Inactivation of a Human Kinetochore by Specific Targeting of Chromatin Modifiers

1In budding yeast, kinetochore assembly begins with the recognition of key DNA sequences by sequence-specific DNA-binding proteins. The cloning of yeast centromere DNA and the subsequent development of artificial chromosomes (YACs; ) allowed detailed characterization of the protein and DNA components, so that the budding yeast has the best-characterized of all kinetochores. Other kinetochores remain less well understood, partly because as first observed in S. pombe, epigenetic effects play a major role in centromere specification.

2The isolation of human artificial chromosomes (HACs) offers a powerful way to study vertebrate kinetochore assembly. HACs have been obtained in HT1080 human fibrosarcoma cells using regular arrays of alphoid DNA derived from human centromeres. HAC kinetochores accurately mimic the structure and mitotic behavior of endogenous kinetochores. The cloned alphoid DNA assembles into kinetochore-specific chromatin with CENP-A and CENP-C, and carries core histone modifications characteristic of centromeric chromatin.

3A specialized structure of kinetochore chromatin was first shown for the central core of the S. pombe kinetochore, which is flanked by heterochromatin-like regions. More recently, a novel form of chromatin termed “centromere chromatin” in which CENP-A clusters are interspersed with clusters of dimethylated histone H3 Lys4 (H3K4me2) has been proposed for human, D. melanogaster, and rice centromeres, as well as a human neocentromeres.

4The goal of the present study was to establish a system permitting manipulation of the epigenetic status of a single kinetochore in living cells. Prior attempts to manipulate centromere chromatin have involved treatment of cells with chemical inhibitors or protein overexpression and/or knockdowns. Such treatments affect all centromeres, and may alter additional aspects of cellular physiology as well. Here, we describe an HAC in which the underlying alphoid DNA repeat contains a tetracycline repressor (tetR) binding site (tetO). This enables us to target any desired protein into the active kinetochore of the alphoidtetO HAC as a tetR fusion protein. Our results reveal that kinetochore activity appears to depend upon the status of the underlying chromatin. We show that although introducing a transcriptional activator into alphoid DNA can lead to centromere inactivation, surprisingly, introducing a transcriptional repressor into the centromere has a much stronger effect, leading to the rapid and quantitative loss of kinetochore function.

5To obtain a human artificial chromosome (HAC) suitable for manipulation of the epigenetic status of the kinetochore, we designed a novel artificial alphoidtetO dimer with one monomer from a chromosome 17 alphoid 16-mer higher order repeat linked to a wholly synthetic alphoid monomer based on a published consensus sequence. The natural monomer contains a CENP-B binding motif (CENP-B box; ). In the synthetic monomer, this was replaced with a 42 bp tetracycline operator (tetO), the binding site for E. coli tetracycline repressor (tetR) (Figure 1A).

6HAC formation in HT1080 cells requires input naked alphoid DNA of at least 30 kb for functional CENP-A core assembly. The artificial alphoidtetO dimer cloned by conventional means to 3.5 kb was further extended by rolling circle amplification using ϕ29 phage DNA polymerase plus yeast transformation-associated recombination (TAR) cloning (Figure 1B; ). This yielded 50 kb of alphoidtetO dimeric repeat cloned head-to-tail in a BAC vector (BAC32-2-mer(tetO); see Figure S1 in the Supplemental Data available with this article online).

7BAC32-2-mer(tetO) formed HACs when introduced into HT1080 cells. FISH analysis of the transformants with probes specific for alphoidtetO dimer or BAC sequences revealed HACs in 2 of 46 transformant cell lines analyzed (Figure 1C and Table S1A). Both the HAC formation efficiency (4.3%) and fraction of HAC-containing cells (35.7% or 28.6%) in the two cell lines were lower than control values obtained using a natural alphoid DNA array (30% and > 50%, respectively, for a 60 kb BAC containing wild-type chromosome 21 type I alphoid DNA [alphoid11-mer]). The alphoidtetO HACs contained 16 or 48 copies of the input DNA (Table S1B), and lacked host chromosomal DNA detectable with inter- and intra-Alu PCR probes.

8Subcloning of these two original cell lines yielded several cell lines in which one copy of the HAC was maintained stably under nonselective conditions (loss rate = 0.0024 or 0.0054, Table S1B). Subsequent experiments were performed on alphoidtetO HAC-containing cell line AB2.2.18.21. FISH analysis with the BAC probe revealed that the HAC behaved normally during mitosis (Figures 1D and 1E). Thus, despite the low HAC formation efficiency, the alphoidtetO HAC segregates correctly as an independent chromosome.

