Construction of Neocentromere-Based Human Minichromosomes for Gene Delivery and Centromere Studies

Construction of neocentromere-based human minichromosomes for gene delivery and centromere studies

LH Wong, R Saffery and KHA Choo The Murdoch Childrens Research Institute, Royal Children’s Hospital, Melbourne, Victoria, Australia

Human neocentromeres are fully functional centromeres that Mb. These minichromosomes exhibit full centromeric arise naturally in non-centromeric regions devoid of ␣-satel- activity, bind to essential centromere proteins, and are lite DNA. We have successfully produced a series of minich- mitotically stable over many generations. They provide a romosomes by telomere-associated truncation of a marker useful system for dissecting the functional domains of com- chromosome mardel(10) containing a neocentromere. The plex eukaryotic centromeres and as vectors for therapeutic resulting minichromosomes are either linear or circular in gene delivery. nature, and range in size from approximately 650 kb to 2 Gene Therapy (2002) 9, 724–726. DOI: 10.1038/sj/gt/3301756

Keywords: neocentromere; human engineered chromosomes; CENP-A; gene delivery

Introduction that appears unaltered from the progenitor chromosome 10.2–6 Recently, we have used a procedure based on Centromeres are essential for the process of stable chro- chromatin immunoprecipitation and array (CIA) analysis mosome inheritance. They are the site of kinetochore for- to determine the distribution pattern of the histone-H3- mation and are responsible for attachment to, and move- related centromere binding protein CENP-A – a key ment along microtubules for the faithful separation of component of the centromeric nucleosome that distingu- chromosomes during mitotic and meiotic cell divisions. ishes functional centromeres from the surrounding In recent years, there have been major efforts in the study chromatin. Using this procedure, which involves immun- of simple and complex centromeres and in the appli- oprecipitation of centromeric chromatin and the cation of centromere knowledge for the construction of hybridization of the extracted DNA to contiguous gen- human engineered chromosomes (HECs) to provide a omic BAC arrays constructed across neocentromeric tool to further understand chromosome and centromere regions, we have identified a 330 kb and 460 kb CENP- properties, and for potential gene delivery and A-binding domain on the 10q25 and a second 20p12 neo- expression studies in gene therapy. Other authors in this centromere, respectively.7,8 volume have covered the use of conventional centrom- Up to the present time, over 51 cases of human neocen- eres that contain repetitive DNA in HEC construction. tromeres originating from 18 different human chromo- We will focus specifically on the use of non-repetitive somes have been described at different genomic DNA-based centromeres (or neocentromeres) to achieve locations, thus establishing neocentromeres as a relatively HEC construction, and discuss some of the unique appli- widespread phenomenon within the human genome.9,10 cations of the ensuing HECs. Construction of neocentromere-based Neocentromere discovery and its wide minichromosomes (NC-MiCs) distribution within the human genome We have employed the strategy of telomere-associated In 1993, we described the first neocentromere lacking the chromosome truncation to remove non-essential chromo- ␣-satellite DNA traditionally associated with human cen- some arms to produce minichromosomes in situ. Human tromere function.1 This neocentromere was identified on telomeric DNA consists of simple tandem repeats of a a mitotically stable marker chromosome, designated hexanucleotide TTAGGG sequence. By introducing small mardel(10), that was devoid of normal centromeric ␣-sat- arrays of cloned telomeric DNA into cells, it is possible to ellite DNA but had acquired a fully functional kineto- truncate the distal portion of a chromosomal arm hence chore at an interstitial 10q25 region with a DNA content leading to the seeding of a new telomere at the site of integration. Targeted truncation is facilitated by the inclusion of homologous genomic sequences within the Correspondence: KHA Choo, The Murdoch Childrens Research Institute, truncation construct. Through sequential rounds of tar- Royal Children’s Hospital, Flemington Road, Melbourne, Victoria 3052, geted truncation using homologous sequences in close Australia proximity to the neocentromere, both chromosomal arms Construction of neocentromere-based human minichromosomes LH Wong et al 725 ation events. In addition to the NC-MiCs shown in Figure 1, our more recent studies have generated NC-MiCs in which both the mardel(10) chromosomal arms were truncated through specific targeting events to yield minichromosomes whose constructions are fully predictable in accordance with the experimental design (data not shown). Structural analyses of the NC-MiCs have included FISH using whole-chromosome paint for each of the 24 human chromosomes, sub-regional paints for chromosome 10, and individual BACs from a 5-Mb contiguous array spanning the 10q25 neocentromeric/perineocentromeric region, as well as selected Figure 1 NC-MiC construction and mapping. Arrows indicate the pos- PCR and/or sequencing studies. The results indicated ition of intended truncations on mardel(10). (T) indicates targeted trunc- that irrespective of whether truncations had occurred ation, whereas (R) indicates random truncation. The long and short arms specifically or randomly, the NC-MiCs were shown to of mardel(10) are denoted as q’ and p’ respectively. Ordered cosmid and contain single-copy DNA derived from the predicted BAC clones mapped to the 10q25 neocentromere region are shown. The 7 10q25 genomic region, with no detectable rearrangements position of the centromere protein CENP-A binding domain is shown. or untoward alterations to the neocentromeric and perin- The approximate sizes of the different NC-MiCs are shown in parentheses, and the structure of each NC-MiC is indicated as either circular (C) or eocentromeric domains. These NC-MiCs ranged in size linear (L). from approximately 650 kb to 2 Mb. The structural morphology of the NC-MiCs was stud- ied by FISH using a pan-telomere probe, immunofluo- can be removed resulting in minichromosome formation. rescence using an antibody to the telomere repeat-bind- Other authors in this volume and elsewhere,11,12 have ing factor TRF1, and pulsed field gel electrophoresis described the use of this procedure for the sequential (which resolves only linear and not circular DNA). The truncation of a human X and Y chromosome to yield a results indicated that some of the NC-MiCs were linear, number of ␣-satellite-containing minichromosomes. while others were apparently circular in structure Using this procedure, we have successfully produced (Figure 1). a series of NC-MiCs by truncation of the mardel(10) chro- The original NC-MiCs showed mitotic stability of mosome at the long- and short-arm regions near the neo- between 35–80% in long-term culture for over 60 cell centromere in the human HT1080 cell line (Figures 1a divisions with or without drug selection pressure. Sub- and 2).13 Despite the use of homologous sequences to cloning of these NC-MiC-containing cell lines consist- encourage site-specific truncation, detailed analysis of the ently yielded cell lines demonstrating >85% NC-MiC resulting NC-MiCs indicated a mixed outcome of specific retention after >100 cell divisions in the absence of selec- targeting and apparently non-targeted, random trunc- tion, suggesting that the decreased instability observed in the original cell lines was likely to be related to the inherent genomic instability of the particular HT1080 cell line used in this study, for which we have additional evi- dence to support.13 The presence of a fully functional neocentromere on the NC-MiCs was further confirmed by the binding of a host of key centromere-associated proteins including CENP-A, CENP-C, CENP-E, CENP-F, hZW10, p55CDC, and Bub1 (see Figure 2 for an example).

