An interphase pool of KIF11 localizes at the basal bodies of primary cilia and a reduction in KIF11 expression alters cilia dynamics

KIF11 is a homotetrameric kinesin that peaks in protein expression during mitosis. It is a known mitotic regulator, and it is well-described that KIF11 is necessary for the formation and maintenance of the bipolar spindle. However, there has been a growing appreciation for non-mitotic roles for KIF11. KIF11 has been shown to function in such processes as axon growth and microtubule polymerization. We previously demonstrated that there is an interphase pool of KIF11 present in glioblastoma cancer stem cells that drives tumor cell invasion. Here, we identified a previously unknown association between KIF11 and primary cilia. We confirmed that KIF11 localized to the basal bodies of primary cilia in multiple cell types, including neoplastic and non-neoplastic cells. Further, we determined that KIF11 has a role in regulating cilia dynamics. Upon the reduction of KIF11 expression, the number of ciliated cells in asynchronously growing populations was significantly increased. We rescued this effect by the addition of exogenous KIF11. Lastly, we found that depleting KIF11 resulted in an increase in cilium length and an attenuation in the kinetics of cilia disassembly. These findings establish a previously unknown link between KIF11 and the dynamics of primary cilia and further support non-mitotic functions for this kinesin.


KIF11 localizes to the basal body of primary cilia in GBM cells. Our previous work with KIF11
demonstrated that the protein is overexpressed throughout the cell cycle in patient-derived GBM cells 4 . Additionally, it has been demonstrated that primary cilia can be found in patient-derived GBM samples, but the role of these cilia is somewhat disputed 21,22 . Because of the elevated levels of KIF11, and the presence, albeit not well understood, of primary cilia in GBM tumorspheres, we sought to understand if KIF11 was associated with these primary cilia. To answer this question, we grew patient-derived GBM (08-387) tumorspheres in suspension, then fixed the spheres and immunolabeled for KIF11 and the primary cilia markers Arl13b and Cep135 to mark both the axoneme and basal body/centrosome, respectively. Confocal imaging of cilia in tumorspheres revealed that KIF11 localized near the basal bodies of primary cilia (Fig. 1a). Because the role of primary cilia in GBM tumorspheres is not well understood, we also aimed to further characterize the ciliated cells in these samples. We first immunolabeled cells for Ki67 (a marker for proliferating cells), Arl13b, and pericentrin (a marker for pericentriolar material). Imaging of these tumorspheres demonstrated that primary cilia could be found on both Ki67 negative and positive cells (Fig. 1b). This finding is interesting, as primary cilia are most often found on quiescent cells, and hence would not stain positively for Ki67 23 . However, there are examples where cilia are present on proliferative cell populations, such as on neural progenitors in fetal and postnatal proliferative zones, as well as in cancer cells 21,22,24,25 . Our data agrees with previous reports that cilia can be present on proliferating cancer cells. We also labeled tumorspheres for Sox2 (a marker for cancer stem cells), Arl13b, and pericentrin to determine if cilia are present on cancer stem cells. We found that almost all ciliated cells were Sox2 positive, indicating a possible association between primary cilia and the cancer stem cell population (Fig. 1c). These data demonstrate that KIF11 is localized at the basal bodies of primary cilia in GBM tumorspheres. Furthermore, they support previous reports that primary cilia in GBM tumorspheres have been identified on Ki67 positive cells 21 . KIF11 localizes to the basal body of primary cilia in non-neoplastic cells and associates with ciliary proteins. As primary cilia in tumorspheres are not well understood and may be differently regulated than primary cilia in non-neoplastic cells, we next wanted to investigate whether KIF11 also localized to basal bodies of primary cilia on non-neoplastic cells. To address this question, we collected a serum starved population of immortalized retinal pigmented epithelial cells (hTERT-RPE1 or RPE1 cells), and immunolabeled for KIF11, Arl13b, and Cep135. We immunolabeled these antibodies simultaneously and each antibody individually to validate KIF11 localization distinctly at the basal body region of primary cilia. Serum starvation in RPE1 led to quiescent/ G0-synchronized cells, which were used to maximize the number of cells with primary cilia in our samples. Akin to what we identified in GBM cancer stem cells, we found that KIF11 was also localized near the basal body of primary cilia in RPE1 cells (Fig. 2a,b). We also used another cell model, immortalized human neural progenitor cells, to examine KIF11 and primary cilia. Again, using immunofluorescence, we determined that KIF11 localized near the basal body of these primary cilia as well (Fig. 2c,d). Lastly, we collected G0-synchronized RPE1 cells for proteomic analysis to determine if KIF11 interacted with any cilia-associated proteins. We immunoprecipitated KIF11 from our samples, and then performed proteomic analyses to identify any interacting partners. We identified over 30 cilia-associated proteins within the interactome for KIF11 in G0-synchronized RPE1 cells; 14 of these proteins were identified in replicate screens (Table 1, bold indicates proteins identified in replicate screens, non-bolded identified in one screen). These results demonstrate that KIF11 is localized to primary cilia in multiple cell types and may directly or indirectly interact with cilia-associated proteins, suggesting that KIF11 may play a role in the regulation of cilia dynamics in normal cells.