9To test the ability of tetR to target the alphoidtetO array within the active kinetochore, constructs expressing fluorescent proteins (FP) fused to tetR (mRFP-tetR and tetR-EYFP) were transfected into AB 2.2.18.21 cells. After expression for 24 hr, mRFP-tetR showed a single bright focus in interphase nuclei (Figures 1F′ and 1G′). This was the HAC, as it colocalized with both centromere and kinetochore markers, including CENP-B and CENP-C (Figures 1F and 1G). A specific signal could also be seen on the HAC in mitotic cells (Figure S2).

10Immuno-FISH (Figure S3) confirmed that the HAC (detected with the BAC probe) binds kinetochore proteins, including CENP-A, CENP-C, and CENP-H, as well as INCENP, a key subunit of the chromosomal passenger complex.

11Together, these results confirm that the alphoidtetO HAC assembles a bona fide kinetochore that can be targeted with tetR fusion proteins.

12Chromatin immunoprecipitation (ChIP) analysis confirmed that the synthetic alphoidtetO monomers are located within the active kinetochore on the HAC. Immunoprecipitation with antibodies to CENP-A and CENP-B yielded an enrichment for the alphoidtetO repeat comparable to that seen at control endogenous and HAC centromeres (Figure 2A). Analysis of extended chromosome fibers revealed that CENP-A is distributed along the length of the HAC, and is associated primarily with regions of alphoidtetO sequence (Figure S4).

13ChIP analysis revealed high levels of trimethylated histone H3 Lys4 (H3K4me3), a marker for transcriptionally active chromatin, on the marker gene of the alphoidtetO HAC. Much lower H3K4me3 levels were found on the alphoidtetO repeat itself (Figure 2A). Consistent with this, trimethylated histone H3 Lys9 (H3K9me3), a marker for silent chromatin, was associated with the alphoidtetO array (Figure 2A), where it alternated with regions containing CENP-A in extended HAC fibers (Figure S4). Similar results were obtained with the 11-mer alphoid repeat of native chromosome 21 (alphoidchr. 21) and the alphoid11-mer repeat of the control HAC (Figure 2A).

14H3K4me2, a marker for transcriptionally competent or neutral chromatin, has recently been proposed to be a marker for centromere chromatin. Indeed, regions containing H3K4me2 tended to alternate with regions containing CENP-A in extended HAC fibers (Figure S4). However, we found H3K4me2 associated with the alphoidtetO repeat at levels significantly above those on alphoidchr. 21 or the alphoid11-mer control HAC (Figure 2A). Thus, the alphoidtetO array appears to form a slightly more open/neutral chromatin structure than occurs at native centromeres. However, because the alphoidtetO and control HAC exhibit comparable high stability, these subtle differences must not interfere with kinetochore function.

15Direct binding of tetR-EYFP to the alphoidtetO array in AB2.2.18.21 cells stably expressing tetR-EYFP fusion protein was confirmed by ChIP analysis using anti-GFP antibody (Figure 2B). A stable subline was isolated and maintained in medium containing doxycycline (Dox) to inhibit tetR-EYFP binding to the alphoidtetO array during selection and growth of the cells. Anti-GFP antibody precipitated alphoidtetO DNA fragments only after the cells were grown without doxycycline for 7 and 14 d (Figure 2B). Thus, targeting of the tetR to the alphoidtetO DNA in vivo can be controlled by the presence or absence of doxycycline.

16Targeting of the transcriptional transactivator tTA to the alphoidtetO HAC produced a mosaic response in which the HAC kinetochore was inactivated in some cells, but not in others.

17In a “cell-by-cell” assay, we transfected cells bearing the alphoidtetO HAC with various plasmids, killed the nontransfected cells with puromycin, and scored the percentage of cells bearing 0, 1, or 2 copies of the HAC (detected by FISH with the BAC probe; Figure 3A) after 11–12 d. To control for effects of the transfection procedure on HAC stability, all values were normalized to those for cells transfected with plasmid carrying only a puromycin resistance cassette.

18Transfection of cells with plasmids expressing tetR-EYFP had essentially no effect on HAC stability (Figure 3A). In contrast, tTA expression caused a reproducible increase in the population of cells lacking the HAC (Figure 3A). Transfection with tTA3 (39% as active as tTA in promoting transcription) or tTA4 (14% activity) also caused significant, albeit reduced, levels of HAC loss. The tTA had no effect on HAC stability if cells were grown in the presence of doxycycline.

