1 and human cervical cancer cell line HeLaS3 were purchased from American Type Culture Collection (Rockville, MD, USA). All cell lines except HeLaS3 were cultured in Iscove’s modified Dulbecco’s medium (IMDM; Invitrogen) supplemented with 10% fetal bovine serum. HeLaS3 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Niccui) supplemented with 10% calf serum. All cell lines were maintained under 5% CO₂ at 37°C.

Anti-E-cadherin antibody (BD610182) was purchased from BD Biosciences (San Jose, CA, USA). Anti-vimentin antibody (sc-6260) was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Anti-actin antibody (A 2066) was purchased from Sigma (St. Louis, MO, USA). Primary polyclonal antibodies against hMena variants were raised in rabbits against amino acid sequences AQSKV TATQD STNLR CIFC (hMena^INV) and RDSPR KNQIV FDNRS YDSLH (hMena^11a). The obtained antibodies were affinity-purified using each immunizing peptide.

Whole cell lysates were prepared with ice-cold RIPA buffer (50 mM Tris-HCl pH 7.5, 1% nonidet P-40, 150 mM NaCl, 0.1% SDS, 0.5% deoxycholic acid) containing 1 μg/ml leupeptin, 1 μg/ml pepstatin, 1 μg/ml aprotinin, 1 mM DTT, 1 mM NaVO₄ and 0.5 mM PMSF. The supernatants were recovered as total cell lysates following centrifugation. Aliquots of the cell lysates (50 μg of protein) were separated by 10% SDS-PAGE and transferred to PVDF membranes (Bio-Rad Laboratories, Hercules, CA, USA). Primary antibodies bound to their antigens on the membranes were detected using appropriate HRP-conjugated secondary antibodies (Amersham Bioscience, Piscataway, NJ, USA) and a Super Signal chemiluminescence detection system (Pierce, Rockford, IL, USA) or the Lumi-LightPLUS western blotting substrate (Roche Diagnostics, Basel, Switzerland) according to the manufacturer’s instructions.

Data are summarized in bar graphs. Bars represent the mean and whiskers the standard deviation (SD). Statistical analysis was performed using Microsoft Excel (Microsoft Corp., Seattle, WA, USA). Paired-sample t-tests, χ² test, or Kruskal-Wallis tests were used as appropriate. Differences were considered statistically significant when the P-value was <0.05.

We sought to determine whether the mRNA expression of hMena splice isoforms hMena^INV and hMena^11a is altered in different IDC-NST lesions [invasive areas (IA-IDC-NST, Fig. 1A) and non-invasive intraductal lesions (NIDL-IDC-NST, Fig. 1B)] compared with corresponding non-neoplastic duct epithelium (Fig. 1C) (50 cases). Fig. 2A demonstrated that an increase in hMena^INV mRNA expression was found in the different IDC-NST lesions IA-IDC-NST and NIDL-IDC-NST, while no detectable expression was observed in non-neoplastic duct epithelia. More intense augmentation of hMena^INV mRNA expression was demonstrated in IA-IDC-NST than NIDL-IDC-NST (P<0.05). We further examined whether there were any differences in hMena^INV mRNA expression among three different subtypes of IDC-NST according to the WHO classification (16) of histological grading (tubule formation, nuclear pleomorphism, and mitotic accounting) or molecular phenotype (luminal A, luminal B, and basal subtypes according to the presence or absence of estrogen receptor, progesterone receptor, or Her2 protein). Fig. 2B showed that higher expression of hMena^INV mRNA was significantly detected in IA-IDC-NST lesions of grade 3 breast carcinomas compared with either those of grade 1 tumors (P<0.05) or those of grade 2 tumors (P<0.05). Additionally, no significant differences in hMena^INV mRNA expression were found in NIDL-IDC-NST lesions of the three different subclasses. Furthermore, a significant increase in hMena^INV mRNA expression was found in IA-IDC-NST lesions of three different molecular phenotypes; basal subtype expressed higher levels of hMena^INV mRNA than either luminal A subtype, or luminal B subtype (P<0.05; Fig. 2C). However, we did not find any statistically significant difference between luminal B and basal subtypes. In NIDL-IDC-NST lesions, there was no statistically significant difference in hMena^INV mRNA expression among the three different molecular phenotypes. Taken together, it would be reasonable to conclude that higher hMena^INV mRNA expression is found in more aggressive histological and molecular subtypes of IA-IDC-NST lesions. Furthermore, our data suggest that no statistically significant difference in hMena^INV expression was found in NIDL-IDC-NST lesions, but increased expression was found in IA-IDC-NST in accordance with tumor progression. It is well known that determination of these subtypes reflect patient prognosis (17), and, therefore, our results are comparable with a previous report that showed increased expression of hMena^INV could confer a potent pro-metastatic phenotype when expressed in breast cancer cells (18,19). Our results also showed that levels of hMena^INV expression also tended to be increased in non-invasive ductal carcinoma in clinical breast carcinoma specimens according to histological grade, but this was not statistically significant. These results indicate that the measurement of hMena^INV mRNA in IA-IDC-NST but not NIDL-IDC-NST may be useful in predicting patient prognosis from histological samples of IDC-NST. It should be stressed that mutually-exclusive alternative hMena mRNA splicing would not necessarily be observed only in invasive breast carcinoma cells in clinical specimens and that cell sheets of non-invasive intraductal lesions in IDC-NST produced significant amounts of hMena^INV mRNA.

