Depletion of CEP55 in OvCa cells reduces the speed of cellular abscission but does not affect the pAkt level
In order to analyze the oncogenic activities of CEP55 in OvCa cells, its mRNA was stably down-regulated via shRNA in two different OvCa cell lines with high CEP55 expression OVCAR-8 and SKOV-3. SKOV-3 sh1 and sh2 cells had a down-regulation efficiency of 83% and 66%, OVCAR-8 cells of 74% and 62% (Figure S1A). In addition, the specificity of CEP55 depletion was controlled by stably re-expressing the protein in CEP55 sh1 cells (Figure S1B).
Since CEP55 has been shown to control PI3K-Akt signaling, midbody separation and CIN, these cellular processes were compared between control and CEP55 manipulated cells. SKOV-3 and OVCAR-8 cells exhibited a high pAkt level even in non-stimulated cells, showing that in these cell lines Akt-signaling is constitutively activated. However, we did not find an effect of CEP55 depletion on the concentration of pAkt (Fig. 1A), indicating that in OvCa cells a high CEP55 level does not increase constitutively activated PI3K-Akt signaling.
On the other hand, measurement of midbody separation in SiR-tubulin-treated cells by life cell imaging revealed that strong and mild knock-down of CEP55 increased the time necessary for cellular abscission (Fig. 1B). To analyze whether this CEP55-controlled midbody separation was dependent on the CEP55-Alix interaction, a CEP55 mutant with deficient Alix-interaction was re-expressed in SKOV CEP55 sh1 cells (CEP55Y187A). Thereby, we found that CEP55Y187A did not restore slowed cellular abscission of CEP55 depleted cells (Fig. 1C). Thus, also in OvCa cells a high CEP55 level accelerates the speed of cellular abscission in an Alix-dependent manner. However, although cellular division was decelerated in CEP55 depleted cells, no cleavage failures were detected as found in Hela cells (13) (data not shown) and hence no alterations of multinucleated cells (data not shown). Thus, a high CEP55 level accelerates cellular abscission in OvCa cells but is not necessary to complete this process.
In conclusion, in OvCa cells CEP55 does not control constitutively activated PI3K-Akt signaling but accelerates cytokinesis.
Depletion of CEP55 decreases CIN
The impact of CEP55 on CIN was analyzed by different assays. As first read out, the number of cells with properly aligned chromosomes (18) was compared between control and CEP55 shRNA cells. This evaluation revealed that strong down-regulation of CEP55 significantly increased the number of cells with aligned chromosomes, while moderate down-regulation had no significant effect (Fig. 2A). Re-expression of CEP55 reversed decreased misalignment in CEP55 shRNA cells, validating the specificity of the CEP55 knock-down effect (Fig. 2B). Likewise, the number of micronuclei, a consequence of lagging chromosomes(8), was reduced in sh1 cells, and this effect was rescued by re-expressing CEP55 (Fig. 1C).
Since chromosome misalignment can result in a higher diversity of chromosome number(3), chromosomes were counted in control and CEP55-manipulated cells (Fig. 2D, E). We found that the number of chromosomes was highly diverse in chromosomal instable scr control cells. This diversity was significantly decreased in sh1 but not in sh2 cells (Fig. 2D), and chromosome number was again highly diverse in cells re-expressing CEP55WT (Fig. 2E).
To further validate the role of CEP55 in genomic heterogeneity, whole genome, low coverage next generation sequencing of control and knock-down cells was performed. For evaluation of genomic heterogeneity, the median chromosome copy number of control cells was determined and this value was defined as baseline (bold black line in Fig. 3). Thereafter, the chromosome regions closer (green) or more distant (red) from this baseline in knock-down as compared to control cells was investigated. This calculation revealed that in knock-down cells a total of 481.5 Mbp were closer and 238.5 Mbp were more distant from baseline relative to control SKOV-3 cells. In OVCAR-8, 229.9 Mbp were closer and 162 Mbp were more distant. These results indicate that the heterogeneity of chromosomal aberrations within the ovarian cancer cell lines are reduced after knock-down of CEP55.
Together, the results from four different assays revealed that reduced CEP55 expression decreases CIN in OvCa cells.
