Steroid Receptor RNA Activator Protein (SRAP): a potential new prognostic marker for estrogen receptor-positive/node-negative/younger breast cancer patients

Breast cancer remains the second leading cause of cancer-related deaths in women worldwide and is one of the most frequently diagnosed cancers with an estimated 1,000,000 new cases detected each year worldwide [1]. Following diagnosis, several critical prognostic and predictive markers are assessed in order to determine, for each patient, the most appropriate treatment to be administered. Estrogen receptor (ER) status, progesterone receptor (PR) status, nodal status, tumor size, and grade of malignancy are the classical parameters used to date by clinicians to narrow down prognoses and weight treatment options [2]. More recently, human epidermal growth receptor (HER)-2, which is over-expressed in about 25% of breast cancers and is associated with a more aggressive disease and a poorer outcome, has also been used as a prognostic and predictive marker [3]. Recent approaches such as gene profiling and tissue micro-arrays (TMAs) have increased our ability not only to identify new potential markers, but also to rapidly test their potential validity [4, 5]. The more such molecules are identified, the higher become the odds of finding the optimal combination of markers allowing the determination of an 'ideal' treatment for any given patient [6].

The steroid receptor RNA activator (SRA) was originally identified as a functional non-coding RNA increasing the transcriptional activity of ligand-bound steroid receptors [7]. It is currently believed that this action is mediated by the formation, at the promoter of target genes, of regulatory complexes containing steroid receptors, SRA, and both positive and negative protein regulators [8‐10]. SRA RNA is over-expressed during breast, ovarian and uterine tumorigenesis and tumor progression [11‐14]. It has therefore been suggested that by increasing the activity of the ER, SRA could participate in the mechanisms underlying these events [7, 12].

It has now been confirmed that coding SRA transcripts co-exist in breast cancer cells, with the previously described non-coding transcripts [15‐17]. The corresponding endogenous protein, steroid receptor RNA activator protein (SRAP), has been detected by western blot in multiple cell lines as well as in muscle and breast tissue [15, 17‐19]. It has been suggested that, as its RNA counter-part, the protein might also regulate the activity of estrogen and androgen receptors [10, 17‐19]. This hypothesis is further corroborated by the identification of the RNA helicase p68, a well-characterized regulator of ER activity [20], in nuclear complexes co-immunoprecipitating with endogenous SRAP [21].

Overall, accumulated data raise the intriguing possibility that SRAP levels could be associated with ER activity and/or expression, and could also potentially reflect on the response of breast cancer patients to endocrine therapy. It has recently been reported that the relative proportion of coding and non-coding SRA transcripts varies from one breast tumor to another and might characterize particular tumor subgroups [22]. Altogether, this suggests that SRAP expression could also differ between cases and potentially be a prognostic and/or predictive indicator in breast cancer. To address this issue, we have herein investigated the use of TMAs for the expression of SRAP in 372 cases with a wide range of clinical parameters.

In the present study, we have established that several SRAP-like peptides were detectable in breast tumor tissues. Interestingly, in ER positive, and more predominantly within node negative and younger patients, higher levels of the corresponding signal were associated with poor survival.

No data reporting specific detection of SRAP employing the antibody used in our experiments (cat # A300-743A) has been published. It was therefore critical to first validate the use of this antibody for western blot and immunohistochemistry analyses. A doublet migrating at the apparent size of about 30 kDa is detected by western blot in MCF-7 as well as in multiple other cell lines (Figure 1 and data not shown). This size corresponds to the size previously observed using different SRAP antibodies [15, 17, 19]. The signal decreases upon peptide neutralization and during SRA RNAi experiments. An additional band, also recognized by anti-V5 antibody (data not shown), appears when cells are transfected with a construct encoding an exogenous V5-tagged SRAP. Altogether, these observations strongly suggest that 743A antibody specifically recognizes SRAP.

The endogenous SRAP signal is detected mainly in the cytoplasm of cancer cells in vitro. In some cases, a speckled staining can also be seen in the nucleus. This is in agreement with both localizations previously observed for exogenous tagged SRAP [16]. This also corroborates data from Jung and colleagues [21], who co-immunoprecipitated endogenous SRAP from both nuclear and cytoplasmic protein extracts from Hela cells [10, 21].

Western blot analysis of breast tumors revealed the presence in some tumors of two additional bands migrating at an apparent size of about 25 kDa and about 40 kDa. Both signals disappear when the antibody is pre-incubated with the corresponding blocking peptide. This suggests that as well as 30 kDa SRAP, these proteins are also specifically recognized by 743A antibody. We have already reported the detection, within some breast tumors, of both 30 kDa and 25 kDa migrating bands [15]. The antibody used was, in contrast to 743A, targeting the N-terminal extremity of SRAP, because it was raised against SRAP amino acids 20 to 34. No bands migrating at the apparent size of 40 kDa had been previously observed. It should however be noted that when blots are overexposed, the 40 kDa band is detectable in MCF-7 cells but not in Hela cells (Figure 3), suggesting potential cell/tissue specificity in the mechanisms responsible for the generation of the corresponding peptide. This raises the possibility that this particular 'higher mass' or 'larger' peptide might be missing SRAP N-terminal extremity but contain additional unknown sequences. The identity of all these SRAP-like peptides remains to be determined. It should however be stressed that several SRA splice variants have now been characterized [10, 13, 17]. Proteins potentially encoded by some of these alternatively spliced transcripts are missing SRAP N-terminal extremity but still contain SRAP amino acids 180 to 236 recognized by 743A antibody. For example, Ensembl transcript [Ensembl: ENSESTT00000035630] containing intron-1, is expected to encode for a peptide [Ensembl: ENSESTP00000035630] corresponding to the last 162 amino acids of SRAP and with a predicted molecular mass of 18 kDa. As well, other alternative splicing events, such as intron-1 retention conjugated with exon-3 deletion [17], predicts a 220 amino acid residue protein (molecular weight 24.68 kDa) identical to SRAP in its N- and C-terminal extremity but differing in the central sequence.

