Centromeres: the missing link in the development of human artificial chromosomes

Making CENs of mammalian artificial chromosomes

Mammalian artificial chromosomes (MACs) hold the promise of providing autonomous vectors for gene therapy in dividing cells. They would not require insertion into the genome and could include sufficient genomic sequences that surround the therapeutic gene to ensure proper tissue-specific and temporal regulation. Several groups have reported successful formation of MACs in human cells using transfection strategies that included alpha satellite DNA, the primary DNA found at normal human centromeres. These results, although extremely encouraging, have limitations such as unpredictable chromosome formation and success thus far in only one transformed human cell line. Examination of other cells where alpha satellite DNA has integrated into ectopic chromosomal locations, as well as naturally occurring dicentric and neocentromere-containing cell lines, suggests that alpha satellite DNA may not be necessary or sufficient for centromere formation. Overall, these results suggest that epigenetic modifications of centromeric DNA are required for efficient centromere formation. Models for this centromere-specific epigenetic modification include a specialized chromatin structure and differential replication timing of centromeric DNA. Thus, further investigation of these centromere-specific epigenetic modifications may suggest strategies for increasing the efficiency of generating human artificial chromosomes for use as gene therapy vectors.

Hypothesis: for the worst and for the best, L1Hs retrotransposons actively participate in the evolution of the human centromeric alphoid sequences

A number of questions concerning the evolution and the function of the alpha satellite DNA sequences present at the centromere of all human chromosomes are still open. In this paper, we present data which could contribute to understanding these points. It is shown here that the alphoid sequences within which L1 elements are found are quite divergent from those of the homogeneous alphoid subsets present at each centromere where none has so far been detected. In addition, a number of L1s are detected close to the ends of the alpha satellite blocks. A fairly high proportion exhibit a polymorphism of presence/absence. Strikingly, several L1s localized at a distance from each other are always either present or absent simultaneously. This is interpreted as resulting from intrachromosomal recombination, through distant L1s, leading to deletion of several of them at once together with their surrounding alphoid sequences. The parameters determining which portion of the several megabases of alphoid sequences is actually involved in the centromeric function are not known. From the above data we suggest that the alpha satellite domain within which DNA sequences are recruited to form a centromere is both homogeneous in sequence and uninterrupted by L1s or any other retrotransposons. Conversely, non-centromere competent alphoid sequences would be both divergent and punctuated by scattered L1 elements, particularly at the borders of the alphoid blocks. On the grounds of these data and hypotheses, a model is presented in which it is postulated that accumulation of L1 insertions within a centromere competent alphoid domain is ruining this competence, the consequence being damage to or even loss of the centromere-forming capability of the chromosome. Restoration of fully centromere-forming competence is supposed to occur by two alternative means, either de-novo amplification of a homogeneous and uninterrupted alphoid domain or by unequal crossing over with a homologue harbouring a large competent one. If L1 retrotransposons are acting detrimentally to centromere integrity (for the worst), one must also consider them as having positive consequences on chromosomes by preventing their centromeres from swelling indefinitely by the addition of alphoid sequences (for the best). The data and ideas presented here fit well with those already put forward by Csink and Henikoff (1998) using the example of Drosophila.

Use of a human minichromosome as a cloning and expression vector for mammalian cells

A natural human minichromosome (MC1) derived from human chromosome 1 was shown to be linear and to have a size of 5.5 Mb. Human IL-2 cDNA and the neo gene were co-transfected into a MC1-containing human-CHO hybrid cell line. Integration of the foreign genes was directed to the pericentromeric region of MC1 by co-transfection of chromosome 1-specific satellite 2 DNA. A number of G418-resistant transfectants were obtained and expression of IL-2 was determined. FISH analysis demonstrated co-localization in the minichromosome of the IL-2 gene and of the satellite 2 DNA. An IL-2-producing clone was used in cell fusion experiments with IL-2-dependent murine CTLL cells to generate CTLL-human hybrids containing the modified minichromosome (MC1- IL2 ). The hybrids were able to grow in medium lacking IL-2 for 17 mean population doublings (MPD), indicating that expression of the cytokine was sufficient to relieve the IL-2 dependence of CTLL proliferation. Endogenous IL-2 production delayed the onset of apoptosis in the IL-2-dependent CTLL cells. Mitotic stability was shown to be 100% in the human-CHO hybrids and 97% per MPD in CTLL cells. These results demonstrate that a natural human minichromosome can be utilized as a cloning and expression vector for mammalian cells and that the MC1 minichromosome can be engineered to deliver IL-2 to two types of cells, fibroblasts and lymphocytes.

Proper metaphase spindle length is determined by centromere proteins Mis12 and Mis6 required for faithful chromosome segregation

High-fidelity chromosome transmission is fundamental in controlling the quality of the cell division cycle. The spindle pole-to-pole distance remains constant from metaphase to anaphase A. We show that fission yeast sister centromere-connecting proteins, Mis6 and Mis12, are required for correct spindle morphogenesis, determining metaphase spindle length. Thirty-five to sixty percent extension of metaphase spindle length takes place in mis6 and mis12 mutants. This may be due to incorrect spindle morphogenesis containing impaired sister centromeres or force unbalance between pulling by the linked sister kinetochores and kinetochore-independent pushing. The mutant spindle fully extends in anaphase, although it is accompanied by drastic missegregation by aberrant sister centromere separation. Hence, metaphase spindle length may be crucial for segregation fidelity. Suppressors of mis12 partly restore normal metaphase spindle length. In mis4 that is defective in sister chromatid cohesion, metaphase spindle length is also long, but anaphase spindle extension is blocked, probably due to the activated spindle checkpoint. Extensive missegregation is caused in mis12 only when Mis12 is inactivated from the previous M through to the following M, an effective way to avoid missegregation in the cell cycle. Mis12 has conserved homologs in budding yeast and filamentous fungi.

Generation of an approximately 2.4 Mb human X centromere-based minichromosome by targeted telomere-associated chromosome fragmentation in DT40

A linear mammalian artificial chromosome (MAC) will require at least three types of functional element: a centromere, two telomeres and origins of replication. As yet, our understanding of these elements, as well as many other aspects of structure and organization which may be critical for a fully functional mammalian chromosome, remains poor. As a way of defining these various requirements, minichromosome reagents are being developed and analysed. Approaches for minichromosome generation fall into two broad categories: de novo assembly from candidate DNA sequences, or the fragmentation of an existing chromosome to reduce it to a minimal size. Here we describe the generation of a human minichromosome using the latter, top-down, approach. A human X chromosome, present in a DT40-human microcell hybrid, has been manipulated using homologous recombination and the targeted seeding of a de novo telomere. This strategy has generated a linear approximately 2.4 Mb human X centromere-based minichromosome capped by two artificially seeded telomeres: one immediately flanking the centromeric alpha-satellite DNA and the other targeted to the zinc finger gene ZXDA in Xp11.21. The chromosome retains an alpha-satellite domain of approximately 1. 8 Mb, a small array of gamma-satellite repeat ( approximately 40 kb) and approximately 400 kb of Xp proximal DNA sequence. The mitotic stability of this minichromosome has been examined, both in DT40 and following transfer into hamster and human cell lines. In all three backgrounds, the minichromosome is retained efficiently, but in the human and hamster microcell hybrids its copy number is poorly regulated. This approach of engineering well-defined chromosome reagents will allow key questions in MAC development (such as whether a lower size limit exists) to be addressed. In addition, the 2.4 Mb minichromosome described here has potential to be developed as a vector for gene delivery.