According to 2020 Globocan data there were 207,252 deaths due to ovarian cancer, which accounts for 4.7% of all deaths due to cancer in women in 20201. The first line treatment of ovarian cancer includes maximal surgical resection of the tumour, followed by neoadjuvant chemotherapy with platinum/taxane drugs. Despite initial complete response to chemotherapy in 60-80% of patients, eventually 80-85% of patients develop chemoresistance. Therefore, there is a constant search for new therapeutic targets in ovarian cancer2. Although several drugs targeting PARPs, angiogenesis and folate receptors have shown promising results, none of them were able to cure ovarian cancer and chemotherapy remains the first line of treatment for ovarian cancer patients3,4. Many of the chemotherapeutic agents exploit the defects in cell cycle machinery of cancer cells and inhibit the cell cycle by blocking mitosis. The majority of the available mitosis targeting chemotherapeutic agents act on tubulin and do not distinguish between healthy and malignant cells. This leads to the development of severe adverse effects including neurotoxicities and myelosuppression5,6. Therefore, the challenge is to identify molecular targets that are preferably required for the mitosis of the cancer cells and not the healthy cells.
Targeting cancer cell mitosis has been effective at treating various forms of cancer. However, most of the targets that have been identified and assessed in the clinic focussed mainly on cell cycle kinase inhibitors. There are many non-kinase proteins that might have deleterious effects on the cancer cell cycle but have remained unexplored. One such family of genes includes microtubule associated proteins (MAPs). MAPs help tubulin molecules to maintain stability as well as dynamics, nucleation, cross-linking, transport and orientation25. Inhibition of some MAPs such as Tau, MAP4 and MAP2 is known to sensitize cancer cells to microtubule-binding chemotherapeutic drugs7-9. In addition, microtubule-associated motor proteins, such as Eg5 have shown potential anti-tumor effects10,11. However there are 200 MAPs and their effect on tumor development remains largely unexplored.
One such MAP that plays an important role during mitosis is CKAP5. In an RNAi lethality test potential targets were screened for multiple myeloma cells compared to non-myeloma cells and CKAP5 along with 3 other gene targets were observed as the most differentially vulnerable targets12. CKAP5 regulates the overall microtubule dynamics in human cells by its microtubule-stabilizing and polymerizing activities. Its absence results in defective mitosis by multipolar spindle formation, reduction in the chromosomal oscillations, reduced tension between kinetochores and decrease in the spindle microtubule length13-20. Therefore, CKAP5 can be a promising cancer cell mitosis target due to the vulnerability of cells with high genetic aberrations and cell cycle abnormalities24. To this end, lipid nanoparticle (LNP) mediated siRNA delivery can be harnessed to silence a specific gene as it is the most advanced non-viral strategy for in-vivo nucleic acid delivery. In addition, application of LNP mediated mRNA delivery for corona vaccine by Pfizer / BioNtech and Moderna have paved the way to LNP mediated therapeutics in the clinic. LNPs are multicomponent lipid nanoparticles that consist of an ionizable lipid, helper phospholipid, cholesterol and PEG-lipid and they enable efficient nucleic acid encapsulation, stabilization and retention in the circulation as well as target cells and improve cell penetration.
Herein, we screened the effect of CKAP5 silencing in various solid cancer cell lines. We demonstrate that cells with high genetic instability are selectively susceptible to CKAP5 depletion and among them, NAR cells, a chemo-resistant ovarian cancer cell line, are the most sensitive to this cellular manipulation. In live-cell imaging we show that any cell that entered mitosis in the CKAP5 siRNA treated group was arrested in metaphase and underwent cell death. Further investigation into the mechanism revealed that the tubulin dynamics are halted during mitosis of CKAP5 knockdown. In an in vivo xenograft intraperitoneal ovarian cancer model, we observed that treatment of the tumor xenografted mice with siCKAP5 LNPs showed significant reduction in the tumor volume as compared to the control groups and increased survival in the treated group. Overall, our data suggests CKAP5 as a promising therapeutic target in ovarian cancer with high genetic instability.
