Background
In men with Western lifestyle the most prevalent cancer is prostate cancer [
1]. Although most cancers show an indolent course, the disease still represents the third most common cause of cancer related death in men. Therefore a specific and sensitive prediction of aggressive forms is warranted to improve decision-making [
2,
3]. At present Gleason grade and tumor extent on biopsies, preoperative prostate-specific antigen (PSA), and clinical stage are established pretreatment prognostic parameters. These parameters are statistically powerful but not sufficiently reliable for optimal individual outcome prediction. For example the Gleason grade suffers from substantial interobserver variation [
4]. Therefore the identification of new clinically applicable molecular markers may enable a more reliable prediction of prostate cancer aggressiveness in the future.
MAPT facilitates tubulin assembly and microtubule stabilization [
5]. MAPT is mainly expressed in neuronal axons and glial cell cytoplasm, but is also present in various non-neuronal cells including lymphocytes, epithelial and glandular cells [
5‐
8]. Aberrant expression of MAPT has been reported for many cancer types such as gastric, breast, and colorectal cancer [
9‐
12], and has been linked to adverse tumor features and poor prognosis in some of them [
12]. Little is known about the role of MAPT in prostate cancer. Only a few studies demonstrated MAPT expression in prostate cancer cell lines and in clinical samples but did not attempt to link MAPT expression to clinical features of the disease [
13‐
15]. However, MAPT might be of interest in prostate cancer since overexpression has been found to represent a prognostic marker in several cancers [
12,
16,
17]. MAPT overexpression has also been linked to resistance to taxane-based therapies in various other cancer types [
10‐
12,
18]. To date, taxanes are the most important cytotoxic agents for advanced and hormone-refractory prostate cancer [
19‐
21].
Here, we employed a large - more than 17,000 prostate cancers - and highly annotated tissue microarray (TMA) to elucidate the role of MAPT expression in this disease.
Discussion
The results of our study identify MAPT overexpression as a moderate prognostic feature occurring in a relatively small subset of prostate cancers.
In this study, detectable MAPT expression was seen in about 8% of prostate cancers whereas normal prostate tissues remained negative under the selected experimental conditions. Only one study has analyzed MAPT expression by IHC in prostate cancer before. Cirak et al. reported 23% MAPT positive cases in a series of 30 prostate cancers [
13]. It is well possible, that the large section approach of Cirak et al. lead to a higher detection rate of tumors with a heterogeneous MAPT expression. Our data indeed suggest that MAPT expression might be heterogeneous in a considerable fraction of tumors. Clear-cut heterogeneity was even found in some TMA spots (Fig.
1d) and a thorough analysis of all cancer-containing tissue blocks of three of our cancers with high MAPT expression on TMA spots always revealed both MAPT positive and MAPT negative cancer areas. Such heterogeneity represents a limitation for TMA studies analyzing only single spots per tumor.
MAPT overexpression was associated with to unfavorable tumor phenotype and early biochemical recurrence in this study (
p < 0.0001 each). The independent prognostic impact of MAPT overexpression from established prognostic parameters and the difference in the five-year recurrence rate of more than 20% between patients with and without detectable MAPT expression argues for a potential clinical relevance of this molecular feature. A similarly strong prognostic role has recently been described for aberrant βIII-tubulin (TUBB3) expression in prostate cancer. TUBB3 is a microtubule protein, which is normally expressed in cells of neuronal origin but not in prostate epithelium [
32]. Overall, the striking prognostic impact of the expression of proteins influencing structure and maintenance of microtubules suggest a considerable impact of composition and function of the cytoskeleton on the behavior of cancer cells.
The extensive molecular database attached to our TMA allowed us to further study the role of MAPT expression in prostate cancer and to search for possible interactions. About 50% of prostate cancers carry gene fusions linking the androgen-regulated
TMPRSS2 with the transcription factor
ERG [
26,
33]. As a result of this rearrangement,
ERG becomes androgen regulated and massively overexpressed. Our data demonstrate strikingly higher MAPT expression levels in ERG positive than in ERG negative cancers. This finding is consistent with data suggesting that ERG may have a regulatory role in microtubule dynamics [
17,
34] and that ERG can even destabilize microtubules by binding soluble tubulin in the cytoplasm [
35]. The exact molecular mechanism for this is unknown. According to the eukaryotic promoter database [
36] MAPT is not a direct target of the
ERG transcription factor. It is possible, however, that ERG has an indirect impact on MAPT transcription through at least one of its more than 1600 target genes [
37‐
39]. Our comparison of MAPT expression with frequent genomic deletions identified
PTEN as the only deletion linked to high MAPT expression. This fits well to earlier work in neurodegenerative diseases reporting that
PTEN can affect MAPT phosphorylation, aggregation or it’s binding to microtubules [
40,
41].
The existing data suggest a general role of MAPT protein in cancer. High rates of MAPT positivity have been reported from several other important cancer types including 43–52% in breast cancer [
16,
42,
43], 63–74% in ovarian cancer [
12,
44], and 55–70% in gastric cancer [
11,
45,
46]. The clinical and prognostic value of MAPT may greatly depend on the tumor type. For example, high MAPT protein expression level has been linked to good prognosis in breast cancer [
47], but to poor prognosis in ovarian cancer [
12]. It is unknown why MAPT exerts a different impact on tumor cell aggressiveness in different cancer types. As the microtubule composition varies between cell types, it may be speculated that MAPT induced modifications of the microtubule dynamics may have a diverse impact on cell behavior depending on the tissue of origin. It is also known that MAPT interacts with other cancer related proteins and pathways. For example, it has been shown that MAPT can cooperate with various growth related kinases such PI3K, Fyn, cSrc, and Fgr [
14,
48,
49]. Such kinases may have a different role in different cell types. Moreover, MAPT interactions depend on its phosphorylation status. Substantial differences in cell lines derived from prostate and brain cancers suggest that MAPT phosphorylation might strongly depend on the tumor type [
14,
50‐
52].
In several tumor types, MAPT has been suggested to represent a potential predictive marker in patients treated with taxanes [
12,
46,
53‐
57]. MAPT competes with taxanes for the same binding site at the microtubules. Although MAPT stabilizes microtubules in the same way as paclitaxel, its binding is more reversible [
18]. Consequently, overexpression of MAPT has been suggested to render microtubules insensitive to paclitaxel therapy [
18,
58,
59]. In prostate cancer, taxanes are the most important cytotoxic agents for advanced metastatic disease. However, response rates in clinical studies (measured as a 50% decline of PSA) are about 45–50% [
60]. It would be interesting to study the relationship between expression of proteins related to the microtubules system - such as MAPT and TUBB3 - and response to taxanes in prostate cancer in clinical trials.