CEP55 depletion reduces spindle MT-dynamics and increases spindle MT-stability
In contrast to the role of CEP55 in midbody separation, its function in CIN has not been elucidated yet(10). Therefore, the main focus of this study was to identify the CEP55 activity required for its CIN-inducing effect.
Since chromosomal misalignment can result from dysregulated spindle MT-dynamics(3) and CEP55 has been shown to directly bind to MTs in vitro(12), we analyzed whether CEP55 may be involved in the regulation of spindle MT-dynamics. For this, spindle MT-dynamics were compared between CEP55-manipulated cell lines from early to late metaphase. MTs were labeled with SiR-tubulin and spindle speed (µm/h) was assessed by life cell imaging with subsequent reconstruction of 3-D-images (for details, see methods). Calculation of spindle speed showed that, as expected, early metaphase spindle MTs were not motile, thus no differences were found between control and CEP55-manipulated cells. At late metaphase, however, the speed of spindle MTs increased 2.5-fold in CEP55 shRNA cells but only very weakly in control cells as well as in shRNA cells re-expressing CEP55 (Fig. 4A-C). Thus, a high CEP55 level seems to reduce the speed of spindle MT dynamics.
In order to assess whether the alteration of MT-dynamics are associated with changed MT-stability, fixed M-phase cells were stained with antibodies against detyrosinated, thus stabilized MTs(19). The fraction of cells having stable or unstable cells was analyzed by counting cells with low centrosome signals (Fig. 4D, ”unstable”) and those with strong signals in centrosomes and in the spindle (Fig. 3D, ”stable”). This evaluation revealed that a strong CEP55 knock-down reduced the number of cells with stable spindle MTs.
In summary, our data show that down-regulation of CEP55 expression increases the speed and decreases stability of spindle MTs in metaphase cells.
Identification of CEP55 peptides required for MT-binding.
Our data show that CEP55 controls the CIN rate as well as spindle MT dynamics in OvCa cells. In order to reveal whether both findings are related, a CEP55 mutant with deleted MT-binding peptides has to be re-expressed in CEP55 sh1 cells and chromosome alignment and micronucleus formation assessed again. However, the CEP55 peptides required for MT-binding had not been identified yet and therefore this issue was addressed in this study.
Since MT-binding domains are not conserved in amino acid sequence or tertiary structure(20), we predicted the domain architecture of CEP55 using bioinformatics. CEP55 is a coiled-coil protein and so far, only the EABR domain has been characterized on a structural level(16). The secondary CEP55 structure mainly consists of α-helices, and coiled-coil (CC) domains, while β-sheets were only found in the C-terminal region (Figure S2). In addition, the 3-D structure of CEP55 was predicted by AlphaFold(21) (Fig. 5A, and Figure S3). Based on the coiled-coil predictions and the resolved EABR-Domain, we predicted CEP55 as a Dimer. In this model, the EABR domain is connected, N- and C-terminally, to large coiled-coil domains through loops, providing flexibility. The N- and C-termini are structured less complex with loops and short α-helices or ß-sheets. A leucine zipper-like structure seems to be formed, among which the N- and C-termini appear highly flexible.
On the basis of both predictions (Fig. 6A, S2, S3), functional segments were defined as indicated in Fig. 5B. Among these, we first deleted large N- and C-terminal segments (aa 1-126 and aa 362–464) or a large segment inside the protein (aa 127–361) (Fig. 6A). These truncation constructs were expressed in E. coli as His-GFP fusion proteins and enriched by nickel chelate chromatography (Fig. 6B). To analyze binding of His-GFP (control), full-length protein (CEP55FL) and its mutants to MTs, three different methods were conducted. (1) Rhodamine labeled MTs were incubated with GFP-CEP55 fusion proteins and the localization of CEP55 proteins to MTs were analyzed by fluorescence microscopy (Fig. 6C). (2) CEP55 proteins were immobilized to GFP-dynabeads, the beads were incubated with MTs, and binding of MTs to CEP55 was analyzed by Western blotting (Fig. 6D). (3) An MT-pull down assay was performed and binding of the CEP55 proteins to MTs was analyzed (Fig. 6E). Among these methods the last one enables to assess the strength of MT-binding as it is a ratiometric measurement.