Applications of neocentromere-based minichromosomes The NC-MiCs described represent the first examples of mitotically and structurally stable engineered mammalian minichromosomes lacking centromeric repeats. The absence of large tracts of simple repeats, and the availability of fully known genomic DNA sequences of the neocentromere-based minichromosomes provide advantages in at least three different areas of studies compared with the use of repetitive DNA-based mammalian artificial chromosomes. The first application is in the dissection of the functional domains of complex mammalian centromeres Figure 2 FISH and/or immunofluorescence analysis of mardel(10) and using human neocentromeres as a model system. In earl- NC-MiC3. (a, b) FISH using a pan-␣-satellite pTRA-7 probe (green), ier studies, we have demonstrated that the availability of ␣ showing absence of -satellite on mardel(10) (a) and NC-MiC3 (b). Com- well-defined genomic reference markers, in combination bined image (c) and DAPI staining (d) of FISH analysis using E8 (green, with the CIA analysis method mentioned above, have a BAC in the 10q25 neocentromere region) and immunofluorescence using anti-centromere CENP-E antibody (red), showing presence of centromere allowed the high-resolution mapping of the CENP-A- proteins on NC-MiC3. Arrowheads indicate NC-MiC3 or mardel (10) and binding domains of two different human neocentrom- arrows indicate normal chromosome 10. eres.7,8 These studies have provided the first direct evi-