KIF11 depletion alters cilia dynamics and length.
Having identified that KIF11 localized at the basal body of primary cilia and interacts with ciliary proteins, we next sought to investigate if KIF11 had a role in regulating primary cilia. To answer this question, we used two siRNAs designed against KIF11 (siKIF11-1 and siKIF11-2), as well as a control siRNA (siControl) with depletion of KIF11 confirmed at the protein level by immunoblotting ( Supplementary Fig. 1a,b). We treated asynchronously growing RPE1 cells with either siControl, or one of the two siKIF11s. siRNAs to KIF11 were titrated to an amount that did not impact the ability  Fig. 1c). 72 h post siRNA treatment, cells were fixed and immunofluorescence was performed to examine primary cilia. KIF11 depletion led to a significant increase in the number of ciliated RPE1 cells over siControl-treated cells (Fig. 3a). Importantly, this shift towards increased ciliation was not merely due to all of the cells becoming quiescent, as evidenced by the incorporation of EdU into an equal number of cells as those treated with siControl ( Supplementary Fig. 1c). Since a decrease in KIF11 via siRNA caused an increase in the number of primary cilia, we also overexpressed GFP and GFP-KIF11 in asynchronous RPE1 cells to determine if this would also impact the number of primary cilia. Both 24 and 48 h after GFP or GFP-KIF11 transfection into RPE1 cells, we did not see any difference in the percent of ciliated cells between the two groups ( Supplementary Fig. 1d). We then co-transfected siKIF11-1 with either GFP alone, or an siRNAresistant, GFP-tagged KIF11 to determine if KIF11 was sufficient to rescue this change in the percentage of cili-  Tumorspheres from GBM 08-387 were fixed and probed with α-Arl13b (green), α-Cep135 (green), α-KIF11 (red), and the DNA was counter-stained with Hoechst (blue). (b) 08-387 tumorspheres were fixed and probed with α-Arl13b (green), α-Pericentrin (green), α-Ki67 (red), and the DNA was counter-stained with Hoechst (blue). (c) 08-387 tumorspheres were fixed and probed with α-Arl13b (green), α-Pericentrin (green), α-Sox2 (red), and the DNA was counter-stained with Hoechst (blue). White boxes in a-c represent regions of interest shown in the additional panels. www.nature.com/scientificreports/ ated cells. When GFP-KIF11 was transfected into the cells, the phenotype of increased cilia was partially rescued in positively-transfected cells (Fig. 3b). We also determined that depleting KIF11 increased the length of primary  www.nature.com/scientificreports/ cilia over siControl-treated cells (Fig. 3c). The length of primary cilia is tightly regulated, and changes in length could indicate changes in cilia-mediated signaling 39,40 . To further explore how reducing KIF11 impacted primary cilia dynamics, we studied the impact on serum-induced primary cilia disassembly. When RPE1 cells are serum-starved, approximately 80% of the cells will become quiescent and form cilia. In this experiment, RPE1 cells were serum-starved for 48 h, and then serum was added back to the cells to induce cilium disassembly and cell cycle reentry. Cells were fixed for immunofluorescence at the time of serum addition (0 h), as well as 9 h later, when cilia in more than 50% of the cells should have been disassembling. At 9 h post-serum addition, there were significantly more ciliated cells remaining when cells were treated with siKIF11 ( Fig. 3d). Together, these data demonstrate, for the first time, that a reduction in KIF11 is enough to alter primary cilia dynamics and presence on a cell, as well as primary cilia length. Furthermore, we have shown that this reduction in KIF11 is permissible in the context of mitosis, but there is a separate, primary cilium-specific effect that cannot function normally with the decrease in protein.