19In a quantitative population assay, AB2.2.18.21 cells were infected with retroviral vectors expressing the tTA, and the copy number of alphoidtetO dimer in selected clones was quantitated 30 d later by real-time PCR. Cells infected with control vectors expressing the marker gene or tetR alone showed no increase in HAC instability (Figure 3B). Cultures expressing the tTA showed a 6-fold increase in the rate of HAC loss (27% decrease in HAC content). No significant HAC loss was observed in parallel cultures grown in doxycycline (Figure 3B).

20tTA expression caused a mosaic pattern of alphoidtetO HAC loss in clonal cell lines after culturing for 37 d without selective drug (Figure S5). Chromosome loss rates were remarkably high in 43% (10/23) of cell lines, but were either moderate or normal in the remaining cell lines. In controls, expression of mRFP-tetR had a minimal effect on the stability of the alphoidtetO HAC, and expression of the tTA or mRFP-tetR had no effect on loss rates for endogenous chromosomes X or 17, thereby confirming that chromosome loss induced by tTA expression was specific for the alphoidtetO HAC.

21tTA expression caused a ∼2-fold increase in the very low level of alphoidtetO transcripts, but had no effect on the much higher levels of transcription from the bsr gene (Figure 4E). However, examination of the HAC chromatin following tTA expression failed to reveal significant structural differences detected by ChIP with antibodies to H3K4me2 or H3K4me3 (data not shown). This is most likely because tTA binding to the HAC elicits a mosaic response, in which a minority of cells respond by increasing transcription, and destabilizing the HAC (Figures 4B and 4E and Figure S5B′). Since the HAC appears to resist effects caused by tTA binding in most cells, this could explain why the chromatin changes analyzed by ChIP remain below our limit of detection.

22These experiments demonstrate that the alphoidtetO HAC kinetochore can be inactivated in a subset of cells by targeting its chromatin with a transcriptional activator. If, indeed, kinetochore inactivation is induced by transcription of the alphoidtetO array, this result might be similar to that observed in budding yeast, where strong transcriptional bombardment can inactivate a conditional centromere. What was more surprising was that a transcriptional silencer had a much stronger effect and appeared to completely inactivate the kinetochore.

23The tTS, a powerful transcriptional repressor, induced a highly penetrant destabilization of the HAC (Figures 3A and 3B). The tTS is a fusion of the tetR and the Kruppel-associated box (KRAB)-AB silencing domain of the Kid-1 protein (kidney, ischemia, development; ).

24Unlike the variable and weak effects seen following expression of the tTA, AB2.2.18.21 cells showed a 72-fold increase in the rate of HAC loss (97% decrease in HAC content) 30 d after infection with retroviral vectors expressing the tTS (Figure 3B). HAC loss was essentially complete by 14 d of tTS expression. This effect of the tTS was also seen in the “cell-by-cell” protocol of Figure 3A. Expression of a tTS point mutant (tTSmut) unable to bind the corepressor KAP-1 (KRAB-associated protein-1; ) had no effect on alphoidtetO HAC stability even after 14 d (Figure 3B). Thus, destabilization of the HAC by the tTS apparently occurs via the KAP-1 pathway.

25In control experiments, we determined that the differential effects of tetR-EYFP, tTA, and tTS on HAC stability could not simply be explained by differences in the affinity of these proteins for the alphoidtetO array. In ChIP experiments, tetR-EYFP appeared to bind to the alphoidtetO array slightly more efficiently than did tTA-EYFP and tTS-EYFP (Figure 4A). The three proteins exhibited similar levels of binding to the HAC in vivo as measured by quantitative fluorescence (Figures 4B and 4C). However, in some cells, HACs with bound tTA-EYFP appeared to be larger, and had higher levels of associated FP. Importantly, the tTS, which dramatically destabilized the HAC, bound to it at levels comparable to the tetR, which had no observable effect on HAC stability.

26tTS binding to the alphoidtetO array converted the centromeric chromatin to a more heterochromatic state, disrupting the CENP-A kinetochore domain (Figures 4 and 5). This conversion was accompanied by dramatic reduction of levels of alphoidtetO transcription detected by quantitative RT-PCR (Figure 4E). tTS binding caused a rapid loss of H3K4me2 and somewhat slower decrease in CENP-A levels on the alphoidtetO array (Figure 5). We also observed a transient increase in H3K9me3 levels. In controls, binding of tetR to the alphoidtetO HAC caused no detectable changes in the HAC chromatin.

