Introduction
Many genes implicated in the sequential, multi-step process of metastasis have been identified [
1,
2]. One of the genes identified is Mena, a member of the Ena/VASP family of proteins, which plays a key regulatory role in actin polymerization [
3‐
6]. It has been shown to be involved in intravasation and motility of tumor cells in model systems [
7,
8]. In breast cancer tumors, its expression has been shown to be a key element of the tumor microenvironment for metastasis (TMEM), whose density correlates with risk of distant metastasis [
9]. Importantly, Mena deficiency in the PyMT mouse breast cancer model suppresses intravasation, eliminates mortality and morbidity, and greatly reduces the frequency of metastatic dissemination to the lung [
10].
Mena is alternately spliced to give rise to multiple protein isoforms that are differentially expressed during tumor progression [
11,
12]. Two of the best characterized isoforms are Mena
INV, expressed exclusively in invasive tumor cells, and Mena11a, an epithelial-specific isoform expressed in primary breast carcinomas and down-regulated in invasive tumor cells [
7]. Mena
INV, (originally termed Mena
+++), expression confers a potent pro-metastatic phenotype when expressed in breast cancer cells by potentiating their chemotactic response to epidermal growth factor (EGF), thereby enhancing their ability to engage in efficient streaming motility via increasing their paracrine signaling with macrophages [
3,
13,
14]. The Mena11a, a non-metastatic isoform, contains an alternately-included exon encoding a 21 amino acid insertion located in the carboxy-terminal [
7]. Consistent with its down-regulation during tumor progression
in vivo [
11,
15], Mena11a is expressed in epithelial-like but not mesenchymal-like tumor cell lines [
16], and is down-regulated when human mammary epithelial cells undergo epithelial to mesenchymal transition (EMT) [
12]. Mena11a expression in breast cancer cells causes formation of poorly metastatic tumors with a highly epithelial architecture that are not capable of responding to EGF chemotactic cues
in vivo [
14]. Therefore, Mena11a expression positively correlates with, and enforces epithelial non-metastatic phenotypes, and negatively correlates with, and suppresses mesenchymal metastatic phenotypes
in vitro and
in vivo.
The mechanistic role of MenaINV raises the hypothesis that measurement of this isoform in tumor tissue could be valuable for prediction of the risk of metastasis. Thus, it is reasonable that the fraction of Mena containing the 11a exon may reflect the abundance of poorly-metastatic tumor cells and, therefore, correlate with decreased metastatic risk. Thus far, no evidence exists 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. Therefore, the overall fraction of Mena lacking 11a may reflect the presence of cells with the potential to express pro-metastatic Mena isoforms. We describe here a multiplexed quantitative immufluorescence-based method (MQIF) in which the fraction of Mena protein that may promote invasion inferred by subtraction of the non-invasive isoform from the total Mena present in tumors. We call this biomarker Menacalc and in the study reported here relate it to metastasis using risk of death from breast cancer.
Discussion
From a mechanistic standpoint, measurement of the relative abundance of Mena isoforms associated specifically with poor or high metastatic potential represents an attractive approach for prediction of metastasis. In this paper, we describe the development of a marker generated using a subtractive approach to estimate the abundance of Mena lacking its anti-Metastatic Mena11a isoform. The advantage of this approach is the ability to infer the actual (Mena11a) or potential (Menacalc) abundance of anti- or pro-metastatic Mena isoforms in a single assay.
Another related mechanism-based assay used to assess metastasis risk is measurement of TMEM density [
9]. TMEM counts performed in a small case control study of 30 pairs of invasive breast cancer patients in which half of them (30) developed distant metastasis, while another 30 patients did not. TMEM counts were significantly higher in the patient group that developed distant metastasis compared to the patient group that had only localized breast tumor (
P = 0.00006), independent of other clinicopathologic variables [
9]. Roughly, the risk of distant metastases doubled as for every 10 TMEMs measured in the area of patient tissue analyzed. Expression of Mena
INV, but not Mena 11a, correlates with TMEM density supporting the hypothesis that Mena11a is not involved in the assembly of TMEM intravasation sites [
15]. While it will be interesting to compare the TMEM assay with the Mena
calc method, such a comparison is technically challenging. The TMEM assay requires larger tissue specimens (whole slides) to count the morphologically defined triad of a Mena-expressing tumor cell, macrophage and blood vessel all contacting each other. While the Mena
calc method may be assessed on whole slides, this initial effort is limited to TMA-based material. While we believe that our MQIF method will be suitable for clinical usage to identify patients that are at high risk for developing metastasis, it awaits testing on routine whole slide specimens. However, our results on TMAs suggest that this approach may permit higher throughput, less sensitivity to heterogeneity and, perhaps, increased sensitivity given the larger tissue requirements and more laborious method required for TMEM counts.
Previously, we reported a significant correlation between Mena
INV mRNA levels measured in fine needle aspirate biopsies of clinical samples and TMEM frequency in histological sections from the same patient [
15]. Clearly it will be interesting to evaluate how Mena
calc compares to Mena
INV levels; however, such analyses will require development of a Mena
INV-specific antibody that can be used for tissue staining. We predict, however, that the two metrics (Mena
calc and Mena
INV) will likely convey different information. Mena
calc measures the total amount of all Mena isoforms expressed that do not contain the 11a exon (pan-Mena-Mena11a). Therefore, Mena
calc represents all Mena isoforms other than the anti-invasive/metastatic Mena11a. While Mena
INV is expressed during tumor progression and has pro-metastatic activity, other Mena isoforms lacking the INV exon, including Mena "classic", which lacks all known alternately-included exons, Mena++ and Mena+ [
3], are still expressed in tumors and in invasive tumor cells isolated from mouse breast cancer models [
11]. Therefore, we do not propose that Mena
calc is a surrogate measure for Mena
INV. Instead, we propose that this metric measures the loss of the anti-metastatic activity of Mena11a. Whether a combined evaluation of Mena
calc and Mena
INV provides more powerful prognostic information than either alone will be an important focus of future efforts.
While the results of this investigation suggest that our assay predicts tumor aggressiveness, this study has a number of limitations. First, we used disease-specific survival as a surrogate marker for metastasis. We believe this is reasonable since essentially every death from breast cancer is due to metastasis. Another, more significant weakness of the study is that the cohorts are both retrospective collections from a single institution. Furthermore, this is a post-hoc study design and the conclusions will need to be validated on other cohorts prior to clinical application. The fact that Menacalc is able to stratify patients in the multivariate Cox model analysis on the combined cohorts provided increases our confidence in the assay. However, it will be important to confirm this observation in larger, prospectively collected, multi-institutional cohorts.
Competing interests
Drs. Condeelis and Gertler are stockholders in Metastat, a company with the exclusive license to the Mena patent suite. Drs. Rohan and Rimm serve as consultants to Metastat. Dr. Rohan is a stockholder of Metastat. Drs. Rimm and Camp are founders, consultant and stockholders of HistoRx.
Authors' contributions
SA, DLR, FBG and JSC participated in the concept, design and coordination of the study, and the drafting of the manuscript. SA was also involved in the execution, acquisition, analysis and interpretation of all data. DLR participated in providing financial support, patient cohorts, data analysis and interpretation of all data. FBG and JSC provided the Mena 11a antibody used in this work. TER, RLC, XX and JL helped with all statistical analysis and the drafting of the manuscript. MB helped in generating and validating the Mena 11a antibody and helped in drafting the manuscript. All authors read and approved the final manuscript for publication.