Miro-Dependent Mitochondrial Pool of CENP-F and Its Farnesylated C-Terminal Domain Are Dispensable for Normal Development in Mice

Home , CENPF, Locus (genetics)

bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Miro-dependent mitochondrial pool of CENP-F and its farnesylated C-terminal domain are dispensable for normal development in mice

Martin Peterka1,2, Benoît Kornmann1,*

1Institute of Biochemistry, ETH Zurich, 8093 Zürich, Switzerland 2Zurich Life-Science Graduate School, Molecular Life Science Program *Correspondence to [email protected]

ABSTRACT

CENP-F is a large, microtubule-binding protein that regulates multiple cellular processes including chromosome segregation and mitochondrial trafficking at cytokinesis. This multiplicity of function is mediated through the binding of various partners, like Bub1 at the kinetochore and Miro at mitochondria. Due to the multifunctionality of CENP-F, the cellular phenotypes observed upon its depletion are difficult to interpret and there is a need to genetically separate its different functions by preventing binding to selected partners. Here we engineer a CENP-F point-mutant that is deficient in Miro binding and thus is unable to localize to mitochondria, but retains other localizations. We introduced this mutation in cultured human cells using CRISPR/Cas9 and show it causes a defect in mitochondrial spreading similar to that observed upon Miro depletion. We further create a mouse model carrying this CENP-F variant, as well as truncated CENP-F mutants lacking the farnesylated C-terminus of the protein. Importantly, one of these truncations leads to ~80% downregulation of CENP-F expression. We observe that, despite the phenotypes apparent in cultured cells, mutant mice develop normally. Taken together, these mice will serve as important models to study CENP-F biology at organismal level. In addition, because truncations of CENP-F in humans cause a lethal disease termed Strømme syndrome and because CENP-F is involved in cancer development, they might also be relevant disease models.

Introduction characterized by microcephaly, intestinal atresia and other ciliopathy phenotypes (Filges et CENP-F is a large coiled coil protein that was al., 2016; Ozkinay et al., 2017; Waters et al., originally found as a kinetochore binding protein 2015). (Rattner et al., 1993). It is apparent from recent Both the expression level and subcellular work that CENP-F also functions in mitochon- localization of CENP-F are regulated in a cell drial transport, nuclear envelope breakdown, cycle-dependent manner. Undetectable in G1, microtubule polymerization, and transcriptional CENP-F accumulates in the nucleus during regulation via its interactions with the retinoblas- S/G2. At this stage, a fraction of CENP-F is ex- toma protein and ATF4 (Feng et al., 2006; ported and recruited to the outer nuclear enve- Kanfer et al., 2015; Ma et al., 2006; Varis et al., lope (NE) where it interacts with the nucleoporin 2006). Despite its multitude of subcellular local- Nup133 and participates in the recruitment of izations and binding partners, the physiological dynein (Bolhy et al., 2011). Subsequently in function of CENP-F remains poorly defined. To prometaphase, CENP-F relocates to kineto- our knowledge, the only known phenotype ob- chores (Rattner et al., 1993). The molecular served upon CENP-F loss is dilated cardiomyo- mechanism of CENP-F kinetochore binding has pathy described in a heart-specific conditional been studied in great detail (Berto et al., 2018; knock-out mice (Dees et al., 2012). Ciossani et al., 2018; Zhu, 1999). CENP-F is In humans, aberrant expression of targeted to the outer kinetochore via a KT-core CENP-F has been implicated in prostate cancer domain (residues 2792 to 2887, Fig 1A) via syn- (Aytes et al., 2014). In addition, mutations in ergistic action of Bub1 kinase and the kinesin CENP-F are known to cause Strømme syn- Cenp-E (Berto et al., 2018; Zhu, 1999). In turn, drome, a rare autosomal recessive disorder

1 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

CENP-F recruits Ndel1/Nde1/Lis1/Dynein com- tip-mediated transport of cellular cargoes such plex necessary for chromosome segregation. In as chromosomes and mitochondria. addition, a recent report showed that CENP-F is Additionally, CENP-F contains a CaaX mo- a receptor for ATR at the kinetochore and there- tif, which is farnesylated (Ashar et al., 2000). fore an upstream trigger of the ATR-Chk1-Au- However, functional consequences of CENP-F rora B pathway necessary for the proper segre- farnesylation have been debated. The CaaX gation of chromosomes (Kabeche et al., 2017). motif of CENP-F has been proposed to be nec- Nevertheless, the relevance of CENP-F at the essary for its kinetochore and NE localization kinetochore is controversial. Numerous studies (Hussein and Taylor, 2002; Schafer-Hales et reported defects in chromosome segregation al., 2007). Moreover, inhibition of CENP-F far- and/or cell cycle progression upon depletion of nesylation led to defective degradation of the CENP-F or inhibition of its kinetochore localiza- protein and delayed G2/M progression in can- tion (Bomont et al., 2005; Evans et al., 2007; cer cells (Gurden et al., 2010; Hussein and Feng et al., 2006; Holt et al., 2005; Laoukili et Taylor, 2002). On the other hand, more recent al., 2005; Vergnolle and Taylor, 2007; Yang et studies showed only a mild or no effect of farne- al., 2005). While conflicting reports showed no sylation on CENP-F kinetochore localization defects in mitosis upon CENP-F loss (Ciossani (Holland et al., 2015; Moudgil et al., 2015). et al., 2018; McKinley and Cheeseman, 2017; Thus, the function and physiological relevance Pfaltzgraff et al., 2016; Raaijmakers et al., of CENP-F farnesylation remain unresolved. 2018). The plethora of interacting partners and After exerting its function at kinetochores in cellular localizations make studying CENP-F a mitosis, CENP-F is recruited in cytokinesis to challenge, and the cellular phenotypes ob- the outer mitochondrial membrane by the atypi- served upon CENP-F depletion are difficult to cal GTPases Miro1 and Miro2 (Kanfer et al., interpret due to the multifunctionality of the pro- 2015). The Miro-binding domain is highly con- tein. Here, using a single point mutation within served and located near the C-terminus (2977- the Miro-binding region of CENP-F, we genet- 3020, Figure 1A) of CENP-F. Mechanistically, ically separated the mitochondrial function from CENP-F appears to be linking mitochondria with other functions of CENP-F. We used the growing microtubule tips and harnessing the CRISPR/Cas9-mediated mutagenesis to intro- force generated by microtubule growth (Kanfer duce this mutation in human cells. Furthermore, et al., 2015; Kanfer et al., 2017). This process to gain a physiological insight into mitochondrial helps the mitochondrial network to properly dis- function of CENP-F, we engineered a similar tribute throughout the cytoplasm during cytoki- mutation in mice. Moreover, as a by-product of nesis. Upon CENP-F depletion, mitochondria CRISPR/Cas9-mediated engineering, we gen- fail to spread to the cell periphery and remain erated animals bearing truncated CENP-F al- clumped in the perinuclear area, phenocopying leles lacking the last two exons of CENP-F, Miro depletion. Of note, a fraction of CENP-F is which encode the C-terminal microtubule-bind- localized on mitochondria also in G2 (Kanfer et ing domain and the farnesylation motif. In addi- al., 2015; Kanfer et al., 2017). The physiological tion to the loss of the C-terminal domain, one of function of the mitochondrial fraction of CENP- the mutations resulted in approximately 80% F remains unknown. decrease of overall CENP-F expression. Strik- Bracketing its coiled-coils, CENP-F har- ingly, despite the plethora of functions attributed bors two microtubule-binding domains of un- to CENP-F, these mice are viable and fertile, known functions at both termini of the protein and do not display any obvious phenotype. (Feng et al., 2006). Both domains have microtu- These different mouse models will be instru- bule tip-tracking properties and a unique ability mental to systematically study the multiple func- to transport cargo in vitro continuously with both tions of CENP-F in different tissues and under growing and shrinking microtubules, with the C- different physiological or pathological condi- terminus being a more robust mediator of these tions. movements (Kanfer et al., 2017; Volkov et al., 2015). In cells, full-length CENP-F can track growing microtubule tips (Kanfer et al., 2017). These properties of CENP-F further substanti- ate its potential role in supporting microtubule

