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 The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 mito- chondria. Due to the multifunctionality of CENP-F, the cellular phenotypes observed upon its deple- tion 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 atre- sia 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.