Genetically unstable cancer cells are highly sensitive to siRNA mediated CKAP5 down-regulation
CKAP5 expression across various cancer cell lines is not very well documented, therefore we determined the expression of CKAP5 at the transcript level in a panel of solid cancer cell lines including ovarian, breast, colorectal, lung, liver and head and neck cell lines. All the cell lines showed CKAP5 expression; among which MDA MB231 showed the highest and HCT15 showed the lowest CKAP5 expression in our panel (Supp. Figure1). Since CKAP5 expression was observed in all the cell lines tested, we included the entire panel to test the effect of CKAP5 silencing on cell viability. To this end, we applied siRNA encapsulated LNPs. Ionizable lipid is the principal component of the LNPs and in the present study we used our previously described Lipid 10 as the ionizable amino lipid in a defined formulation21. LNPs were formed by microfluidic mixing, as previously described21. This produced uniformly sized LNPs with an average size distribution of 70nm and partially negative surface charge (Figure 1A). The size distribution and LNPs uniformity were further supported by cryo-EM data (Figure 1B). Confocal analysis of particle internalization in a few representative cell lines demonstrates particle uptake 2 hours post incubation, which was further increased after 4 hours (Figure 1C). After the initial particle characterization, the dose response effect of CKAP5 siRNA on cell viability was tested by XTT assay. Effect on cell viability was compared with the control siRNA particles at the highest treatment concentration. Cell viability was determined at 3 and 6 days post transfection of the cells with LNPs in the concentration range of 0.015-0.25 µg siRNA/ml. Control siRNA particles did not show toxic effects in this range, except for the Detroit cell line (Supp Figure 2A). Therefore, further experiments were performed with 0.12µg/ml of siRNA specifically for this cell line. CKAP5 silencing was confirmed on the transcript and protein level by qPCR and western blot analysis, respectively (Figure 2C & D). CKAP5 silencing resulted in reduction of cell viability 72 hours post treatment. However, the effects were more prominent 6 days post treatment in the majority of the cell lines screened for all the concentrations (Figure 2A & B). Best effect was observed at the highest concentration, which is 0.25µg/ml of siRNA. Any cell line that had below 50% cell viability post CKAP5 knock down was considered sensitive to the therapeutic effect. Out of the 20 cell lines tested, only 7 cell lines did not show any effect on cell viability (Figure 2A & B). Increasing the siRNA concentration up to the limit of non-toxic doses did not show increased sensitivity to treatment in almost all cells tested, even though efficient silencing was established (Figure 2F & G). The response to siCKAP5 treatment was neither related to its gene expression in these cell lines nor related to the cell doubling time. Since cancer cells are known to harbour genetic instability and cells with such defects can be more sensitive to further mitotic damage, we planned to analyse the genetic abnormalities in our 20 cell line panel. To this end, we determined the micronuclei formation and mitotic damage by DAPI and tubulin staining followed by confocal microscopy and observed that there was significant increase in micronuclei formation and mitotic spindle defects in vulnerable cell lines (p<0.05), suggesting selective sensitivity of genetically unstable cells to CKAP5 depletion (Supplementary Figure 3A & B).
Among all the cell lines tested for cell viability, A2780, NAR, MM468, BT549, HCT116, SK-Hep-1 and HepG2 were most sensitive to CKAP5 silencing. Interestingly, all the ovarian cancer cells were sensitive to CKAP5 down regulation. Among them, the chemoresistant NAR (NCI-ADR/Res) cell line, was most sensitive to CKAP5 silencing. Therefore, we used the NAR cell line as a model of interest for our further studies.
Overall, our data suggest that CKAP5 is expressed in all the cell lines included in the present study. CKAP5 was successfully downregulated by siRNA-LNPs in all the cell lines. Yet, only cells with high genetic instability lost viability in response to CKAP5 depletion. Since a highly chemoresistant ovarian cancer cell line, NAR showed marked sensitivity to CKAP5 depletion, we used it as our model cells for the additional studies.
