Article DNA Sequence-Specific Binding of CENP-B Enhances the Fidelity of Human Centromere Function Graphical Abstract Authors Daniele Fachinetti, Joo Seok Han, ..., Alex J. Wong, Don W. Cleveland Correspondence [email protected] (D.F.), [email protected] (D.W.C.) In Brief Fachinetti et al. uncover the functional importance of CENP-B, the only human centromere protein known to bind in a DNA sequence-dependent manner. The authors show that by directly enhancing CENP-C recruitment and CENP-C- dependent nucleation of kinetochore assembly, CENP-B increases the fidelity of epigenetically defined human centromere function. Highlights d CENP-B binding to alphoid DNA repeats stabilizes CENP-C and kinetochore nucleation d Centromere function is enhanced by mutual dependencies of CENP-A, CENP-B, and CENP-C d The CENP-B free Y and neocentromere chromosomes mis- segregate at elevated frequencies d CENP-B binding to alphoid DNA enhances fidelity of epigenetically defined centromeres Fachinetti et al., 2015, Developmental Cell 33, 314–327 May 4, 2015 ª2015 Elsevier Inc. http://dx.doi.org/10.1016/j.devcel.2015.03.020 Developmental Cell Article DNA Sequence-Specific Binding of CENP-B Enhances the Fidelity of Human Centromere Function Daniele Fachinetti,1,* Joo Seok Han,1 Moira A. McMahon,1 Peter Ly,1 Amira Abdullah,1 Alex J. Wong,1 and Don W. Cleveland1,* 1Ludwig Institute for Cancer Research and Department of Cellular and Molecular Medicine, University of California at San Diego, La Jolla, CA 92093, USA *Correspondence: [email protected] (D.F.), [email protected] (D.W.C.) http://dx.doi.org/10.1016/j.devcel.2015.03.020 SUMMARY sequence-specific binding protein (Verdaasdonk and Bloom, 2011). Human centromeres are specified by a stably in- Nevertheless, human centromere position is epigenetically herited epigenetic mark that maintains centromere specified (Ekwall, 2007; Karpen and Allshire, 1997). Among the position and function through a two-step mecha- strongest evidence for an epigenetically defined centromere nism relying on self-templating centromeric chro- was the discovery of neocentromeres in humans (Amor et al., matin assembled with the histone H3 variant 2004; du Sart et al., 1997; Ventura et al., 2004; Warburton, CENP-A, followed by CENP-A-dependent nucle- 2004) in which the initial functional centromere has moved from its previous location to a new site in formerly euchromatic ation of kinetochore assembly. Nevertheless, natu- DNA without a-satellite repeats or binding of CENP-B (Depinet ral human centromeres are positioned within spe- et al., 1997; du Sart et al., 1997; Warburton et al., 1997). Often cific megabase chromosomal regions containing associated with chromosomal rearrangements and found in a-satellite DNA repeats, which contain binding sites some types of cancer, each neocentromere is marked with chro- for the DNA sequence-specific binding protein matin stably assembled with CENP-A (Amor et al., 2004), the CENP-B. We now demonstrate that CENP-B centromere-specific histone H3 variant (Earnshaw and Rothfield, directly binds both CENP-A’s amino-terminal tail 1985; Palmer et al., 1987). and CENP-C, a key nucleator of kinetochore as- CENP-A is essential for centromere identity (Black et al., sembly. DNA sequence-dependent binding of 2007b). It marks and maintains centromere position (Black CENP-B within a-satellite repeats is required to sta- et al., 2004; Hori et al., 2013; Mendiburo et al., 2011) and recruits bilize optimal centromeric levels of CENP-C. Chro- additional centromere and kinetochore components (Carroll et al., 2009, 2010; Foltz et al., 2006; Liu et al., 2006) to both mosomes bearing centromeres without bound normal and ectopic centromere locations (Barnhart et al., CENP-B, including the human Y chromosome, are 2011; Guse et al., 2011; Hori et al., 2013; Mendiburo et al., shown to mis-segregate in cells at rates several- 2011). Indeed, use of gene targeting in both human cells and fold higher than chromosomes with CENP-B-con- fission yeast has demonstrated that CENP-A-containing chro- taining centromeres. These data demonstrate a matin is the primary epigenetic mark of centromere identity (Fa- DNA sequence-specific enhancement by CENP-B chinetti et al., 2013). of the fidelity of epigenetically defined human Centromere identity and function is achieved through a two- centromere function. step mechanism. First, new CENP-A assembly to centromeric chromatin is mediated by its CENP-A targeting domain (CATD) (Black et al., 2004, 2007a; Shelby et al., 1997) in conjunction INTRODUCTION with the CENP-A selective chaperone HJURP (Dunleavy et al., 2009; Foltz et al., 2009; Shuaib et al., 2010) whose activity is The centromere is the fundamental unit for ensuring chromo- tightly controlled across the cell cycle (Mu¨ ller and Almouzni, some inheritance. Correct centromere function is required for 2014). In the second step, either the amino- or