Mislocalized cytoplasmic p27 activates PAK1‐mediated metastasis and is a prognostic factor in osteosarcoma

The development of pulmonary metastasis is the leading cause of death in osteosarcoma (OS), which is the most common malignant bone tumor in children. We have previously reported that the tumor suppressor p27 (KIP1, CDKN1B) is frequently mislocalized to the cytoplasm of OS. However, its prognostic significance and metastatic mechanism are still elusive. Here, we show that cytoplasmic p27 significantly correlated with a higher metastatic status and poorer survival of OS patients (n = 136, P < 0.05), highlighting the clinical significance of p27 mislocalization in OS. Mechanistically, cytoplasmic p27 is co‐immunoprecipitated with p21‐activated kinase 1 (PAK1), which resulted in higher PAK1 phosphorylations, actin polymerization, and cell motility in p27‐mislocalized OS cells. Silencing PAK1 expression in different p27‐mislocalized OS cell lines decreased the migratory and adhesion abilities in vitro, as well as the development of pulmonary metastases in vivo. Similar PAK1‐dependent motility was also observed in other p27‐mislocalized cancer cell lines. In summary, our study suggests that cytoplasmic p27‐mediated PAK1 activation is crucial for OS metastasis. A biomarker‐guided targeted therapeutic approach for metastatic OS and other cancers harboring p27 mislocalization can be developed, where cytoplasmic p27 is used for risk stratification and PAK1 can be exploited as a potential therapeutic target.


Comments to the Author
In the manuscript entitled "p27 mislocalization is a poor prognostic factor of osteosarcoma by activating PAK1-mediated metastasis", Chen and colleagues analyse the correlation between cytoplasmic p27 and survival of osteosarcoma patients. By using MS, authors identify PAK1 as a novel interactor of p27. Interestingly, this interaction apparently leads to the regulation of cytoskeleton rearrangements which are of pivotal importance in cell adhesion and metastasis development. The manuscript is well written and the conclusions are strongly supported by the data presented. To note, authors both use a broad range of cell lines and perform in vivo experiments (IHC/TMA and xenograft model). The manuscript is almost suitable for publication in Molecular Oncology however I have some issues to address: 1. Please, could you demonstrate the representative images of IHC from your TMA to make it easier to the reader to understand the scale of IHC intensity you are using (i.e. for each level of intensity show one representative image).
>> As the reviewer suggested, representative TMA core images of low (4X) and high (20X) magnification powers of different IHC staining scores (Proportion scores 1-4) used in the survival analysis have been included in Supplementary Figure  S1A. Additionally, we have also included representative TMA images for nuclear, cytoplasmic, and negative staining of p27 (Supplementary Figure S1B).  Fig. 2F contains the RHO-GTPase results. We believe that the revised Fig. 2 is much easier to read as the order the sub-figures follows the order of description in the main text.
3. Scale bars for images and MW for WB must be added, as well as uncropped versions for all WB must be provided (as a Supplementary file or only for revision). >> We agree with the reviewer that quantification of the Phalloidin staining will better discriminate the differences of the actin stress fiber results in Fig. 3C. Thus, we quantified the Phalloidin staining images using the ImageJ software and performed a statistical analysis (t-tests) of the quantification results. The new analysis showed that the stress fiber amounts in osteosarcoma cell lines was significantly lowered by shRNA-mediated PAK1 gene silencing (p<0.05). Since a new PAK1-shRNA mutant was created as described in Comment #6, new Phalloidin staining figures are added. 5. Could you perform the IF staining of p27 in these cell lines ( Fig 3C)?
>> Since IF can only produce semi-quantitative results of the subcellular localization of p27 in osteosarcoma cell lines, we performed a more quantitative approach to analyze the p27 subcellular localization in the three osteosarcoma cell lines used in Fig. 3C to better address the reviewer's comment, i.e. subcellular fractionation followed by Western blotting. Nuclear and cytoplasmic protein controls were added. The same approach was also used to analyze the other cancer cell lines in Fig. 6A. The new results showed that p27 was predominantly localized in the cytoplasm of the three osteosarcoma cell lines. The new result is added in Supplementary Fig. S4.
6. Since it is broadly accepted that at least 2 si/shRNA must be used to avoid any possible off-target effect, it would be nice if you tried to use another shRNA for PAK1 in the migration assay / phalloidin staining. >> We thank you for the reviewer's suggestion to further improve our results. As suggested, we used another PAK1 shRNA (shRNA#2) to develop an additional PAK1-shRNA mutant in the three osteosarcoma cell lines. As indicated in the revised Fig. 3, the new shRNA efficiently reduced the PAK1 expression in NES-p27 and 143B cell lines, but not in U2OS, suggesting this new shRNA is not as commonly effective as the original shRNA (shRNA#1) used. We then examined the migration ability of the new PAK1-shRNA osteosarcoma mutants and found that the migration of the silenced mutant cells was significantly decreased in NES-p27 and 143B when compared with the scramble controls ( Fig. 3B-C), ruling out that the decrease of tumor cell migration is due to off-target effects of the original PAK1-shRNA used. The migration of the new shRNA mutant in U2OS was similar to that in the scramble control, which is consistent with the low knock-down efficiency in this cell line. In addition, Phalloidin staining on the new shRNA mutants showed a consistent result that the actin stress fiber formation is affected by PAK1 expression. This new information has been added to the Result section of the revised main text. In addition, as indicated in our PAK1 inhibitor analysis, the tumor cell migration was also decreased in the FRAX-597 treated cells when compared with DMSO-treated cells (Fig. 2G). This inhibitor result further support the shRNA result that tumor cell migration is dependent on PAK1 in osteosarcoma cells.