2 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure 1 - Disruption of CENP-F-Miro complex by single point mutations in CENP-F (A) Domain organization of CENP-F and alignment of Miro-binding domains from chordates. Red asterisks mark residues G2988A and F2989A essential for Miro binding. (B) Yeast two-hybrid assay using strains containing denoted plasmids. Top row: X-gal overlay assay, bottom row: growth assay on leucine dropout medium.

Results of the highly conserved CENP-F residue G2988 or F2989 completely abrogated Miro binding. We have previously mapped the Miro1/2-bind- On the other hand, alanine substitution of the ing domain in CENP-F to a C-terminal region non-conserved residue T2997 did not influence spanning residues 2977-3020 of human Miro1 binding in this assay (Figure 1B). There- CENP-F, which is among the best conserved fore, the interaction between CENP-F and Miro parts of the protein (Figure 1A) (Kanfer et al., in yeast two-hybrid assay can be disrupted by a 2015). Here, in order to specifically disrupt the single point mutation, offering an opportunity to interaction between CENP-F and Miro, we used genetically separate the mitochondrial from alanine-scanning mutagenesis in the yeast two- other functions of this large multifaceted protein. hybrid assay and found that alanine substitution To characterize the properties of CENP-FF2989A in human cells, we engineered the F2989A substitution into the endogenous CENP-F locus using CRISPR/Cas9-mediated genome editing (Figures 2A, B, C). We performed the substitution in human osteosarcoma cell line U2OS, which displays high levels of Miro-dependent mitochondrial recruitment of CENP-F in S/G2 and cytokinesis (Kanfer et al., 2015). Our approach using co-electroporation of Figure 2 - Introduction of the F2989A CENP-F mutation in human cells (A) Target- Cas9 RNP complexes ing strategy to engineer F2989A mutation in CENP-F using CRISPR/Cas9 in U2OS and a 130nt single- cells. 2nt substitution includes TseI restriction site to facilitate screening for correctly stranded DNA homology- modified clones. (B) TseI digested PCR products of the targeted CENP-F locus directed repair template from WT, heterozygous and homozygous CENP-FF2989A U2OS cells. The expected restriction fragment sizes are 341 and 62 bp for the wild type allele, and 260, 81 yielded homozygous and 62 bp for the F2989A allele. (C) Sequencing electropherograms of the targeted F2989A substitution in one CENP-F area from WT and homozygous CENP-FF2989A U2OS cells. (D) Western out of 43 clones screened. blot on WT or CENP-FF2989A U2OS cells using Ab5 antibody. 14 additional clones were

3 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

heterozygous. This relatively high ratio of heter- of the WT CENP-F. It was well expressed (Fig- ozygous versus homozygous modifications ure 2D), accumulated in the nucleus during the might be a consequence of polyploidy of U2OS S/G2 (Figure 3A), localized to the nuclear en- cells (Janssen and Medema, 2012). To obtain velope in early prophase and to the kineto- additional homozygous clones for robust analy- chores in mitosis (Figure 3B, C). However, sis, we re-targeted CENP-F in one heterozy- while WT CENP-F could be seen accumulated gous clone obtained in the first round of target- on mitochondria in S/G2 and cytokinetic cells as ing. In this second round, 8 out of 26 clones previously described, CENP-FF2989A remained were homozygous for CENP-FF2989A (clones cl.2 diffuse in the cytosol throughout these cell cycle to cl.9). phases (Figure 3A). Taken together, these re- CENP-FF2989A recapitulated the cell cycle- sults indicate that F2989A mutation in CENP-F dependent expression and localization patterns solely affects the mitochondrial localization of the protein.

Figure 3 - F2989A mutation disrupts CENP-F localization at mitochondria in human cells without affecting its localization at the nuclear envelope and kinetochores (A) Immunofluorescence of mitochondrial marker expressing (mtBFP) S/G2 or cytokinetic CENP-FWT or CENP-FF2989A cells using a CENP-F antibody (Ab5) (B) Immunofluores- cence of CENP-FWT or CENP-FF2989A U2OS cells in late G2 using CENP-F (Ab5) and AuroraB antibodies. (C) Im- munofluorescence CENP-FWT or CENP-FF2989A U2OS cells in mitosis using CENP-F (Ab5) and AuroraB antibodies. Scale bars 5 μm.

4 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

clumping in CENP-FF2989A cells, we used an in- We have previously shown that siRNA-me- formatics approach that we have developed diated depletion of CENP-F leads to defective previously and that computes the “moment of distribution of the mitochondrial network inertia” of the mitochondrial network. This anal- throughout the cytoplasm. This defect most ysis measures the distance of each mitochon- likely originates in cytokinesis during which drial pixel to the cell’s center of mass, and thus, CENP-F-depleted cells fail to redistribute mito- reflects the spreading of the mitochondrial net- chondria to the cell periphery (Kanfer et al., work (Kanfer et al., 2015). Two independent 2015). To validate that this defect was due to clones (cl.1 and cl.2) bearing the homozygous the absence of the mitochondrial pool of CENP-FF2989A mutation displayed a reduction in CENP-F, and not a side effect of global CENP-F mitochondrial spreading, similar to what has depletion, we imaged mitochondria in the been observed upon complete depletion of Miro CENP-FF2989A background. CENP-FF2989A cells or CENP-F (Figure 4C, and (Kanfer et al., recapitulated the mitochondrial phenotypes of 2015)). Therefore, the Miro-recruited mitochon- global CENP-F or Miro depletion in cytokinesis, drial fraction of CENP-F ensures that mitochon- displaying typical “clumped” mitochondria that dria are properly distributed throughout the cy- failed to extend to the cell periphery (Fig- toplasm in cytokinesis. ure 4A). Moreover, in interphase cells, mutant To gain insight into the physiological func- cells often displayed similar phenotype. This tion of mitochondrial CENP-F at the organismal was not due to a failure of cell spreading, as level, we turned to genome editing in mice to bright-field imaging did not reveal aberrant cell mutate the conserved F2872 residue (cognate shapes (Figure 4B). To quantify mitochondrial to human F2989) to alanine. For this purpose,

Figure 4 - CENP-F-Miro complex is necessary for normal mitochondrial spreading in human cells (A) Immunoflu- orescence of mtBFP-expressing cytokinetic CENP-FWT or CENP-FF2989A U2OS cells using a CENP-F antibody (Ab5). (B) Examples of mitochondrial morphology in live CENP-FWT or CENP-FF2989A interphase U2OS cells stably expressing mtBFP. (C) Quantification of the moment of inertia (mitochondrial spreading) of cells exemplified in (B) Significance calculated using Mann-Whitney U test (N = at least 228 cells per condition) Two independent CENP- FF2989A U2OS clones were used for quantification. Scale bars 5 μm.

