Background
Syndecans are a conserved family of transmembrane heparan sulfate proteoglycan receptors that participate in cell-cell and cell-extracellular matrix adhesion and the actions of peptide growth factors in normal tissues [
1,
2]. In land vertebrates, the syndecan gene family contains 4 members of which syndecan-4 is the most widely expressed [
3]. Syndecan-1 is expressed predominantly in epithelia, but is also found on fibroblasts, myoblasts and differentiating B cells and is up-regulated in multiple myeloma [
1,
2,
4]. Syndecan-1 null mice are resistant to Wnt1-activated mammary tumors and carcinogen-induced tumor development and have increased susceptibility to allergen-induced airway inflammation [
5‐
7]. Syndecan-1 null mice are also defective in repair of corneal and epidermal wounds [
8,
9]. Overall, these data demonstrate important contributions of syndecan-1 in the regulation of epithelial homeostasis, proliferation and migration.
In cell culture models, E-cadherin dependent loss of cell-surface syndecan-1 is associated with epithelial-mesenchymal transition and the epithelial phenotype is restored by over-expression of syndecan-1 [
10‐
12]. Syndecan-1 over-expression also inhibits cell invasion into collagen gels [
13]. Similarly, malignant transformation of Caco-2 epithelial cells is associated with loss of syndecan-1 [
14]. Although these studies implicated tumor suppressor roles of syndecan-1, the relationship of syndecan-1 expression to tumor progression or clinical outcomes in human carcinomas has proved to be more complex. Thus, in gastric cancers, tumor cell expression of syndecan-1 has been correlated with patient survival and the loss of syndecan-1 with a poor prognosis [
15,
16]. A fraction of patients have abnormal up-regulation of syndecan-1 on stromal cells in the vicinity of the tumor, and this also correlates with poor prognosis [
16]. Similar observations have been made in other carcinomas, for example [
17‐
20]. However, for pancreatic and breast carcinomas there are conflicting reports that increased syndecan-1 expression correlates with a poor prognosis or resistance to chemotherapy [
21‐
24]. These complex data point to a need for continuing assessment of syndecan-1 status in relation to the clinico-pathological characteristics of carcinomas from different tissue sources.
Colorectal carcinoma is the third most common form of cancer in both men and women in the USA and Europe. It remains a major cause of cancer mortality, with a 5 year survival rate of 60%, and its incidence is expected to increase in association with the ageing of western populations [
25]. The major therapeutic approach is surgical resection and there is an urgent need to identify new biomarkers to improve strategies for adjuvant therapies or post-operative monitoring. We and others have recently demonstrated that expression of the actin-bundling protein, fascin, has prognostic significance in colorectal adenocarcinoma [
26,
27]. In non-transformed cells, syndecan-1 acts as a transducer of extracellular matrix cues that regulate the organization of actin and fascin in lamellipodia [
28]. It has been reported that syndecan-1 expression is decreased in colorectal adenocarcinomas in comparison to adenomas and the normal tissue [
29‐
31], and that reduced expression correlates with the incidence of local metastases [
30]. Increased levels of syndecan-1 in the local stroma have also been described [
32]. However, the prognostic relevance of changes in syndecan-1 expression in colorectal carcinoma remains unclear from the published studies [
30,
31]. In view of the health burden imposed by colorectal carcinoma, we examined a tissue microarray of clinically annotated colorectal carcinoma specimens and report on our analysis of epithelial or stromal syndecan-1 expression in relation to patient outcomes and the expression of the recently identified potential biomarker, fascin.
Discussion
The results of our study demonstrate that the loss of expression of syndecan-1 from colonic epithelial cells in colorectal adenocarcinomas correlates with tumor TNM stage and incidence of local lymph node metastasis but nevertheless does not correlate statistically with patient survival. Two previous studies reached conflicting conclusions on whether reduced syndecan-1 correlated with decreased patient survivial [
30,
31]. Our data are in agreement with the study of Lundin et al. [
31]. We also examined the relationship between syndecan-1 and a recently identified novel independent prognostic factor for colorectal carcinoma, fascin [
26,
27]. A novel significant correlation between decreased syndecan-1 staining on tumor cells and increased stromal fascin staining was detected.
