Here, we show that experimental manipulation of transgelin expression levels affects the metastatic potential of CRC cells in vivo. In the experimental metastasis assay, control HCT116 cells and RKO cells with stable overexpression of transgelin produced more and larger tumors in the lung than their isogenic, low transgelin-expressing counterparts. The widely used tail vein assay provides a composite measure of the efficiency of sequential events required for metastasis, including survival of isolated cells in the bloodstream, extravasation into the lung, and subsequent colonization and proliferation to form observable tumors [
27]. Each of these steps has a parallel counterpart that was previously measured in cell culture [
1], including resistance to anoikis, invasiveness in a Transwell assay, and colony formation at low density, respectively. Although a number of proteomic studies (but not all) have shown a correlation between transgelin levels and advanced stage disease and poor prognosis, data here imply a direct causal relationship between transgelin levels and metastatic potential in vivo.
Mechanism of the transgelin effect
In normal cells and tissue, transgelin interacts directly with the actin, influencing cell motility [
3‐
5]. This mechanism may apply in cancer cells as well. In addition to direct interaction with actin, transgelin may also influence cell phenotype by affecting gene regulation. This idea is based on transgelin’s partial nuclear localization, its homology to transgelin 3, which has transcriptional regulatory activity [
28], and other evidence [
1,
25,
26]. To address the influence of transgelin on gene regulation in greater depth, we performed comprehensive gene expression profiling comparing low- and high-expressing RKO cells. Altering transgelin levels significantly affected the expression of a set of ~250 other genes, with statistical overrepresentation of cytoskeletal and actin-binding proteins, together with various other regulatory genes.
The five most-highly upregulated genes in Table
1 are connected with cancer, and in some cases specifically with tumor metastasis. HOOK1 (up-regulated 96-fold) interacts with microtubules and is up-regulated in breast cancer [
29]. SDCCAG8 (up-regulated 86-fold) organizes the centrosome and is related to a colon cancer autoantigen [
30]. ENAH (up-regulated 18-fold), which is the mammalian enabled homology and is also known as Mena, is a transcriptional target of the wnt/beta-catenin pathway [
31]. It facilitates cell invasion via phosphatidylinositol 3-kinase-dependent local accumulation of actin filaments [
32]. Increased ENAH/Mena expression levels correlate with invasiveness of breast and salivary gland tumors[
33,
34], and are also seen in colorectal cancer and in polyps with high grade dysplasia [
35]. TNS1 (up-regulated 14-fold), which is also known as Tensin 1, has actin cross-linking activity and localizes to focal adhesions. Prior studies show that increased expression of Tensin 1 correlates with tumor cell migration [
36,
37]. A sialyltransferase (ST8SIA4) has been shown to promote metastatic dissemination in pancreatic cancer by interfering with E-cadherin dependent cell adhesion [
38].
Of the most highly down-regulated genes, EMB (down-regulated 16-fold) is thought to mediate adhesion to the extracellular matrix [
39]. BCL11B (down regulated 12-fold) is a tumor suppressor that is widely mutated in acute lymphocytic leukemia [
40] and has recently been shown to be methylated, and thus likely down-regulated, in prostate cancers [
41]. PTPRD (down-regulated 11-fold), a tumor suppressor that is frequently mutated in glioblastoma [
42,
43], Ewing’s sarcoma [
44], lung cancer [
45], cutaneous squamous cell carcinoma [
46], and laryngeal squamous cell carcinoma [
47]. PTPRD has previously been shown to suppress colon cancer cell migration in cooperation with β-catenin/TCF signaling [
48].
Based on an analysis of a more limited set of genes, we had previously hypothesized that transgelin regulates the epithelial-to-mesenchymal transition (EMT). In one cell line, fibronectin and vimentin (mesenchymal markers) and occludin (an epithelial marker) were affected. In another, only fibronectin was affected. In the RKO model, however, neither these nor other markers of EMT were altered. Thus the effect of transgelin on EMT, if any, appears to be cell-line specific.
It remains unclear whether the effect of transgelin on gene regulation occurs by a direct or indirect mechanism. Transgelin localized partially in the nucleus of the CRC cells; it shares 85 % similarity to
TAGLN3, which is believed to be a transcriptional regulator [
28]; Bioinformatic analysis using the DP-bind Web server (
http://lcg.rit.albany.edu/dp-bind) identified several segments of transgelin as having high DNA-binding potential (data not shown); studies also suggests that nuclear actin-binding proteins participate widely in the transcription processes (reviewed in [
49]). We were, however, unable, to detect direct association of transgelin with RNA polymerase II or chromatin (HMZ, YYF, and YL, unpublished results), which suggests that transgelin most likely works indirectly, through interaction with other transcriptional regulatory proteins, rather than by direct interaction with the polymerase or template.
Transgelin in the context of human CRC
The original impetus for the studies reported here was to identify biomarkers that could be quantified at the time of surgical resection and used to predict metastatic risk. Transgelin is abundant in normal tissue and is not a marker of cancer
per se. In our work, however, it discriminated well between node-positive and node-negative CRC specimens [
1].
Evaluation of potential biomarkers should ideally include evidence that they influence the biological process of interest; that is, adding or subtracting transgelin in a given context influences phenotypic behaviors associated with metastatic risk. Combining the results presented with those of a previous study [
1], we have created four isogenic pairs (HCT116, SW480, RKO, and DLD-1), three have been characterized with respect to in vitro phenotypes (HCT116, SW480, and RKO) and two (HCT116 and RKO) have been characterized in an animal model of experimental metastasis. The combined results are consistent with a contribution of transgelin to metastatic potential.
Human CRC is characterized by genomic instability and variability, and caution is warranted in generalizing from results with individual cell lines. However, laboratory findings provide a rationale for continuing to explore the potential of transgelin, perhaps in conjunction with other biomarkers, as a predictor of individual risk. As noted earlier, proteomic studies of human cancer are not in universal agreement with respect to the role of transgelin, and it will be important to identify the root cause of the discrepancies, which could relate either to different patient populations or to technical factors. Separately, the finding that a number of gene products are co-regulated with transgelin suggests that they may be attractive targets for drug therapy in advanced CRC.