5 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure 5 - Introduction of the F2872A CENP-F mutation in mice (A) Tar- geting strategy to generate mice carrying F2872A in CENP-F using CRISPR/Cas9. (B) Sequencing electropherograms of the targeted CENP-F area from WT and F2872A-carrying F0 heterozygous or F1 homozygous mice. (C) 8 weeks old WT and homozygous F2872A litter- mate mice (D) Western blot on primary fibroblasts derived from wild type or F2872A mice using (i) SCF.M or (ii) Ab5 antibody.

we used pronuclear microinjection of preformed Cas9-gRNA complexes together with a single- Although raised against the human protein, stranded 130 bp homology-directed repair tem- Ab5 cross-reacts with the mouse protein, owing plate (Figure 5A). In two microinjection ses- to the good primary sequence conservation in sions, we obtained 4 (20%) founders bearing that area. In fact, the Miro-binding domain is the heterozygous modification of the targeted most conserved region between human and F2872. These mutations were heritable, as mouse species. The discrepancy between the crossing two heterozygous founders resulted in SCF.M and Ab5 antibodies can therefore be ex- homozygous F1 mutant pups in expected Men- plained as follows: the Ab5 polyclonal antibody delian ratios (Figure 5B, Table 1). These ho- preparation might contain antibodies recogniz- mozygous animals (Figure 5C) were viable, fer- ing several epitopes on the human protein, but tile and reached adulthood without any appar- only a limited set of epitopes on the mouse pro- ent phenotype. tein, among which, the Miro-binding domain is a We used western blotting with two different preeminent one. Therefore, we conclude that antibodies to examine if the expression level of the F2872A mutation does not affect protein CENP-F was affected by the F2872A mutation stability but reduces the affinity of the Ab5 anti- in fibroblasts derived from mutant animals. Us- body for mouse CENP-F. ing a polyclonal antibody specifically raised Since the presence of CENP-F on mito- against the central area of mouse CENP-F chondria has not been previously demonstrated (SCF.M antibody targeting residues 1408-1836. in mice, we sought to validate the localization of Courtesy of Stephen S. Taylor), we observed CENP-F in cytokinetic mouse cells using immu- comparable expression levels between WT and nofluorescence. The Ab5 antibody that we cur- the F2872A mutant (Figure 5Di). However, rently use for CENP-F immunofluorescence in contrary to what we observed in the case of the human cells failed to generate a specific signal mutant human cells (Figure 2D), we observed for murine CENP-F, consistent with a difference a reduced signal for the F2872A point mutant in in epitopes between mouse and human pro- western blot when using the widely used com- teins. We therefore performed all immunofluo- mercial CENP-F antibody Ab5 (Figure 5Dii). rescence experiments using the SCF.M anti- Ab5 is a polyclonal antibody raised against the body. Surprisingly, we did not detect any enrich- C-terminal portion of human CENP-F (residues ment of WT CENP-F at mitochondria during cy- 2760-3114), an area that contains the mutated tokinesis in primary mouse fibroblasts (Figure Miro-binding domain, suggesting that the 6A), but another commonly used mouse cell F2872A mutation disrupts an important epitope.

6 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

line NMuMG (mammary gland epithelium), dis- is expressed and localized normally as shown played characteristic CENP-F enrichment on by its nuclear accumulation in G2 and distinct mitochondria during cytokinesis (Figure 6B). kinetochore localization in mitosis (Figure 6D) These data suggest that CENP-F-Miro interac- Taken together, these results demonstrate tion is cell-type specific and prompted us to ex- cell-type specific mitochondrial recruitment of amine CENP-F localization in primary epithelial CENP-F in mice, and show that CENP-FF2872A is cells derived from mammary gland. We thus genetically uncoupled from mitochondria, but harvested mammary glands from WT and mu- retains its other localizations. tant mice and extracted primary Mouse Mam- As a consequence of homologous recom- mary Epithelial Cells (MMECs) by digesting the bination being the limiting step in CRISPR tissue and culturing epithelial mammary gland knock-in experiments, we obtained two addi- organoids (Karantza-Wadsworth and White, tional alleles that resulted from the repair of the 2008; McCaffrey and Macara, 2009). Immuno- Cas9-induced cuts by non-homologous end fluorescence staining of joining, in an error-prone manner (all mouse CENP-F in these cells revealed strong mito- strains created here are summarized in Table chondrial enrichment of CENP-F during cytoki- 2). The first mutation was a 2bp deletion with a nesis in WT cells. On the other hand, no such 4bp insertion at the very beginning of exon 17, enrichment was observed in cells derived from which resulted in a frameshift causing a CENP-FF2872A mutant animals (Figure 6C), indi- frameshift starting from residue G2871 (Table cating that, as in human cells, the F2872 resi- 2, Figure 7A, MUT1). The second allele con- due was essential for CENP-F-Miro interaction tains a 13bp deletion encompassing the intron in mice. Nevertheless, unlike what was ob- 16/exon 17 junction (Table 2, Figure 7A, served in U2OS cells, mitochondria spread MUT2). Thus, both of these mutants are ex- equally well to the cell periphery, when com- pected to yield a truncation of the last two ex- pared to WT cells (Figure 6C). This discrepancy ons, which encode for (1) the C-terminal micro- might be a consequence of differences in mor- tubule-binding domain of CENP-F, (2) part of phology between MMECs and U2OS cells. Un- the Miro-binding domain and (3) the CaaX far- like U2OS cells, MMECs grow in clusters and nesylation motif. The truncation may also affect remain rounded throughout, and after, cytokine- the predicted Rb-interaction domain by remov- sis. Thus, mitochondrial spreading might not re- ing its last five residues (Ashe et al., 2004). Both quire CENP-F in MMECs. of these genotypes resulted in viable homozy- Importantly, as in U2OS cells, CENP-F immunofluorescence revealed that CENP-FF2872A Figure 6 - Mitochondrial CENP-F in mice is cell-type specific and can be disrupted by the F2872A mutation. (A) Immunofluores- cence of mouse fibroblasts stained using CENP-F antibody (SCF.M), Aurora B antibody and mitotracker. (B) Immunofluorescence of mouse mammary epithelial cell line NMuMG stained using CENP-F antibody (SCF.M), mitotracker and DAPI. (C) Immunofluores- cence of CENP-F (SCF.M antibody) in primary MMECs derived from WT or F2872A mice and labelled with Mito- tracker and DAPI. (D) Immu- nofluorescence of CENP-F (SCF.M antibody) in primary MMECs derived from WT or F2872A mice and labelled with DAPI. Scale bars 5 μm.