Heterozygous KIF11 alters cilia dynamics and length. Because MCLMR and FEVR patients are reported to have a mutation in only one allele of KIF11, we next created two heterozygous CRISPR/Cas9 clonal lines of RPE1 cells to experimentally represent this genetic state (crKIF11-1 and crKIF11-2). Gene editing via CRISPR/Cas9 more closely reflects the patient mutational state than RNA interference via siRNA, and hence is critical for evaluating the impact of reduced KIF11 on primary cilia [12][13][14]16,41 . Depletion of KIF11 was confirmed by immunoblotting ( Supplementary Fig. 2a-c). Asynchronously growing wild type RPE1 cells, as well as asynchronous crKIF11-1 and crKIF11-2 cells were collected for immunofluorescence. Similar to our siRNA data, crKIF11-1 and crKIF11-2 cells were significantly more ciliated than wild type RPE1 cells (Fig. 4a). We also confirmed by EdU labeling that crKIF11 cells were still cycling and had similar numbers of EdU positive cells as wild type RPE1 (Supplemental Fig. 2d). Furthermore, we monitored cell growth via the Incucyte ZOOM live-cell imaging system, and again found similar growth rates between wild type RPE1 and crKIF11 cells (Supplemental Fig. 2e). Lastly, we found that cilium length was significantly increased in crKIF11 cells over wild type RPE1 www.nature.com/scientificreports/ (Fig. 4b). Again, these results suggest that reducing KIF11 can alter cilia dynamics, without disrupting its ability to go through mitosis, indicating that there is a primary cilia-specific role for KIF11.

Discussion
Our results are the first to demonstrate that KIF11 is localized at the basal body of primary cilia in multiple cell types, including neoplastic and non-neoplastic cells. Additionally, we have shown a previously unknown ability for KIF11 to regulate primary cilia dynamics. Typically, primary cilia are present when cells are quiescent and are only found on a small percentage of asynchronously growing cells, which are likely in early G1 phase 23 . Our results indicate that by depleting KIF11, we are able to shift this percentage of ciliated cells, implicating KIF11 as a regulator of cilia dynamics. KIF11's role as a primary cilia-regulator is further supported by our finding that transfection of KIF11 partially rescues the increase in ciliation in asynchronous cells. Further, in heterozygous KIF11 RPE1 cells, we see the same increase in ciliation. An increase in the amount of ciliated cells is important, as primary cilia regulate the careful balance in signaling patterns required for homeostasis [42][43][44][45] . Primary cilia are covered in receptors for different pathways, and oftentimes, these receptors are not expressed anywhere else on the cell [46][47][48] . Therefore, careful timing of when primary cilia (and therefore, their receptors) are present is critical for proper pathway activity. For example, during neurodevelopment the expansion of the neural progenitor pool relies upon cilia-mediated sonic hedgehog signaling 45 . If there are too many cilia, signaling could be disrupted or aberrant, and conversely, if cells are not ciliated, signaling will not occur. This leads us to consider a model where KIF11 is necessary for proper expression of primary cilia, and may suppress ciliation in inappropriate contexts. Furthermore, if KIF11 is depleted or mutated in a biological system, we hypothesize that there could be an increase in cilia at a time when cells should not assemble cilia. Ultimately, such a loss of suppression of ciliation could lead to aberrant signaling (Fig. 5a). In addition to the number of ciliated cells impacting signaling, the length of the cilium is also important when discussing relative signaling activity. Change in cilium length has been shown to alter signaling as receptor density changes when cilia are shorter and therefore have less surface area to be covered by receptors 39 . Therefore, we hypothesize that the increase in cilium length brought on by changes in KIF11 expression could potentially alter signaling. Similar to shortened cilia having altered signaling abilities, lengthened cilia could impact receptor density or receptor functionality, which may result in abnormal signaling (Fig. 5b) 49,50 . Further, we propose that these impacts on signaling patterns might explain some of the phenotypes seen in patients with MCLMR and FEVR. Future studies could ascertain the role of KIF11 in primary cilia-mediated signaling by examining downstream gene expression in pathways such as sonic hedgehog and Wnt. Our study is the first to establish a role between KIF11 and regulation of primary cilia, but further exploration will be required to understand the mechanism of how KIF11 impacts primary cilia dynamics. It is worth noting that multiple other mitotic kinesins have been shown to regulate primary cilia, independent of their role in mitosis. KIF14, a kinesin involved in bipolar spindle formation, has recently been shown to also play a role in basal body and primary cilia formation. Upon KIF14 depletion, cilium elongation and assembly is altered, as well as hedgehog signaling, through an association with Aurora Kinase A 51 . KIF2, KIF3, and KIF13 have all been shown to function in mitosis, as well as to have critical primary cilia regulatory roles 52,53 . Additionally, Mps1, a kinase essential for mitosis, has been shown to coordinate with VDAC3 to negatively regulate the inappropriate formation of primary cilia through accumulation at the centrosomes 54 . Perhaps KIF11 is acting in a similar way to one or more of these proteins, and is coordinating