CKAP5 downregulation shows cell cycle arrest and spindle defects in NAR cells
To study the underlying mechanism of CKAP5 downregulation mediated cell death, an apoptotic assay was performed by PI-Annexin V staining 3 and 6 days post CKAP5 silencing in NAR cells. Transfecting cells with 0.25µg/ml siCKAP5 led to 16% and 60% apoptotic cell death, at 3 and 6 days post transfection, respectively (Figure 3A & B). Further, CKAP5 silencing arrested 40% of the cells in the G2-M phase for 36 hours post CKAP5 downregulation. This arrest was not reversed at any given time point up to 96 hours, when most of the cells already entered apoptosis (Figure 3C & D). The effects of CKAP5 downregulation on spindle assembly abnormalities has been mainly studied in HeLa cells previously by others13,16. We investigated if these effects will be similar in NAR cells. To investigate the spindle damage in response to CKAP5 depletion, cells were stained with tubulin antibody 48 hours post CKAP5 silencing. We observed a significant increase in cells arrested in metaphase with multicentric spindle formation (80%) as compared to siControl (19.9%) and untreated cells (18.8%) (Figure 3E & F). Approximately 20% of control cells also showed multicentric spindle, suggesting innate defects in these cells. The detailed analysis of the spindles in CKAP5 silenced cells showed reduced spindle axis (2 folds, p<0.005), reduced spindle density towards the chromosomes and loss of proper metaphase plate formation (Figure 3G). The decrease in the density and axis length of spindle could be due to reduced microtubule polymerization and shorter centrosomal microtubules in the absence of CKAP513-16. Any mitotic damage results in activation and upregulation of damage specific spindle assembly checkpoint genes. Identification of upregulated genes can provide mechanistic insight into the kind of spindle damage. Therefore, we tested the expression of several spindle checkpoint genes in NAR cells 48 and 72 hours post CKAP5 silencing. We observed significant upregulation in BUB1, BUB1B and TTK genes 48 hours post CKAP5 silencing compared to control treated cells, which was normalized at 72 hours post silencing whereas there was no effect on the gene expression of AURKB, MAD2L1 and MAD2L2 (Figure 3H). BUB1, BUB1B and TTK genes are known to function in response to defects in kinetochore-microtubule attachments. Upregulation of these genes in response to CKAP5 downregulation suggests possibility of a similar disruption mechanism in NAR cells14,17. These defects in the cell cycle and spindle assembly formation, resulted in increased gamma H2A.X foci formation in the CKAP5 downregulated cells, which is a gold standard effect of DNA damage (Supp. Figure 4).
Overall, we observed that CKAP5 depletion leads to multicentric spindle formation, G2-M cell cycle arrest followed by apoptosis in NAR cells, which is accompanied by up-regulation of spindle check point genes and gamma H2A.X foci formation.
CKAP5 silenced cells show unique cell death mechanisms
To understand the fate of cells with multicentric spindles, we engineered NAR cells for stable expression of Tubulin.GFP and H2B.mCherry and tracked the spindle formation by live cell imaging using spinning disk microscopy post siCKAP5 or a siControl treatment. We started the live cell imaging 12 hours post treatment by capturing images every 15 minutes until the 60 hour timepoint (schematic shown in Figure 4A). Images from the CKAP5 silenced group show that in the initial few hours of imaging, all the metaphases observed were completed with bipolar spindle; however, after 12 hours of imaging, at 24 hours post CKAP5 silencing, the majority of the metaphase observed was multicentric (Figure 4D, Supp. Video 3). Average time spent in metaphase was 350 minutes in CKAP5 silenced cells whereas it was only 80-90 minutes for cells in the siControl and untreated control groups (Figure 4B & C). Interestingly, this increased average metaphase time was irrespective of the spindle polarity state, suggesting various other spindle abnormalities in these cells in addition to multicentric spindle formation (Supp. Video 3). The majority of the cells that entered mitosis underwent apoptosis (70%) in the CKAP5 silenced group. In 6% of the cells imaged, cells divided even with multicentric spindles however it was difficult to track the fate of these daughter cells due to time as well as 3D plain limitations. The apoptotic cell death occurred via metaphase arrest of cells. In fact, none of the cells that were arrested in metaphase could rescue the arrest and enter anaphase, suggesting the vulnerability of these cells to such damage. Interestingly, we observed a few events where cells transitioned between multipolar and bipolar spindle but in all such events observed, cells remained arrested in metaphase and eventually underwent cell death. In contrast, in the untreated and siControl group there was only 1 event of apoptosis and only single event of multipolar spindle formation was observed (Supp. Video 1 and 2).
Since even cells with bicentric spindle could not complete metaphase and entered apoptosis in CKAP5 silenced group, we questioned the spindle abnormalities in this set of cells. CKAP5 is a +end tubulin binding protein, so we tracked the +ends of microtubules by EB3 tracking for following the tubulin kinetics. Live cell imaging of EB3.eGFP transfected NAR cells through super resolution spinning disc microscopy showed significant differences in EB3 localization as well as kinetics between siCKAP5 and siControl treated cells (Supp. Video 4 & 5). In the siControl group, the majority of the cells showed EB3.eGFP localization only at the + end tips which were highly dynamic during metaphase (73%) whereas the majority of the siCKAP5 treated cells showed EB3.eGFP localization throughout the spindle, with severely non dynamic spindles (Figure 5A.). In both the groups there were a few cases where only the + ends were marked but they were not dynamic (Figure 5A). In the siControl treated group only 15% of non dynamic spindle phenotype was observed. Application of the Utrack analysis program of the MATLAB to measure the dynamics as well as lifetime of EB3 comets clearly showed a significant reduction in the microtubule growth rate in the CKAP5 silenced group (1.6-fold reduction, p<0.05) (Figure 5B). There were small non-significant differences in growth lifetime and microtubule growth length in the CKAP5 silenced group compared to the siControl group (Figure 5C & D). Reduction in spindle length and reduced spindle density in response to CKAP5 silencing was observed (Figure 5A) in CKAP5 silenced groups irrespective of its dynamic status, further confirming the reduced tubulin density in response to CKAP5 silencing. Overall, our results clearly demonstrate a dramatic difference of EB3 phenotype as well as kinetics between CKAP5 siRNA and siControl treated cells.