7 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

gous animals that are fertile and reached adult- We used Immunofluorescence staining to hood without any apparent phenotype (Figure specifically observe the early mitotic population 7B). of cells, which were marked by histone H3 phos- We first analyzed the expression levels of phorylation. While late G2/early M-phase WT the mutant proteins using western blotting on cells displayed the expected nuclear accumula- primary fibroblasts. The Ab5 antibody raised tion of CENP-F, MUT2 cells did not display any against the C-terminus of CENP-F did not de- CENP-F signal above background levels (Fig- tect either allele, confirming that the C-terminus ure 7F). Nevertheless, we could detect residual is missing in both mutants (Figure 7C). When CENP-F in prometaphase at kinetochores (Fig- using the SCF.M antibody raised against an ure 7G). This allowed us to quantify the remain- epitope upstream of the mutations, we ob- ing fraction of MUT2 CENP-F by measuring flu- served that both mutants showed reduced orescence intensities of single CENP-F kineto- CENP-F levels. MUT2 displayed the most sub- chore foci. The mean fluorescence intensities stantial loss of CENP-F expression (Figure 7D), (a.u.) of 341.5 in WT vs. 68.3 in MUT2 suggest prompting us to further analyze the abundance an 80% loss of CENP-F expression in MUT2 and localization of CENP-F throughout the cell (Figure 7H). cycle in these mutant animals.

Figure 7 - Engineering and characterization of CENP-F-truncating mutations in mice (A) CRISPR/Cas9-modified, indel-containing (red) CENP-F alleles in two different mice mutants (MUT1 and MUT2). (B) Adult WT and homozygous MUT1 and MUT2 mice. (C) Western blot using Ab5 antibody on primary fibroblasts derived from WT, MUT1 and MUT2 mice. (D) As in (C) but using the SCF.M. antibody. (E) Average brain weight of 8-week-old WT and MUT2 animals. Significance calculated using Mann-Whitney U test (N=4). (F) Immunofluorescence of late G2 primary fibroblasts from WT and MUT2 mice stained with CENP-F (SCF.M) and p-H3 S10 (06-570) antibodies, DNA labelled with DAPI. (G) Immunofluorescence of mitotic primary fibroblasts from wild type and MUT2 mice using CENP-F (SCF.M) and p-H3 S10 (06-570) antibodies, DNA labelled with DAPI. (H) Quantification of kinetochore CENP-F foci intensities in mitotic wild type or MUT2 cells exemplified in (E). At least 160 single kinetochores from 5 individual cells were quantified for each condition. Significance calculated using Mann-Whitney U test. Mean values depicted in red. (I) Mean intensities of p-H3 areas from mitotic cells exemplified in (E). Significance calculated using Mann-Whitney U test (N=5).

8 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

of CENP-F, which disrupts its interaction with Miro and thus its ability to be recruited to mitochondria. We thoroughly characterize the be- havior of this variant by engineering it in cultured cells, and show that it is completely removed from mitochondria, while retaining other known Table 1 - Heritability of genome-edited CENP-F alleles localizations. To study the physiological relevance of the mitochondrial fraction of CENP-F, we mutated this residue also in mice and validated that CENP-F is removed from mitochondria. Surpris- ingly, these animals are viable and fertile with no apparent burden. The transport of mitochondria is Table 2 - list of mouse lines generated in this study with description a crucial process in long cells like of modified alleles. Numbering of nucleotides and amino acids as in neurons. Its role in non-neuronal ENSMUSG00000026605 cells is much less clear. Preventing A recent report showed that CENP-F re- CENP-F-Miro interaction causes a drastic de- cruits ATR to kinetochores, which leads to acti- fect in mitochondrial spreading in U2OS cells, vation of the Chk1-Aurora B axis, ultimately re- yet does not prevent the development of appar- sulting in S10 phosphorylation of histone H3 ently healthy and fertile mice. This disconnect and suppression of chromosome segregation between in vitro and in vivo phenotypes sup- errors (Kabeche et al., 2017). Our analysis did ports a cell-type specific function of CENP-F- not show any decrease of pH3 S10 in MUT2- mediated mitochondrial trafficking. Indeed, we derived fibroblasts (Figure 7G, I). observe that mouse CENP-F is recruited to mi- CENP-F truncation in MUT2 is similar to al- tochondria in a Miro-dependent fashion in pri- leles responsible for Strømme syndrome. A mary and immortalized mouse mammary epi- hallmark of Strømme syndrome patients is in- thelial cells, but not in primary mouse fibro- testinal atresia, which becomes lethal unless re- blasts. This suggests that Miro-mediated sected. While we did not perform an in-depth CENP-F recruitment to mitochondria serves a phenotypic analysis of MUT2, the normal sur- tissue-specific rather than housekeeping func- vival of these mice excludes the presence of in- tion. What is this function? An in-depth pheno- testinal atresia (Filges et al., 2016; Gao et al., typic analysis of this mouse model is required to 2009). Moreover, 8 weeks old MUT2 mice answer this question. Since the Miro-binding showed normal brain weight, suggesting no dra- domain is one of the best conserved features of matic microcephaly phenotype, another charac- CENP-F, being almost identical from tunicates teristic of Strømme syndrome patients (Figure to human, it is likely that this function of the 7E). Miro-CENP-F interaction is also conserved. What is the molecular basis for the cell-type- Discussion specific Miro-CENP-F interaction? CENP-F Miro-binding domain is heavily phosphorylated CENP-F is an enigmatic protein with a plethora and contains consensus sequences for several of suggested cellular functions ranging from mitotic kinases. It is thus tempting to speculate chromosome segregation to mitochondrial traf- that differential phosphorylation in different tis- ficking. Because of the number of its binding sues and at different cell cycle stages promote partners and involvement in different cellular or inhibit Miro-CENP-F interaction and the sub- processes, a careful genetic dissection is sequent recruitment of CENP-F to the mito- needed to study specific roles of CENP-F. Vari- chondria. ants of CENP-F have been characterized that While the mitochondrial function of separate its functions at the nuclear envelope CENP-F has been recognized only recently, and kinetochores (Berto et al., 2018; Zhu, many conflicting reports exist regarding its func- 1999). tion in mitosis. Some studies advocate that Here, we identify a point mutation CENP-F ensures correct chromosome segrega- (F2989A) in the conserved Miro-binding domain