with other primary cilia proteins. The cell cycle and primary cilia formation are inextricably linked so it stands to reason, and has been experimentally demonstrated, that many of the proteins involved in the cell cycle might also be critical for the proper formation and regulation of primary cilia. Our data are the first step in exploring KIF11 in this context, and future studies should elucidate the mechanism for this previously unknown role for KIF11. Depleting KIF11 leads to alterations in cilia dynamics and length. (a) Asynchronous RPE1 cells were treated with either siControl, siKIF11-1, or siKIF11-2. After 72 h, cells were fixed and probed with α-Arl13b (green), α-Cep135 (green), α-KIF11 (red), and the DNA was counter-stained with Hoechst (blue). Percentage of ciliated cells was calculated for each condition and graphed (n = 3 biological replicates with 150-250 cells counted per n). Data analyzed by Student's t-test. Error bars represent standard deviation. Representative images for each condition are shown in the image panels. (b) Asynchronous RPE1 cells were treated with siKIF11-1, and after 24 h were then subsequently transfected with GFP, or GFP-KIF11. 48 h later cells were fixed and probed with α-Arl13b (red), α-Cep135 (red), α-GFP (green), and the DNA was counter-stained with Hoechst (blue). Percentage of ciliated, GFP-positive cells was calculated from the total number of GFP positive cells for each condition, and results were graphed (n = 3 biological replicates with 50-100 cells counted per n). Data analyzed by Student's t-test. Error bars represent standard deviation. Representative images for each condition are shown in the image panels. (c) Cilium length was measured for each condition and graphed (n = 3 biological replicates with 20-30 cells counted per n). Data analyzed using a Student's t-test. Error bars represent standard deviation. Median and quartiles denoted by dashed lines within violin plots. Representative images for each condition are shown in the image panels. (d) To evaluate disassembly of primary cilia, asynchronous RPE1 cells were treated with either siControl, siKIF11-1, or siKIF11-2. After 24 h, cells were serum starved. 24 h later cells from each condition were collected, and serum was added back to the remaining cells. Cells were collected 9 h later. All cells were fixed and probed with α-Arl13b, α-Cep135, α-KIF11, and the DNA was counter-stained with Hoechst. Percentage of ciliated cells was calculated for each condition and graphed (n = 3 biological replicates with 125-250 cells counted per n). Data analyzed by 2-way ANOVA with Sidak's post-test. Error bars represent standard deviation.    www.nature.com/scientificreports/ progenitors were obtained commercially (SigmaMillipore) and were grown in ReNcell neural stem cell maintenance media (SigmaMillipore) with basic fibroblast growth factor (10 ng/mL; Gibco) and EGF (10 ng/mL; Gibco). All cells were cultured at 37 °C at 5% CO 2 . For cell counting before each experiment, a single-cell suspension was achieved using TrypLE (Gibco). Mycoplasma testing was done quarterly (LookOut Mycoplasma PCR Detection Kit; Sigma-Aldrich) and cell line verification was done annually (microsatellite genotyping; OSUCCC Genomics Shared Resource).
Briefly, 25 µL of beads were incubated with 3 µg of KIF11 antibody (Novus; NB500-181; rabbit) for 30 min at room temperature followed by washes and then an overnight incubation at 4 °C with 25 mg of G0 synchronized RPE1 whole cell extract. For analysis of proteins off beads, immunoprecipitated proteins were released from the beads by boiling in Laemmli 2X concentrate sample buffer (Sigma).

Samples for LC-MS/MS analysis.
For "on-bead" analysis, KIF11 from G0 synchronized RPE1 was immunoprecipitated as described above with resulting immmunoprecipitated protein taken on beads for processing by the Proteomics Shared Resource at The Ohio State University Comprehensive Cancer Center for protein quantification and identification of protein interactors. For KIF11 that was boiled off beads, samples were resolved on 4-20% Mini-PROTEAN TGX precast gels (Bio-Rad Laboratories, Inc.) and visualized with GelCode Blue Stain Reagent as per manufacturer instructions (ThermoFisher Scientific). The KIF11 band was then cut from the gel and processed by the Lerner Research Institute Mass Spectrometry Laboratory for Protein Sequencing for proteomic analysis and identification of interacting proteins.
Cell growth. Cell growth was monitored after depletion of KIF11 by EdU labeling and by Incucyte ZOOM live-cell imaging. EdU labeling was accomplished by exposing cells to a 4-h pulse of EdU prior to harvesting the cells at the end of the experiment. Cells that were exposed to EdU were then fixed in 100% methanol, permeabilized in PBS-Triton X-100 (0.2% vol/vol), and labeled with the reagents and protocol associated with the Click-iT EdU Cell Proliferation Kit for Imaging (Thermo Fisher). Nuclei were counterstained with Hoechst. Coverslips were then mounted onto slides using Fluoromount-G (SouthernBiotech). Images were taken using a Zeiss LSM 800 confocal microscope. To track cells via Incucyte live-cell imaging, 50,000 WT RPE1 or crKIF11 cells were plated in wells of a 6 well plate and placed on the Incucyte. Cell confluence was used as a measure of growth, and was monitored over 40 h. Multiple areas of each well were monitored.

Statistical analyses.
Statistical significance was calculated with GraphPad Prism Software using a Student's t-test or 2-way ANOVA with Sidak's post-test (GraphPad Software, Inc.).