Next, we examined the expression of hMena^11a mRNA at the afore-mentioned lesions of IDC-NST. The expression of hMena^11a mRNA was hardly found in non-neoplastic breast duct epithelium but was dramatically increased in the different lesions of IDC-NST, as in the case of hMena^INV mRNA. Fig. 2D shows that significant expression of hMena^11a mRNA was consistently found in NIDL-IDC-NST as well as IA-IDC-NST in contrast to non-neoplastic duct epithelia, while the relative expression in IA-IDC-NST was downregulated to approximately 85% that of the level in NIDL-IDC-NST. This reduction reflects previous reports that Mena^11a is downregulated in invasive tumor cells (11). Compared with the elevation in hMena^INV mRNA expression in accordance to adverse histological grade, hMena^11a mRNA expression did not change among the three subgroups classified according to WHO histological grades or molecular phenotypes, as shown in Fig. 2E and F. Rather, grade 3 subtype of both NIDL-IDC-NST and IA-IDC-NST tended to have increased hMena^11a mRNA expression, although this was not statistically significant. Our results indicated that significant hMena^11a mRNA expression was found in different NIDL and IA lesions of IDC-NST but was decreased in IA-IDC-NST, which is supposed to undergo tumor progression. However, our results also showed that hMena^11a mRNA expression was not downregulated in different lesions with either histological or molecular tumor progression. Accordingly, our results do not necessarily coincide with previous reports in vitro and in vivo that hMena^11a is downregulated in invasive tumor cells (11,20). Collectively, our results showed that cell sheets of non-invasive intraductal lesions in IDC-NST produced significant amount of hMena^11a mRNA as well as hMena^INV mRNA. This would reflect whether some populations of invasive cells are intermingled in intraductal lesions of NIDL-IDC-NST or some populations of non-invasive intraductal lesions could produce both hMena^INV and hMena^11a mRNA simultaneously. Nevertheless, there has been no evidence indicating that both the INV and 11a exons are included in the Mena mRNA at the same time or expressed at high levels within the same cell (21).