Our findings from a large, clinically annotated tissue microarray of colorectal carcinoma specimens add to the body of evidence that loss of epithelial syndecan-1 is a general feature of carcinoma progression. In agreement with other analyses of colorectal carcinoma, loss of epithelial syndecan-1 correlated with tumor TNM stage [
29‐
31] and incidence of metastases to local lymph nodes [
30,
31]. Epithelial to mesenchymal transition (EMT) is a major process in tumor progression and loss of expression of syndecan-1 is well established to regulate aspects of EMT [
1,
10,
11]. Thus the loss of epithelial syndecan-1 is likely permissive for development of higher stage, more biologically aggressive tumors. The correlation of loss of syndecan-1 with male gender documented in our study (Table
1) has not been observed in previous studies and the biological significance of this observation is unclear at this time. Loss of syndecan-1 did not correlate with tumor location. Thus there does not appear to be a close biological relationship between syndecan-1 status and microsatellite instability, which is strongly associated with tumor location in the proximal colon [
34].
In a minority of the specimens (16.6%) stromal staining for syndecan-1 was increased in comparison to the normal tissue. This percentage is markedly lower than in one prior report, where stromal syndecan-1 immunoreactivity was observed in 58% of specimens [
31], but is consistent with an analysis of micro-dissected tissues that identified 7.4% of specimens to have elevated syndecan-1 content, largely due to stromal expression [
32]. By comparing
in situ hybridization with immunohistochemistry, the latter study demonstrated that stromal syndecan-1 immunoreactivity is due to expression of the SDCN1 transcript in a stromal cell population, likely myofibroblasts [
32]. It is likely that stromal syndecan-1 has impact on the tumor microenvironment by alterations to the retention of heparin binding growth factors and extracellular matrix components in the vicinity of the tumor. These factors have been proposed to facilitate tumor cell invasion and, in general, the tumor stroma has an important role in cancer development [
1,
4,
35]. Nevertheless, from the current analysis, the presence of stromal syndecan-1 did not correlate with a more biologically aggressive tumor phenotype or altered patient survival outcomes. A previous analysis is consistent with this conclusion [
31]. We observed a wide variation of staining intensity for stromal syndecan-1 and also variations in the distribution of staining, which tended to be patchy within limited areas of the stroma. We speculate that expression of stromal syndecan-1 might be transient and reflective of short-term phenotypic changes in stromal myofibroblasts.
Although our data did not support syndecan-1 status as an independent prognostic factor in colorectal carcinoma, it was of interest to examine whether additional information could be obtained from analyzing syndecan-1 status in combination with another candidate biomarker. We selected fascin for this analysis, because syndecan-1 is a functionally significant regulator of the cytoskeletal organization of actin and fascin in several normal cell types [
28,
36]. Moreover, fascin is absent from normal colonic epithelium and its upregulation in colorectal adenocarcinomas correlates with poor prognosis [
26,
27]. In the normal stroma, fascin is detected at low levels in fibroblasts and at higher levels in dendritic cells and vascular endothelial cells. These studies also uncovered that stromal fascin is increased in at least 47% of tumor specimens, irrespective of the fascin status of the tumor [
26,
37]. From study of contiguous sections of the same Cleveland Clinic colorectal tissue microarray specimen set stained for syndecan-1, we identified that loss of tumor cell syndecan-1 did not correlate with the upregulation of fascin in clinically aggressive adenocarcinomas. However, loss of syndecan-1 immunoreactivity of the tumor did correlate significantly with strong stromal fascin staining. This novel finding brings further support to the idea that upregulation of stromal fascin may represent an aspect of the host-tumor interaction. The altered adhesive and motility properties of both tumor cells and adjacent stromal cells may jointly contribute to tumor progression. Future studies of independent datasets will be needed to validate the statistical as well as the clinical significance of the combined biomarker data.
We also identified that, in the tumors where stromal syndecan-1 was elevated, stromal syndecan-1 frequently overlapped with areas of increased stromal fascin (Fig.
1d). However, the fractions of positive specimens (16.6% for stromal syndecan-1 and 47% for stromal fascin) were very different. We infer that either the two molecules are under separate regulation, or that fascin is expressed by multiple cell types within the stroma. Abnormal expression of fascin by foci of stromal fibroblasts has also been observed in idiopathic pulmonary fibrosis [
38]. We speculate that for stromal cells in which both syndecan-1 and fascin are upregulated, syndecan-1 could provide pro-migratory cues through its intracellular regulation of fascin that promotes formation of lamellipodia [
28]. On the basis of the small total number of cases that were positive for stromal syndecan-1 in our dataset, the observation of co-staining for stromal syndecan-1 and fascin did not have statistical significance.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MS organized the stainings, scored sections and contributed to the study design and drafting of the manuscript. YH scored sections, carried out the statistical analysis, prepared figure panels and the Table and contributed to the drafting of the manuscript. JCA designed the study, participated in data analysis, prepared Fig.
1 and drafted the manuscript. Allauthors read and approved the final manuscript.