9 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

tion by regulating kinetochore-microtubule at- straightforward explanation to this discrepancy tachments, thus affecting mitotic checkpoint lies in the fact that most studies to date have and cell cycle progression (Bomont et al., 2005; been performed in heavily transformed cancer Feng et al., 2006; Holt et al., 2005; Kabeche et cell lines. Some of these cell lines might have al., 2017; Yang et al., 2005). On the contrary, adapted their kinetochores to the fast pace of other studies failed to detect any mitotic pheno- their divisions, possibly by becoming heavily de- types upon CENP-F deletion in HeLa cells pendent on CENP-F. Indeed, CENP-F emerges (McKinley and Cheeseman, 2017) or in mouse as a marker that is overexpressed in cancers fibroblasts (Pfaltzgraff et al., 2016). Likewise, (Aytes et al., 2014; Falagan et al., 2018; O'Brien the function of the C-terminal farnesylation of et al., 2007; Zhuo et al., 2015). Here, we find CENP-F remains controversial. A number of re- that CENP-F can be downregulated without ob- ports showed diminished CENP-F localization vious consequences on animal health. There- at kinetochores upon farnesyltransferase inhib- fore, inhibiting CENP-F might be a viable ave- itor (FTI) treatment or upon mutagenesis of the nue for treating CENP-F-dependent cancers. CaaX motif (Gurden et al., 2010; Holland et al., CENP-F is also involved in human genetic 2015; Hussein and Taylor, 2002; Schafer-Hales diseases (Aytes et al., 2014; Filges et al., 2016; et al., 2007). However, a recent study reported Ozkinay et al., 2017; Waters et al., 2015). In that FTI treatment or mutagenesis of the CaaX particular, familial mutations in human CENP-F motif did not prevent kinetochore localization of lead to the Strømme syndrome, a disease char- CENP-F (Moudgil et al., 2015). Thus, further in- acterized by severe ciliopathy phenotypes such vestigations into the function of CENP-F farne- as microcephaly and intestinal atresia. Interest- sylation are needed to clarify these conflicting ingly, one of the non-functional CENP-F vari- observations. ants (p.Arg3094*) detected in Strømme patients Here, in addition to the F2872A mutant, we is only predicted to lack the last 20 residues, in- have generated mice carrying CENP-F alleles dicating that minimal perturbations in CENP-F entirely lacking the C-terminal microtubule-bind- sequence can have drastic effects on human ing domain and the farnesylation motif. Both of health (Filges et al., 2016). Some of the these have been implicated in possible CENP- CENP-F mutant alleles found in Strømme pa- F mitotic functions (Hussein and Taylor, 2002; tients were shown to be hypomorphic (Waters Volkov et al., 2015). The normal survival of et al., 2015). We did not find a significant re- these animals excludes a drastic global defect duction in brain weight in young adult MUT2 an- in chromosome segregation, but it remains pos- imals. However, the possibility is open that sible that certain cell types will be affected. Im- more subtle changes in brain structure might be portantly, one of the mutated CENP-F alleles present. The mouse models presented here of- displays marked decrease of CENP-F expres- fer tools to study specifically the physiological sion. A recent report placed CENP-F as an im- functions of the mitochondrial fraction CENP-F, portant activator of mitotic checkpoint at the ki- of the C-terminal microtubule-binding domain netochore, where it triggers a novel ATR - Chk1- including the farnesylation site, and the conse- AurB - pH3 pathway. The authors observed de- quences of global CENP-F downregulation. In crease in H3 phosphorylation upon expression the future, it will be important to examine the rel- of a presumed dominant-negative form of evance of these mutants as potential models for CENP-F. We did not see any decrease in H3 CENP-F involvement in cancer and Strømme phosphorylation in fibroblasts derived from mu- syndrome. tant animals. While it is possible that residual CENP-F in MUT2 might be fully capable of ATR activation, our observation is in line with the fact that CENP-F-null MEFs and HeLa cells do not display any defect in chromosome segregation (McKinley and Cheeseman, 2017; Pfaltzgraff et al., 2016), supporting that the mitotic function of CENP-F might be cell type or context specific. Despite a host of important functions attributed to CENP-F, the protein appears mostly dispensable for development. The most

10 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Materials and methods wide-field microscope (Olympus IX71) equipped with 100x 1.4NA Oil UPlanSApo ob- Yeast two-hybrid assay jective, DAPI-FITC-TRITC-CY5 filter set LexA operator-driven yeast two-hybrid assay (Chroma) and Roper CoolSnap HQ2 camera. was performed as described in (Golemis et al., Images were deconvoluted with Softworx 4.1.0 2011). Briefly, bait (pEG202-CENP-F2977-3020) (Applied Precision). Final image processing and and prey (pJG4-5-Miro11-594) plasmids were analysis were performed using Fiji (Schindelin constructed as described previously (Kanfer et et al., 2012). al., 2015) and transformed into yeast strain EGY48. For the X-gal assay, transformants Quantification of mitochondrial spreading were plated on Gal/SD-His-Ura-Trp medium For quantification of mitochondrial spreading, and assayed for LacZ reporter induction using live U2OS cells stably expressing mtBFP the X-gal overlay method (Serebriiskii and (Kanfer et al., 2015) were grown and imaged in Golemis, 2000). For the growth assay, trans- Labtek chambers. Images of mitochondria were formants were plated on Gal/SD-Leu medium acquired at 37°C using the wide-field micro- and grown for 48hrs at 30°C. CENP-F alanine scope setup described above. For each image point mutants G2988A, F2989A and T2997A of the mitochondrial network, the moment of in- were prepared by PCR-based site-directed mu- ertia (equivalent of mitochondrial spreading) tagenesis of the CENP-F bait plasmid using pri- was calculated using custom Fiji and Matlab mer pairs 1+2, 3+4 and 5+6 respectively (all pri- scripts as detailed previously (Kanfer et al., mers listed in Table 3). 2015).

Immunofluorescence Quantification of CENP-F kinetochore local- For immunofluorescence staining (IF) using the ization and p-H3 signal intensity αCENP-F (Ab5, Abcam), αAuroraB (Clone To quantify residual CENP-F at kinetochores, 6/AIM-1, BD Biosciences) and p-H3 (S10) (Mili- CENP-F MUT2 and WT mitotic fibroblasts were pore, 06-570) antibodies, cells were grown on fixed and labelled using the SCF.M αCENP-F coverslips to reach ~80 % confluency and fixed antibody, and DAPI to visualize mitotic chromo- using PBS/4 % PFA for 30 minutes at room tem- somes. Z-sections of prometaphase cells were perature. Subsequently, the samples were acquired in 0.3 µM steps using the wide-field mi- washed three times with PBS/0.1 % Triton- croscope described above and then deconvo- X100 and incubated in the blocking solution luted. Images were subsequently processed (PBS/10 % FCS) for 30 minutes. The samples and analyzed in Fiji. In short, the sections were were then incubated with primary antibodies for Z-projected (maximum intensity) and kineto- 1 hour (dilution 1:300 in blocking solution), chore CENP-F signal was calculated using the washed three times with PBS/0.1 % Triton- IntDen - (Area * Mean) formula (IntDen = inte- X100 and incubated with fluorophore-conju- grated signal density of a single kinetochore, gated secondary antibodies (dilution 1:1000 in Area = the size of the selected kinetochore area the blocking solution) for 30 minutes. Finally, and Mean = the mean fluorescence of the the samples were washed three times with CENP-F signal outside the kinetochores). At PBS/0.1 % Triton-X100, three times with PBS least 164 single kinetochores were measured and then mounted to slides using Vectashield for each condition. To quantify phosphorylation mounting medium, optionally containing DAPI of histone H3 phosphorylation in mitotic cells, to visualize DNA. For immunofluorescence MUT2 and WT fibroblasts were immunostained staining of mouse cells using SCF.M antibody for p-H3 (S10) and processed similarly. Mean (kind gift from Stephen Taylor, The University of fluorescence intensities of p-H3 (S10) areas in Manchester), the same steps were performed mitotic cells were measured and normalized to except that cells were fixed with ice-cold meth- the background. anol (-20°C, 10 minutes). To visualize mitochondria, MitoTracker Deep Red FM (Invitro- Western blotting gen/Molecular Probes) was added to the growth Primary skin fibroblasts or U2OS cells were har- media (100nM, 45 minutes) before cell fixation. vested and lysed in 8M urea with 3 % SDS. Imaging was performed using a DeltaVision Equal amounts of protein lysates as measured by the Bradford assay (Bio-Rad) were resolved

