In this study we provide evidence that SKIP acts as a co-regulator of AR transcription. Using mammalian one- and two-hybrid-experiments, we showed that SKIP augments AR AF-1 ligand-independent transcriptional activity and AR ligand-dependent AF-1 and AF-2 (AR N/ C-terminal) transcriptional interaction. Moreover, SKIP augmented AR-dependent transcription with two ARE-containing reporters in prostate cancer PC3 cells. Immunoprecipitation and FRET-experiments showed that AR interacts with SKIP and that the AR translocates to the nucleus within minutes after ligand stimulation to interact with SKIP.
FRET detects proximities of 3–7 nm for ECFP/EYFP-FRET pairs , which is a distance that can in most cases be regarded to report on direct interactions, considering that the size of the fluorescent proteins is already 2 nm × 4 nm . The observed high FRET efficiency between AR-ECFP/ EYFP-SKIP on the speckle patterned SKIP-positive structures is therefore in agreement with a direct interaction of these proteins.
Our FRET data are supported by our one- and two-hybrid data, which suggest that SKIP can activate both the N- and C-terminus of AR. Therefore, FRET-fluorophores on the C-terminus of AR and N-terminus of SKIP may come into very close proximity, despite the relatively large size of both proteins. The strong increase of FRET upon DHT-stimulated nuclear translocation of AR-ECFP to EYFP-SKIP positive structures that resemble nuclear speckles, can be explained by the fact that an excess of the acceptor, EYFP-SKIP, leads to high FRET-levels in a molecular complex, where the donor-acceptor-ratio is 1: >1 .
Nuclear speckles form a functional compartment within the nucleus that include splicing factors. Active gene transcription coupled with pre-mRNA splicing is thought to occur at the periphery of the splicing factor compartment [49–51]. It was therefore intriguing to see that the FRET between SKIP and AR increased from the outside to the inside of the speckle-like SKIP-positive structures. However, as fluorescently tagged proteins had to be overexpressed, care must be taken when relating the magnitude of the FRET with the affinity of the two proteins under physiological conditions. On the other hand, overexpression does not seem to induce random interaction in our experiments, as the interaction only significantly increased a) after DHT-stimulation and b) involving a translocation from the cytoplasm to the nucleus. We cannot say, whether AR is in addition targeted to other nuclear compartments.
It is of interest that the AR AF-1 domain has been shown to interact within nuclear speckles with another splicing factor ANT-1 (homologous to yeast splicing factor Prp6p) . This distribution resembles the AR-SKIP co-localization that we have observed (Figure 5). Thus AR interaction with SKIP and SKIP enhancement of AR AF-1 transcriptional activity is similar to AR/ ANT-1-dependent augmentation of transcription and splicing. Therefore, further studies will be required to determine if SKIP, like ANT-1 may recruit AR into a transcription-splicing coupled machinery.
Interestingly, the SKIP-EYFP signal disappeared completely ~16 minutes after DHT mediated translocation of AR to the nucleus, suggesting AR-mediated degradation or masking of SKIP. Loss of the SKIP-EYFP signal started soon after the maximum FRET was reached, raising the possibility that a regulated degradation is triggered at high AR concentrations. The AR has been previously linked to proteasome-mediated degradation, by its interaction with ubiquitin protease USP10 , the ubiquitin ligase ARNIP  and sensitivity of AR transactivation to proteasome inhibitors .
AR transcriptional activity is dependent on its N-terminal and C-terminal domains, which if artificially isolated can act in a ligand-independent (AF-1) and ligand-dependent (AF-2) manner, respectively [23, 34]. Immunoprecipitation and FRET-data allow us to explain the transactivation data by the direct interaction of SKIP with AR, however details of the exact mechanism are still unclear. We showed that the AR N-terminal domain was sufficient to engage SKIP to increase transactivation. It is possible that in a transcriptionally active complex, such as that induced by the N-terminal AR fragment, SKIP is engaged in the recruitment of general co-regulators such as N-CoR and p300 . In line with the latter interpretation, it was previously found that AR ligand-dependently interacts with the nuclear receptor co-repressor, N-CoR , which also interacts with SKIP .
Moreover, our two-hybrid data indicate that SKIP bi-functionally interacts with the N- and C-termini of AR, or at least facilitates this N/C- interaction. Interestingly, Saitoh et al.  observed something similar for the effect of the co-regulator CBP, which is also required for nuclear foci or speckle formation. In addition, SRC1 also modulates N/C-terminal interactions , however, this and its co-regulator activities are handled by distinct parts of the protein with an apparently higher significance for its LXXLL-motif independent interaction . Moreover, co-regulators were shown to modulate the N/C-interaction, in a cell type dependent manner. Correspondingly, their effect on the N/C-interaction did not always correlate with their effects on AR-mediated transactivation or cell growth, which may highlight the importance of specific microenvironments that influence this interaction . This may be explained by the assumption that the AF-1 may be the major transactivator under normal physiological conditions, while type I co-regulators come into play, if their concentration is high, such as observed in recurrent prostate cancer .