Further studies are urgently needed to establish the nature of the different splicing events potentially involved in the generation of the different SRA transcripts and their potential protein products. This is critical if we consider that SRA was originally described as a functional RNA whose activity depended on the integrity of the core sequence determined by exon-2 to exon-5 [29]. It has recently been shown that the balance between fully spliced SRA and transcripts still containing intron-1 varied between breast tumors and characterized particular tumor subgroups [30]. Interestingly, alteration of this balance led to a change in breast cancer cell growth [30]. Taken together with size differences between SRA transcripts detectable by Northern blot in breast cancer cells and cells from other origins [7], this suggests that alternative splicing events, leading to different SRA RNAs and potentially to different SRAP-like proteins, might participate in the development of different breast cancer cell phenotypes.

Alternative post-translational modifications might also be responsible for the apparent differences in SRAP molecular mass. For example, a lysine residue at position 21 of the SRAP human sequence is conserved in all vertebrates and is predicted to be sumoylated (data not shown). Sumoylation of this residue could potentially participate to the shift in migration from 30 kDa to 40 kDa [31, 32]. Further studies are warranted to address this possibility.

Although only six normal breast cases were analyzed, they all expressed only 30 kDa SRAP at a level lower than that seen in most tumors. This is in agreement with the previous observation that SRA RNA is overall expressed at lower levels in normal compared with matched adjacent tumor tissue [14]. This also raises the possibility that mechanisms responsible for the generation of the 25 kDa and 40 kDa SRAP-like peptides are less frequent in normal than in tumor cells.

Using 743A antibody, significant higher levels of SRAP are found in ER-positive tumors, in PR-positive tumors, and in lower grade tumors. This fits with the pattern of expression of a known activator of steroid receptor, the steroid receptor co-activator 1 (SRC1). Indeed, high SRC1 protein levels have recently been shown to positively correlate with ER, PR levels and low grade in a large cohort of 426 breast cancer patients [33]. In their study, Green and colleagues reported that patients whose tumors expressed higher SRC1 levels had an overall longer survival and a lower recurrence rate [33]. In contrast, we found that high SRAP expression is a strong indicator of poor survival in ER-positive patients. It should be noted that a similar finding was obtained when using a training set, validation set method and X-tile software [34] (data not shown).

All 271 women whose tumors were ER positive had been treated with tamoxifen following surgery. Further analyses are needed to establish whether SRAP should be considered as a prognostic factor, a predictive factor, or both.

Interestingly, the potential indicator power of SRAP appears limited to younger patients whose tumors are ER positive and node negative. It should be stressed that the age cut-point of 64 years (corresponding to median age at time of surgery) has been chosen to divide the cohorts in subgroups with comparable number of cases. A cut-point of 51 years, average age for spontaneous menopause [35, 36], would have been more informative to establish the potential link between menopausal status and SRAP prognostic value. The number of cases younger than 51 was, however, too low to perform meaningful statistical analyses. For example, only 21 out of the 271 ER-positive patients were younger than 51 years. Menopause age follows a Gaussian distribution ranging roughly from 40 to 60 years [35, 36]. As patients' menopausal status is not registered in the Manitoba Breast Tumor Database, women younger than 51 years could also potentially be menopaused. It is therefore not possible to extrapolate whether SRAP prognostic value is directly linked to menopausal status. Further studies, performed on a cohort consisting of comparable numbers of known pre- and post-menopausal women should be initiated to clarify this issue.

The particular group of younger patients identified in this cohort (i.e ER positive, node negative) usually benefits from better prognosis and is considered as having an increased likelihood of responding to tamoxifen therapy. Our data, however, suggest that among this group, women with higher levels of SRAP might need additional treatment compared with women with lower SRAP levels.

We have previously reported that the detection in breast tumor extracts of 30 kDa SRAP by western blot, but not 25 kDa SRAP, might correspond to a better outcome for ER-positive/node-negative patients [17]. There is therefore an apparent disagreement with the present TMA results that show that higher SRAP is an indicator of shorter survival in ER-positive/node-negative patients. This apparent divergence could be due a variety of reasons. It may potentially result from differences in the cohorts studied. Our previous cohort was indeed smaller (n = 74 versus n = 130), the median ER levels was higher (45.5 fmol/mg versus 37 fmol/mg), the number of events was different (n = 7 deaths versus n = 26) and patients' ages were not available. Further, in the previous western blot analysis, which was performed as outlined earlier with an antibody recognizing the N-terminal region of SRAP, 30 kDa SRAP was detected in only 24 of 74 cases [17]. Using 743A, we have in the present study detected SRAP signal in all but 2 out of 130 ER-positive/node-negative patients. It should also be stressed that, as mentioned above, SRAP staining assessed in the present TMAs corresponds to multiple SRAP-like peptides (25 kDa, 30 kDa and 40 kDa SRAP) whereas only the 30 kDa SRAP was considered in the first study. This also raises the possibility that different peptides might have different prognostic or predictive values.

Altogether, we have found that several SRAP-like peptides are expressed in breast tumors and that their detection by immunohistochemistry could be used as a new prognostic/predictive marker in younger patients with ER-positive/node-negative breast cancer. Additional studies, performed with other antibodies are warranted to confirm this observation and further establish whether specific SRAP-like peptides have additional predictor values.