11 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

by SDS-PAGE and analyzed by western blot- sgRNA and 100 ng/µl ssDNA donor in the injec- ting using Ab5 or SCF.M antibodies (1:500 dilution buffer (8 mM Tris-HCl, pH 7.4, 0.1 mM tion). EDTA). The mix was injected into pronuclei of fertilized single-cell embryos using standard mi- Generation of CRISPR/Cas9-edited U2OS croinjection procedures. Injected embryos were cells re-implanted into foster mothers. The pups were CENP-F was targeted using electroporation genotyped by PCR on genomic DNA extracted (Neon system, Invitrogen) of Cas9 RNPs and from ear clips using primers 14+15. The result- DNA template for homologous recombination ing PCR product was sequenced using primer (also see Figure 2 for strategy details). Briefly, 14. ~200.000 cells were harvested, washed with PBS and resuspended in 5 µl of Neon R buffer In vitro transcription of sgRNAs (Invitrogen). Shortly before electroporation, sgRNAs were prepared using MEGAshortscript Cas9 RNP mix (100 pmol of Cas9 (NLS-Cas9, T7 Transcription Kit and MEGAclear Transcrip- a kind gift from Martin Jinek, University of Zur- tion Clean-Up Kit (Ambion) according to manu- ich), 120 pmol of sgRNA and 200 pmol of single facturer instructions. The DNA template for in stranded donor DNA template (10, Table 3) in vitro transcription was generated by PCR-as- total volume of 5 µl of Neon R buffer) was added sembly of partially overlapping oligonucleotides to the cell suspension and electroporation was containing T7 promoter sequence, mouse or performed using 10 µl Neon tips with electro- human CENP-F gRNA target sequence (Figure poration parameters set to 1400V, 15 ms, and 4 2A and 5A) and an invariant Cas9 scaffold se- pulses. Immediately after electroporation, cells quence (human: primers 7+8, mouse: primers were seeded into 0.5 ml of pre-warmed growth 9+8). medium and grown for 3 days. Cells were then single-cloned and colonies arising from single Primary mouse cells isolation clones were used for PCR amplification of the Primary skin fibroblasts were isolated as de- targeted CENP-F locus using primers 12+13. scribed step-by-step in (Seluanov et al., 2010). The resulting PCR product was sequenced us- In brief, approximately 1cm2 of skin was col- ing primer 12 or digested by TseI to screen for lected from the underarm area of freshly eu- the intended modification (Figure 2B). Clones thanized ~ 8-week-old animals, cut into small containing homozygous F2989A mutations in pieces and dissociated in 10ml of Liberase TL CENP-F were collected for further analysis. (Roche) solution (0.14 Wunsch units/mL in DMEM/F12 + 1X antibiotic/antimycotic) at 37°C Generation of CRISPR/Cas9-edited CENP-F for 1 hour while stirring. Dissociated tissue frag- mutant mice ments were collected by centrifugation (520g/5 Mice were generated and all animal work was minutes) and washed three times in DMEM/F12 performed in the animal facility of ETH Zurich with 15 % FBS to remove remaining Liberase. (EPIC) according to the Swiss Federal Veteri- The tissue fragments were then plated in nary Office (BVET) guidelines. Animals were DMEM/F12 with 15 % FBS + 1X antibiotic/anti- regularly checked for potential burden. CENP-F mycotic and grown at 37°C, 5 %CO2 for 7 days knock-in and truncated alleles in mice (B6D2F1 allowing fibroblasts to exit the tissue. Mouse x C57Bl/6NTac mixed background) were gener- mammary epithelial cells (MMECs) were iso- ated using a modified Cas9-RNPs approach lated as described in (Karantza-Wadsworth and (Sung et al., 2014). gRNA sequence targeting White, 2008; McCaffrey and Macara, 2009). In the proximity of the CENP-F F2872 codon was brief, 3rd, 4th and 5th pairs of mammary glands selected using the CRISPR design tool were collected from freshly euthanized ~ 8- (http://crispr.mit.edu/) (also see Figure 5A for week-old animals, minced using scissors and details of the targeting strategy). In vitro tran- dissociated in Collagenase A (Roche) solution scribed sgRNA was prepared as described be- (2 mg/ml in DMEM/F12 + 1X antibiotic/antimy- low. ssDNA containing F2872A codon change cotic) at 37°C for 2 hours while stirring. The dis- flanked by 60nt homology arms was used as a sociated template for homologous recombination (11, Table 3). Injection mix was assembled on ice by mixing 200 nM of Cas9-NLS (NEB), 400 nM

12 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

No. name 5'-3' sequence 1 CenpF_G2988Afwd GTTGTAAAGAAAGCGTTTGCTGACATC 2 CenpF_G2988Arev GATGTCAGCAAACGCTTTCTTTACAAC 3 CenpF_F2989Afwd GTAAAGAAAGGGGCTGCTGACATCCCG 4 CenpF_G2988Arev GATGTCAGCAAACGCTTTCTTTACAAC 5 CenpF_T2997Afwd CCGACAGGAAAGGCTAGCCCATATATC 6 CenpF_T2997Arev GATATATGGGCTAGCCTTTCCTGTCGG 7 hCenpFgRNA8fwd GAAATTAATACGACTCACTATAGGGGGATGTCAG- CAAACCCTAAGTTTTAGAGCTAGAAATAGC 8 t7gRNAtemprev AAAAGCACCGACTCGGTGCCACTTTTTCAAGTT- GATAACGGACTAGCCTTATTTTAACTT- GCTATTTCTAGCTCTAAAAC 9 T7_mcenpf_gRNA7 GAAATTAATACGACTCACTATAGGTGGGATGTCAG- CAAACCCTAGTTTTAGAGCTAGAAATAGC 10 dnr_h_gRNA8 ATTATCCTTGTTAAAACCACCTAATGAATATCTTTT- CCACTGTAGATAGAATTGGAC- CTAGCAGTGCTGACAGTTATTTTTTTTGCCCTTAGGGGC TGCTGACATCCCGACAGGAAAGACTAGCCCA 11 cenpf_donor_7 GCCCAGCGAACCCAGCCAAACAGCTAATCAG- TGCCATTACCAACAAGCAGCTCTTTT- GCCCTTAGGGGCTGCTGACATCCCAACTGGAAAGACAA GCCCATATATCCTTCGGAGAACAACCATGGCAACC 12 h_cf_cas9restr_F CTGTGTAGTGTTATTTTCCC 13 h_cf_cas9restr_R GGAGGACTCTGCAAGATTTT 14 cenpfnestfwd TCTTAGCAAGTGCAGCCT 15 cenpfnestrev TCTTCACCTTTTGCCCAT