All three methods showed that the full-length protein (CEP55FL) bound to MTs while the CEP55127–361 mutant did not. These data reveal that the MT-binding peptides reside in the N/C termini and assume that the CEP55Δ127–361 binds to the same extent to MTs as full-length CEP55 does. However, binding of CEP55Δ127–361 to MTs was reduced by 60% compared to full-length CEP55.
To further locate the N/C peptides required for MT-binding, the N-terminus (aa 1–58; CEP5559–464), the C-terminus (aa 429–464; CEP551–428) and the N- and C-termini together (CEP5559–428) were deleted (Fig. 7A) and assessed by the dynabeads assay. CEP551–428 and CEP5559–464 showed an about 25% reduced, but not a deficient binding to MTs, while binding of CEP5559428 to MTs was nearly completely abolished (> 90%) (Fig. 7B). This result was confirmed by the pull-down assay and by fluorescence microscopy (Fig. 7C, D). Thus, both the N- and the C-terminus are required for MT-binding. The isolated CEP55Δ59428, however, did not bind to MTs as assessed by fluorescence microscopy and by the pull-down assay. Since our AlphaFold prediction did not indicate strong intramolecular interactions (Fig. 8, Figure S3) of peptides inside CEP55Δ59428, masking the MT-binding sites, we conclude that full-length CEP55 is necessary to position the N/C-termini at MTs. Our finding that also CEP55Δ127–631 showed reduced MT-binding (Fig. 6E and 8) supports this conclusion.
Together, our data demonstrate that binding of CEP55 to MTs is mediated by the N- and the C-terminal peptides.
Re-expression of a MT-binding deficient CEP55 mutant does not rescue reduced CIN of CEP55 shRNA cells
The identification of the CEP55 peptides required for MT-binding has now enabled us to analyze the impact of CEP55 MT-binding on CIN. For this purpose, the N/C-terminus of CEP55 was deleted (CEP5559428) and cloned into the lentiviral Lego vector. OVCAR-8 and SKOV-3 CEP55 shRNA cells were infected with the virus coding for CEP5559428 and after selection of the cells expression of CEP5559428 was validated by qPCR (Figure S4).
Before analyzing the impact of CEP5559428 on CIN, we validated the relevance of the C-terminus for cellular CEP55 localization as in U2OS bone osteosarcoma cells the C-terminal amino acids 355–464 are required for CEP55 localization to the midbody and the centrosomes(13). However, as shown in Fig. 9A, the CEP5559428 mutant is still bound to the centrosome and midbody.
In order to analyze the role of CEP55 MT-binding on the CIN rate, chromosomal alignment and micronuclei formation were compared between CEP55 sh1 cells and cells re-expressing CEP55WT or CEP5559428. The results of these experiments revealed that re-expression of CEP5559428 did not rescue the reducing effect of CEP55 depletion on CIN (Fig. 9B, C). Thus, binding of CEP55 to MTs is required for its promoting effect on CIN.
Impact of CEP55 on MT-dynamics in vitro
Our cellular data show that CEP55 controls MT-dynamics but do not indicate whether CEP55 directly controls this process. To assess this, the effect of bacterial expressed CEP55 on MT-polymerization was analyzed in vitro by a turbidity assay. As negative controls, the CEP5559428 and CEP55127–361 mutants were employed. Thereby, we found that CEP55 increased the rate of MT-polymerization 4-fold while the mutants did not affect MT-polymerization (Fig. 10A-C).
Since increased MT-polymerization can result from attenuation of MT-depolymerization (catastrophe) and in cells CEP55 promotes MT-stabilization, we next assessed the impact of CEP55 on MT-stability. For this, Rhodamine labeled MTs were depolymerized in presence of 4 mM CaCl2 at 4°C in absence and presence of CEP55. This experiment revealed that CEP55 protected MTs from cold-induced depolymerization (Fig. 10D).
From these results, we conclude that CEP55 directly promotes MT-polymerization, most likely by attenuating catastrophe. These data are in line with our results obtained by cell culture experiments, showing that a high level of CEP55 stabilizes MTs