carboxy-terminal preventing errors in chromosome delivery that would lead to tail of CENP-A is required for nucleation of assembly of a kinet- genomic instability and aneuploidy, both hallmarks of many hu- ochore that mediates high fidelity chromosome segregation man cancers. Although not conserved across species, centro- indefinitely (Fachinetti et al., 2013). Curiously, amino-terminal mere regions from fission yeast to man have arrays of repetitive tail-dependent kinetochore nucleation requires the presence of sequences (Fukagawa and Earnshaw, 2014). Human centro- CENP-B (Fachinetti et al., 2013). Starting from this suggestion meres carry extensive (1,500 to >30,000) copies of imperfectly of a CENP-B-dependent role in kinetochore function, we now repeated arrays of a 171 bp element, termed a-satellite DNA. use a combination of gene targeting and replacement in human In all but the centromere of the Y chromosome (Earnshaw and mouse cells, coupled with in vitro approaches, to identify a et al., 1987, 1989; Miga et al., 2014), this array contains a DNA sequence-dependent contribution to fidelity of human 17-base pair (bp) consensus motif that is the binding site of centromeric function that is mediated by CENP-B binding to CENP-B, the only known mammalian centromeric DNA centromeric a-satellite repeats. 314 Developmental Cell 33, 314–327, May 4, 2015 ª2015 Elsevier Inc. RESULTS decrease of centromere-bound CENP-C [Fachinetti et al., 2013]). CENP-A’s Amino-Terminal Tail Directly Binds the To determine if short-term reduction of CENP-B also had con- Alphoid DNA-Binding Protein CENP-B sequences on CENP-C maintenance at centromeres, we inte- To test the effect that complete loss of the CENP-A amino-termi- grated (at a unique genomic locus using the Flp-In system in nal tail has on centromere-bound CENP-B and on overall cell DLD-1 cells) an siRNA-resistant, doxycycline-inducible gene en- viability, we stably expressed (by retroviral integration) a full- coding CENP-B that was dually tagged with EYFP and AID length CENP-A or a CENP-A variant lacking its amino-terminal (auxin-inducible degron), the latter to enable rapid degradation tail (DNH2CENP-A) in human cells containing one disrupted upon addition of the synthetic auxin indole-3-acetic acid endogenous CENP-A allele and one floxed allele (CENP-AÀ/F) (IAA) (Holland et al., 2012). In agreement with our initial gene (Figure 1A). After Cre-recombinase mediated inactivation of the inactivation approach (Figure 2D), siRNA-mediated CENP-B floxed allele and subsequent loss of endogenous CENP-A pro- depletion led to decrease in centromeric CENP-C by half (Fig- tein (Figures S1A and S1B), long-term cell viability was rescued ures S2B–S2D). In contrast, doxycycline-induced expression of by DNH2CENP-A (Figure 1B), albeit with a 4-fold increase in chro- CENP-BAID-EYFP maintained the initial level of CENP-C at each mosome mis-segregation and micronuclei formation (scored centromere (Figures S2B–S2D), while IAA-induced rapid degra- by live cell imaging in cells stably expressing H2B-mRFP to dation of CENP-BAID-EYFP was accompanied by reduction (again label chromosomes) (Figures 1A and 1C). Furthermore, loss of by half) of CENP-C bound at centromeres (Figure S2D). the CENP-A amino-terminal tail was accompanied by reduced To directly measure the level of CENP-C after siRNA-medi- CENP-B binding at centromeres (Figures 1A and 1D), as ated CENP-B depletion, we EYFP-tagged one or both alleles measured by quantifying centromeric CENP-B intensity by of the endogenous human CENP-C gene using TALEN-medi- immunofluorescence. ated gene targeting (Figures 2E–2G). As expected, quantifica- To determine if this CENP-A-dependent binding of CENP-B at tion of centromere fluorescence intensity of single centromeres centromeres could result from a direct interaction, recombinant in live cells revealed that CENP-C in CENP-C+/EYFP cells was CENP-B was incubated with GST or GST-tagged CENP-A frag- half that of centromeres in which both alleles were targeted ments and GST-containing proteins were affinity purified on (CENP-CEYFP/EYFP)(Figure 2H). Importantly, in agreement with glutathione-immobilized beads (Figures 1E and 1F). CENP-B our indirect immunofluorescence analysis, siRNA-mediated bound directly to the amino-terminal tail of CENP-A (GST- depletion of CENP-B led to loss of half of centromeric EYFP in- CENP-A1–44), but not to GST alone or to a CENP-A mutant lack- tensity (Figure 2H). ing its amino-terminal tail (GST-CENP-AD1–44)(Figure 1F). The first 29 amino acids of the CENP-A tail were sufficient for this Direct Binding of CENP-B to CENP-C interaction (Figure S1C), in agreement with the observation that To determine whether the CENP-B influence on the level of the first 29 amino acids of CENP-A’s amino terminal tail stabilize centromere-bound CENP-C in cells could be mediated by CENP-B binding at centromeres (Fachinetti et al., 2013).