Table 3 - list of oligonucleotides used in this study

epithelial organoids were pelleted from the sus- Acknowledgments pension by centrifugation (250g/5 minutes) and resuspended in 5ml DMEM/F12. The organoids We are grateful to the EPIC facility of ETH Zur- were again pelleted (250g/5 minutes) and re- ich & Thomas Hennek for help with generating suspended in 10ml PBS/5 % FBS. The pellet and handling CRISPR-edited mice, Martin Jinek was then resuspended and washed with 10ml (University of Zurich) for kindly providing recom- PBS/5 % FBS five times (400g/15 seconds) to binant Cas9 and Stephen S. Taylor (The Uni- remove single cells from the organoids. After versity of Manchester) for kindly sharing the the last washing step, the pellet containing pure SCF.M antibody. We also thank all members of epithelial organoids was resuspended in the Kornmann lab for providing comments on DMEM/F12, 10 % FBS, 5 µg/ml insulin (Sigma, the manuscript. Microscopy was performed at I0516), 1 µg/ml hydrocortisone (Sigma, H0888), the Scientific Center for Optical and Electron Mi- 5 ng/ml hEGF (Sigma, E9644) + 1X PSG mix) croscopy of the ETH Zurich. This work was for 48 hours and switched to serum-reduced funded by grants from the ERC (337906-Orga- medium (DMEM/F12, 5 % FBS, 5 µg/ml insulin, Net) and the Swiss National Science Founda- 1 µg/ml hydrocortisone, 5 ng/ml EGF + 1X PSG tion (PP00P3_13365) to BK. mix) to prevent growth of remaining fibroblasts. Epithelial cells escaped the organoids within 2- 3 days and were immediately processed for im- munofluorescent staining.

13 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

References tumors. Journal of Clinical Oncology 36, e13565-e13565. Feng, J., Huang, H. and Yen, T. J. (2006). CENP-F Ashar, H. R., James, L., Gray, K., Carr, D., Black, is a novel microtubule-binding protein that S., Armstrong, L., Bishop, W. R. and is essential for kinetochore attachments Kirschmeier, P. (2000). Farnesyl and affects the duration of the mitotic transferase inhibitors block the checkpoint delay. Chromosoma 115, 320- farnesylation of CENP-E and CENP-F and 329. alter the association of CENP-E with the Filges, I., Bruder, E., Brandal, K., Meier, S., microtubules. J Biol Chem 275, 30451- Undlien, D. E., Waage, T. R., Hoesli, I., 30457. Schubach, M., de Beer, T., Sheng, Y., et Ashe, M., Pabon-Pena, L., Dees, E., Price, K. L. al. (2016). Stromme Syndrome Is a Ciliary and Bader, D. (2004). LEK1 is a potential Disorder Caused by Mutations in CENPF. inhibitor of pocket protein-mediated cellular Human mutation 37, 359-363. processes. J Biol Chem 279, 664-676. Gao, N., White, P. and Kaestner, K. H. (2009). Aytes, A., Mitrofanova, A., Lefebvre, C., Alvarez, Establishment of intestinal identity and Mariano J., Castillo-Martin, M., Zheng, epithelial-mesenchymal signaling by Cdx2. T., Eastham, James A., Gopalan, A., Dev Cell 16, 588-599. Pienta, Kenneth J., Shen, Michael M., et Golemis, E. A., Serebriiskii, I., Finley, R. L., Jr., al. (2014). Cross-Species Regulatory Kolonin, M. G., Gyuris, J. and Brent, R. Network Analysis Identifies a Synergistic (2011). Interaction trap/two-hybrid system Interaction between FOXM1 and CENPF to identify interacting proteins. Curr Protoc that Drives Prostate Cancer Malignancy. Cell Biol Chapter 17, Unit 17.13. Cancer Cell 25, 638-651. Gurden, M. D., Holland, A. J., van Zon, W., Tighe, Berto, A., Yu, J., Morchoisne‐Bolhy, S., A., Vergnolle, M. A., Andres, D. A., Bertipaglia, C., Vallee, R., Dumont, J., Spielmann, H. P., Malumbres, M., Ochsenbein, F., Guerois, R. and Doye, Wolthuis, R. M., Cleveland, D. W., et al. V. (2018). Disentangling the molecular (2010). Cdc20 is required for the post- determinants for Cenp‐F localization to anaphase, KEN-dependent degradation of nuclear pores and kinetochores. EMBO centromere protein F. J Cell Sci 123, 321- reports 19. 330. Bolhy, S., Bouhlel, I., Dultz, E., Nayak, T., Holland, A. J., Reis, R. M., Niessen, S., Pereira, Zuccolo, M., Gatti, X., Vallee, R., C., Andres, D. A., Spielmann, H. P., Ellenberg, J. and Doye, V. (2011). A Cleveland, D. W., Desai, A. and Nup133-dependent NPC-anchored Gassmann, R. (2015). Preventing network tethers centrosomes to the nuclear farnesylation of the dynein adaptor Spindly envelope in prophase. J Cell Biol 192, 855- contributes to the mitotic defects caused by 871. farnesyltransferase inhibitors. Mol Biol Cell Bomont, P., Maddox, P., Shah, J. V., Desai, A. B. 26, 1845-1856. and Cleveland, D. W. (2005). Unstable Holt, S. V., Vergnolle, M. A. S., Hussein, D., microtubule capture at kinetochores Wozniak, M. J., Allan, V. J. and Taylor, depleted of the centromere‐associated S. S. (2005). Silencing Cenp-F weakens protein CENP‐F. The EMBO Journal 24, centromeric cohesion, prevents 3927-3939. chromosome alignment and activates the Ciossani, G., Overlack, K., Petrovic, A., Huis In 't spindle checkpoint. Journal of Cell Science Veld, P. J., Koerner, C., Wohlgemuth, S., 118, 4889-4900. Maffini, S. and Musacchio, A. (2018). The Hussein, D. and Taylor, S. S. (2002). Farnesylation kinetochore proteins CENP-E and CENP-F of Cenp-F is required for G2/M progression directly and specifically interact with distinct and degradation after mitosis. J Cell Sci BUB mitotic checkpoint Ser/Thr kinases. J 115, 3403-3414. Biol Chem 293, 10084-10101. Janssen, A. and Medema, R. H. (2012). Genetic Dees, E., Miller, P. M., Moynihan, K. L., Pooley, instability: tipping the balance. Oncogene R. D., Hunt, R. P., Galindo, C. L., 32, 4459. Rottman, J. N. and Bader, D. M. (2012). Kabeche, L., Nguyen, H. D., Buisson, R. and Zou, Cardiac-specific deletion of the L. (2017). A mitosis-specific and R loop– microtubule-binding protein CENP-F driven ATR pathway promotes faithful causes dilated cardiomyopathy. Disease chromosome segregation. Science. models & mechanisms 5, 468-480. Kanfer, G., Courtheoux, T., Peterka, M., Meier, S., Evans, H. J., Edwards, L. and Goodwin, R. L. Soste, M., Melnik, A., Reis, K., (2007). Conserved C-terminal domains of Aspenstrom, P., Peter, M., Picotti, P., et mCenp-F (LEK1) regulate subcellular al. (2015). Mitotic redistribution of the localization and mitotic checkpoint delay. mitochondrial network by Miro and Cenp-F. Experimental cell research 313, 2427- Nat Commun 6. 2437. Kanfer, G., Peterka, M., Arzhanik, V. K., Falagan, S., Vlagea, A., Moreno-Rubio, J., Drobyshev, A. L., Ataullakhanov, F. I., Casado, E. and Sereno, M. (2018). A Volkov, V. A. and Kornmann, B. (2017). retrospective analysis of the correlation CENP-F couples cargo to growing and between anti-CENP-F titres and malignant shortening microtubule ends. Molecular Biology of the Cell 28, 2400-2409.