Several lines of evidence indicate that molecular expression of hMena^11a is strictly regulated during breast carcinoma development and is predominantly found in the non-invasive stage of breast carcinoma (12,20). However, our results showed that both hMena splice isoforms were dramatically increased in the different IDC-NST lesions, IA-IDC-NST (Fig. 2A) and NIDL-IDC-NST (Fig. 2D). Additionally, our results also revealed that the expression of hMena^INV is further augmented in cells at the invasive front, while hMena^11a is downregulated in cells at the invasive front. It should be stressed that cell sheets of NIDL-IDC-NST produced hMena^11a mRNA as well as significant amounts of hMena^INV mRNA in clinical breast carcinoma specimens. This would reflect whether some populations of invasive breast carcinoma cells are intermingled in non-invasive intraductal lesions of IDC-NST or some cell populations of non-invasive intraductal lesions could produce both Mena^INV and Mena^11a, simultaneously. To explore these possibilities, we examined whether cell populations in ductal carcinoma in situ (DCIS) contain hMena splice isoform mRNA. We examined 45 cases of DCIS, 10 cases of intraductal papilloma, and 55 lesions of the corresponding non-neoplastic duct epithelia around the in situ carcinoma and papilloma. Fig. 3A and B showed that cells of DCIS constantly expressed significant amounts of hMena^11a mRNA as well as hMena^INV mRNA. In papilloma lesions (n=10), weak expression of hMena^11a was also found, but hMena^INV mRNA expression was not detectable. Accordingly, cells in DCIS produced increased amounts of both hMena^INV and hMena^11a mRNA in contrast to non-neoplastic epithelia. We further examined the different DCIS subtypes according to either WHO histological classification; the expression levels of hMena^11a and hMena^INV mRNA did not differ among the three different grades of DCIS lesions (Fig. 3C and D), although we did not investigate the molecular phenotypes. Our results are comparable with a recent report that showed the rate of local recurrence for low- and intermediate-grade DCIS is not significantly different from the rate of local recurrence for women with high grade DCIS (15.1%) (22). Fig. 3E showed that among three different breast carcinoma lesions, cells in DCIS always showed the highest level of hMena^11a mRNA expression, followed by those in NIDL-IDC-NST. Cells of IA-IDC-NST consistently showed the lowest levels of hMena^11a mRNA expression among the three different lesions. Conversely, the expression of hMena^INV mRNA was reversed and increased in cells of NIDL-IDC-NST, and showed the highest expression levels in cells of IA-IDC-NST (Fig. 3F). These results indicated that the expression of hMena^11a suppressed metastatic potential and hMena^INV expression promoted tumor progression. Previous reports showed that Mena^INV is exclusively expressed in invasive tumor cells in in vitro and in vivo (18,23). Nevertheless, our results indicate that cells in non-invasive breast duct carcinoma produce Mena^11a mRNA as well as Mena^INV mRNA in clinical breast carcinoma samples.

To validate our results, we examined the expression of hMena splice isoforms in several non-invasive and invasive cancer cell lines by western blotting. Fig. 4 shows the expression of hMena splice variants in several breast cancer cell lines. Seven cancer cell lines, including four E-cadherin-positive and vimentin-negative cell lines and three E-cadherin-negative and vimentin-positive cell lines were used. The E-cadherin-positive cell lines are epithelial and show non-invasive phenotypes. All four lines expressed hMena^11a protein. However, the E-cadherin-positive and vimentin-negative non-motile breast cancer cell line, MDA-MB468, simultaneously expressed both hMena^INV and hMena^11a. Conversely, E-cadherin-negative and vimentin-positive HeLaS3 cells presumably undergo EMT, and were found to express hMena^11a protein but not hMena^INV protein. These results suggest that a proportion of the non-invasive carcinoma cells could express hMena^INV as well as hMena^11a, while invasive carcinoma cells such as HeLaS3 express only hMena^11a but not hMena^INV. These findings are fairly consistent with our results in clinical breast cancer samples, and suggest that a subset of non-invasive cancer cells simultaneously express hMena^11a and hMena^INV. Accordingly, our results indicate that the differential expression of either hMena^INV or hMena^11a in cancer cells is not strictly regulated during tumor progression.

To examine the expression of whole hMena mRNA expression in different non-neoplastic and neoplastic breast epithelial lesions, we investigated 210 foci (non-neoplastic lesions: n=105; papilloma: n=10; DCIS: n=45, IDC-NST: n=50). Fig. 5 showed that levels of whole hMena mRNA expression were not statistically different among the non-neoplastic and neoplastic breast epithelial foci used in this study.

Our data resulting from semi-quantitative mRNA expression in microdissected samples of clinical surgical specimens are compatible with the mRNA expression of hMena^INV and hMena^11a splice variants in previous studies (18,24) using xenograft models; however, whole hMena expression levels were different from our results. These researchers claimed that there was 3–4-fold augmented expression of whole hMena during tumor progression. Although we could not determine the cause of this discrepancy, we speculate that it stems from different model systems; we investigated clinical human breast cancer samples while the previous report used xenograft models.