Biodistribution of LNPs
After establishing the potential and mechanistic aspects of CKAP5 silencing, our next aim was to determine the in vivo therapeutic potential. To this end, we first evaluated the biodistribution of lipid nanoparticles in an intraperitoneal ovarian cancer xenograft mouse model. NAR cells labelled with mCherry and FLUC were intraperitoneally implanted in athymic Nude-Foxn1nu mice and Cy5 labelled control siRNA encapsulated LNPs (Cy5-LNPs) were administered at a 1 mgkg-1 dose. As observed in Figure 6A, in vivo imaging showed a Cy5-LNP signal in the peritoneal cavity of the mice at 2 hours post administration, but the majority of the signal was lost at 4 hours post administration. To better understand the biodistribution and particle localization to the tumor site, we extracted lung, liver, spleen, kidneys and tumor tissue from all the mice and tested the Cy5 and mCherry signal through IVIS imaging (Figure 6B). Cy5 signal was mainly observed in the spleen, liver and tumor tissue area at 2 hours post administration. In the composite image, Cy5 colocalization with mCherry positive tumor cells clearly suggests that the LNPs localize to the tumor tissue. Imaging the tissues at 4 hours post Cy5-LNP administration showed a sparse signal in the liver and tumor tissue area whereas signal from spleen was lost entirely, suggesting particle clearance from the system. These results were further confirmed by quantitative measurement of the Cy5 signal obtained, confirming LNP localization to tumor tissue in addition to liver and spleen tissues (Figure 6C & D).
CKAP5 knockdown leads to reduction in xenografted tumor growth and increased survival
To test in vivo therapeutic gene silencing of CKAP5, NAR mcherry.Fluc cells were intraperitoneally implanted into mice and tumor growth was monitored through IVIS imaging. At 8 days post implantation mice were randomly grouped into the siCKAP5-LNP, siControl-LNP and untreated group (N=20 / group). The mice were treated with a 1 mgkg-1 dose of siCKAP5 or siControl LNPs via an intraperitoneal administration. Overall, 6 doses were administered at 4 day intervals and tumor growth was monitored every 4th day of the treatment by in vivo fluorescence as well as luminescent imaging (Experimental scheme in Figure 7A). Out of 20 mice in each group, 5 mice were sacrificed after 6 doses of treatment and the tumor tissues were collected to test the in vivo silencing efficiency of the LNPs and to check the tissue anatomy by H&E. As shown in Supp. Figure 5 delivery of siCKAP5-LNPs resulted in a significant reduction in CKAP5 expression compared to the control groups, suggesting effective functional effects of our LNP mediated siRNA delivery. In addition, H&E staining of various tissue organs did not show any differences between the siCKAP5, siControl and untreated control groups, suggesting that the particles were not toxic. The remaining mice were monitored for tumor growth kinetics and survival.
Tumor kinetic data demonstrate a significant increase over time in the mCherry as well as Fluc signal in untreated and siControl groups; whereas in the siCKAP5 treated group tumor growth was significantly inhibited by 6-fold (p<0.0001) and only 3 out of 15 mice showed an increased tumor growth post 42 days. More importantly, in 5 mice tumors were completely eradicated and did not relapse for the entire 90 day experiment period. Out of the 3 relapsed tumors in CKAP5 knockdown mice, 2 formed ascites post 80 days of implantation. This decrease in tumor growth resulted in a significantly increased survival in the siCKAP5 treated group. Follow-up of the mice weight suggested an ascites associated weight increase in the siControl-LNPs and untreated groups. The weight increment in the siControl group was similar to the untreated group suggesting no toxic effects of the siControl-LNPs. In the CKAP5 siRNA treated mice, the weight remained relatively constant up to 40 days and thereafter a small increase in mice weight could be noticed.