14 bioRxiv preprint doi: https://doi.org/10.1101/415315; this version posted September 12, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Karantza-Wadsworth, V. and White, E. (2008). A Schafer-Hales, K., Iaconelli, J., Snyder, J. P., Mouse Mammary Epithelial Cell Model to Prussia, A., Nettles, J. H., El-Naggar, A., Identify Molecular Mechanisms Regulating Khuri, F. R., Giannakakou, P. and Breast Cancer Progression. Methods in Marcus, A. I. (2007). Farnesyl transferase enzymology 446, 61-76. inhibitors impair chromosomal Laoukili, J., Kooistra, M. R., Bras, A., Kauw, J., maintenance in cell lines and human Kerkhoven, R. M., Morrison, A., Clevers, tumors by compromising CENP-E and H. and Medema, R. H. (2005). FoxM1 is CENP-F function. Mol Cancer Ther 6, required for execution of the mitotic 1317-1328. programme and chromosome stability. Nat Schindelin, J., Arganda-Carreras, I., Frise, E., Cell Biol 7, 126-136. Kaynig, V., Longair, M., Pietzsch, T., Ma, L., Zhao, X. and Zhu, X. (2006). Mitosin/CENP- Preibisch, S., Rueden, C., Saalfeld, S., F in mitosis, transcriptional control, and Schmid, B., et al. (2012). Fiji: an open- differentiation. Journal of Biomedical source platform for biological-image Science 13, 205-213. analysis. Nature methods 9, 676-682. McCaffrey, L. M. and Macara, I. G. (2009). The Seluanov, A., Vaidya, A. and Gorbunova, V. Par3/aPKC interaction is essential for end (2010). Establishing Primary Adult bud remodeling and progenitor Fibroblast Cultures From Rodents. Journal differentiation during mammary gland of Visualized Experiments : JoVE, 2033. morphogenesis. Genes & development 23, Serebriiskii, I. G. and Golemis, E. A. (2000). Uses 1450-1460. of lacZ to study gene function: evaluation McKinley, K. L. and Cheeseman, I. M. (2017). of beta-galactosidase assays employed in Large-Scale Analysis of CRISPR/Cas9 the yeast two-hybrid system. Analytical Cell-Cycle Knockouts Reveals the Diversity biochemistry 285, 1-15. of p53-Dependent Responses to Cell-Cycle Sung, Y. H., Kim, J. M., Kim, H. T., Lee, J., Jeon, Defects. Developmental Cell 40, 405- J., Jin, Y., Choi, J. H., Ban, Y. H., Ha, S. 420.e402. J., Kim, C. H., et al. (2014). Highly efficient Moudgil, D. K., Westcott, N., Famulski, J. K., gene knockout in mice and zebrafish with Patel, K., Macdonald, D., Hang, H. and RNA-guided endonucleases. Genome Chan, G. K. (2015). A novel role of research 24, 125-131. farnesylation in targeting a mitotic Varis, A., Salmela, A.-L. and Kallio, M. J. (2006). checkpoint protein, human Spindly, to Cenp-F (mitosin) is more than a mitotic kinetochores. J Cell Biol 208, 881-896. marker. Chromosoma 115, 288-295. O'Brien, S. L., Fagan, A., Fox, E. J., Millikan, R. Vergnolle, M. A. S. and Taylor, S. S. (2007). Cenp- C., Culhane, A. C., Brennan, D. J., F links kinetochores to McCann, A. H., Hegarty, S., Moyna, S., Ndel1/Nde1/Lis1/Dynein microtubule motor Duffy, M. J., et al. (2007). CENP-F complexes. Curr Biol 17, 1173-1179. expression is associated with poor Volkov, V. A., Grissom, P. M., Arzhanik, V. K., prognosis and chromosomal instability in Zaytsev, A. V., Renganathan, K., patients with primary breast cancer. McClure-Begley, T., Old, W. M., Ahn, N. International journal of cancer 120, 1434- and McIntosh, J. R. (2015). Centromere 1443. protein F includes two sites that couple Ozkinay, F., Atik, T., Isik, E., Gormez, Z., efficiently to depolymerizing microtubules. Sagiroglu, M., Sahin, O. A., Corduk, N. J Cell Biol 209, 813-828. and Onay, H. (2017). A further family of Waters, A. M., Asfahani, R., Carroll, P., Bicknell, Stromme syndrome carrying CENPF L., Lescai, F., Bright, A., Chanudet, E., mutation. American journal of medical Brooks, A., Christou-Savina, S., Osman, genetics. Part A 173, 1668-1672. G., et al. (2015). The kinetochore protein, Pfaltzgraff, E. R., Roth, G. M., Miller, P. M., CENPF, is mutated in human ciliopathy Gintzig, A. G., Ohi, R., Bader, D. M. and and microcephaly phenotypes. Journal of Blanchoin, L. (2016). Loss of CENP-F medical genetics 52, 147-156. results in distinct microtubule-related Yang, Z., Guo, J., Chen, Q., Ding, C., Du, J. and defects without chromosomal Zhu, X. (2005). Silencing mitosin induces abnormalities. Molecular Biology of the Cell misaligned chromosomes, premature 27, 1990-1999. chromosome decondensation before Raaijmakers, J. A., van Heesbeen, R. G. H. P., anaphase onset, and mitotic cell death. Blomen, V. A., Janssen, L. M. E., van Mol Cell Biol 25, 4062-4074. Diemen, F., Brummelkamp, T. R. and Zhu, X. (1999). Structural requirements and Medema, R. H. (2018). BUB1 Is Essential dynamics of mitosin-kinetochore interaction for the Viability of Human Cells in which the in M phase. Mol Cell Biol 19, 1016-1024. Spindle Assembly Checkpoint Is Zhuo, Y. J., Xi, M., Wan, Y. P., Hua, W., Liu, Y. L., Compromised. Cell Reports 22, 1424- Wan, S., Zhou, Y. L., Luo, H. W., Wu, S. 1438. L., Zhong, W. D., et al. (2015). Enhanced Rattner, J. B., Rao, A., Fritzler, M. J., Valencia, D. expression of centromere protein F W. and Yen, T. J. (1993). CENP-F is a .ca predicts clinical progression and prognosis 400 kDa kinetochore protein that exhibits a in patients with prostate cancer. Int J Mol cell-cycle dependent localization. Cell Motil Med 35, 966-972. Cytoskeleton 26, 214-226.