Gene Therapy Construction of neocentromere-based human minichromosomes LH Wong et al 726 dence for the extent of CENP-A binding on any higher HECs for gene therapy applications. Whether these theor- eukaryotic centromere. The same strategy can be readily etical advantages are realisable will depend on the out- applied to other centromere-binding proteins allowing an come of further detailed investigation of the NC-MiCs. accurate mapping of the relative distribution profile for Such investigations should include further attempts at each of these proteins. In addition, it can be used for downsizing existing NC-MiCs, determining the fidelity determining other centromeric and pericentromeric of the regulation and expression of inserted genes, stab- properties, such as the pattern of DNA replication tim- ility in different cell lines in culture and in whole animal, ing,7 scaffold attachment, cohesion, and histone acety- and establishing efficient methods for their isolation and lation. Although such mapping studies can be performed transfer into cells. on the original neocentromere-containing marker chromosomes as we have shown,7,8 chromosome truncation References experiments provide an important means to allow the direct ascertainment of the functional role of the different 1 Voullaire LE, Slater HR, Petrovic V, Choo KHA. A functional domains. In this respect, we have demonstrated, using marker centromere with no detectable alpha-satellite, satellite our smallest 650-kb NC-MiC5, that the minimal 330-kb III, or CENP-B protein: activation of a latent centromere? Am J Hum Genet 1993; 52: 1153–1163. CENP-A-binding domain of the mardel(10) neocentro- 2 du Sart D et al. A functional neo-centromere formed through mere and approximately 320 kb of additional DNA activation of a latent human centromere and consisting of non- flanking this domain are sufficient to provide full kineto- alpha-satellite DNA. Nat Genet 1997; 16: 144–153. chore activity,7 (see Figure 1). The 650-kb NC-MiC5 has 3 Barry AE et al. Sequence analysis of an 80 kb human neocentro- also allowed us to infer that a neocentromere-related 1- mere. Hum Mol Genet 1999; 8: 217–227. Mb region of significantly delayed replication flanking 4 Barry AE et al. The 10q25 neocentromere and its inactive pro- one side of the CENP-A-binding domain can be deleted genitor have identical primary nucleotide sequence: further evi- without affecting neocentromere function, suggesting dence for epigenetic modification. Genome Res 2000; 10: 832–838. that the formation of such a domain of delayed repli- 5 Saffery R et al. Components of the human spindle checkpoint 7 control mechanism localize specifically to the active centromere cation timing is not obligatory to centromere function. on dicentric chromosomes. Hum Genet 2000; 107: 376–384. Future work aimed at further terminal or interstitial 6 Saffery R et al. Human centromeres and neocentromeres show truncation of the minichromosomes, including portions identical distribution patterns of >20 functionally important of the CENP-A-binding domain, should allow us to study kinetochore-associated proteins. Hum Mol Genet 2000; 9:175– in detail the functional roles and importance of individ- 185. ual centromeric and pericentromeric domains. 7LoAWet al. A 330 kb CENP-A binding domain and altered The second application extends those described above replication timing at a human neocentromere. Embo J 2001; 20: to the whole chromosome. Thus, in addition to the dissec- 2087–2096. tion of functions specific to the centromere, the NC-MiCs 8LoAWet al. A novel chromatin immunoprecipitation and array provide excellent models of fully operational mammalian (CIA) analysis identifies a 460-kb CENP-A-binding neocentromere DNA. Genome Res 2001; 11: 448–457. chromosomes in which both their small sizes and fully 9 Choo KHA. Domain organisation at the centromere and neocen- known nucleotide sequence contents should make them tromere. Dev Cell 2001; 1: 165–177. unique and tractable experimental tools for defining the 10 Warburton PE et al. Molecular cytogenetic analysis of eight properties of the higher eukaryotic chromosomes. Some inversion duplications of human chromosome 13q that each of these properties overlap those described above for the contain a neocentromere. Am J Hum Genet 2000; 66: 1794–1806. centromere, but extend across the entire minichromo- 11 Farr CJ et al. Generation of a human X-derived minichromosome some, including the telomeric ends. In addition, such using telomere-associated chromosome fragmentation. Embo J minichromosomes may facilitate the investigation of 1995; 14: 5444–5454. agents that damage chromosome function, cellular mech- 12 Heller R, Brown KE, Burgtorf C. Brown WR. Mini-chromosomes derived from the human Y chromosome by telomere directed anisms that repair such damages, and other processes chromosome breakage. Proc Natl Acad Sci USA 1996; 93: 7125– that affect chromosome dynamics. 7130. A third potential application of the NC-MiCs is to serve 13 Saffery R et al. Construction of neocentromere-based human as an improved tool for the delivery of genes in gene minichromosomes by telomere-associated chromosomal trunc- therapy. Numerous authors have recently discussed the ation. Proc Natl Acad Sci USA 2001; 98: 5705–5710. pros and cons of using mammalian artificial chromo- 14 Brown WRA, Mee PJ, Shen MH. Artificial chromosomes: ideal somes for this application in general,14–17 and have also vectors? Trends Biotech 2000; 18: 218–223. specifically compared and contrasted the use of repetitive 15 Willard HF. Neocentromeres and human artificial chromosomes: an unnatural act. Proc Natl Acad Sci USA 2001; 98: versus 15–17 DNA-based neocentromere-based HECs. 5705–5710. Because of their predictable mode of construction, fully 16 Choo KHA. Engineering human chromosomes for gene therapy definable sequence nature, and the potential to be smaller studies. Trends Mol Med 2001; 7: 235–237. than the repetitive DNA-based HECs, the NC-MiCs bear 17 Saffery R, Choo KHA. Strategies for engineering human chro- a number of theoretical advantages over other types of mosomes with therapeutic potential. J Gene Med 2002; 4:5–13.

Gene Therapy