27tTS binding also reduced H3K4me2 and H3K4me3 levels and increased H3K9me3 levels on the marker gene as well as strongly repressing its transcription (Figures 4 and 5). Thus, the closed chromatin structure induced by tTS binding spreads laterally to the BAC vector region. Furthermore, although only 4% of the HAC DNA remained in the population at later times, the fact that H3K9me3 levels remained high on the marker gene while dropping on alphoidtetO DNA suggests that even after tTS binding, subtle differences in chromatin structure remain between the (now inactive) kinetochore and flanking regions.

28The tTS induces HAC loss by disrupting kinetochore structure. If AB2.2.18.21 cells expressing tTS-EYFP were cultured in doxycycline-free medium for 7–8 d to allow tTS binding to the kinetochore, the CENP-A, CENP-B, and CENP-C signals on HACs with associated FP signals were either undetectable or significantly weaker than those on endogenous centromeres (Figures 6, 7B, and 7G and Figure S6C). AlphoidtetO HACs lacking detectable CENP-A were observed in 34% of interphase cells and 39% of mitotic cells (Figures 6E and 6F). In those cells, the FP signal overlapped with significant levels of H3K9me3, a marker for heterochromatin that was not enriched on HACs targeted by tetR-EYFP (Figures 7C and 7D). In parallel control experiments, tetR-EYFP colocalized with CENP-A and CENP-C on the alphoidtetO HAC, confirming that this protein does not disrupt kinetochore structure (Figures 6E and 7A, and Figures S6A and S6B).

29In cells expressing the tTA, we frequently observed that one or both copies of the HAC was spatially separate from the other chromosomes during mitosis (Figure 3C). Similarly, when AB2.2.18.21 cells were examined after 7–8 d of tTS-EYFP expression, we saw HACs with FP signals lying outside the mass of congressed chromosomes in 28% of mitotic cells (Figures 6A, 6D, and 6F). These nonaligned HACs lacked detectible kinetochore proteins, including CENP-A, CENP-B, and CENP-C.

30Loss of the alphoidtetO HAC induced by the various forms of tTA and tTS could occur by two mechanisms: nondisjunction (2:0 segregation) or chromosome loss (failure to replicate, 1:0 segregation). All of our assays revealed that the increase in cells with no HAC was not balanced by a corresponding increase in cells with two HACs. (Note that the cell-by-cell assay scores only HAC signals in the nucleus. See below.) Thus, the HAC initially appeared to be undergoing chromosome loss. However, the fact that tetR-EYFP had no effect on HAC stability suggested strongly that tetR binding to the alphoidtetO array does not interfere with HAC replication, and that loss must be occurring via another mechanism. This is consistent with data shown in Figure 4D in which a HAC with bound tTS is present as two replicated sister chromatids in mitosis.

31Detailed examination of cultures revealed that cells expressing either the tTA or tTS often had minute DAPI-stained structures in the cytoplasm (Figures 3D, 6B, 6C, and 6E and Figure S6E). These resemble tiny versions of micronuclei typically observed when chromosomes fail to segregate correctly in mitosis, and we refer to them as nanonuclei. After 7 d of tTS expression, 11% of interphase cells had nanonuclei containing the alphoidtetO HAC (Figures 6B, 6C, 6E, and 6F and Figure S6E). Nanonuclei lack detectible ACA, CENP-A, or CENP-B staining, indicating that they do not contain functional kinetochores. Although detailed characterization of nanonuclei is beyond the scope of this study, we suspect that sequestration of the alphoidtetO HAC in these tiny structures ultimately results in its loss from the population.

32Together these results indicate that targeting of chromatin-modifying activities into the alphoidtetO kinetochore disrupts centromere chromatin, evicting CENP-A, CENP-B, and other kinetochore proteins, and ultimately inactivating the kinetochore.

33tTS binding to the alphoidtetO array could in theory inactivate the kinetochore by a number of pathways, including the formation of heterochromatin involving H3K9me3 and HP1α. We therefore stained HAC-bearing cells for HP1α following exposure to tetR, the tTA, and the tTS (Figure 7). Of these treatments, only the tTS induced a robust accumulation of HP1α at the kinetochore (Figures 7G″ and 8D). This accumulation was accompanied by disruption of the kinetochore, as detected by the loss of CENP-C staining.

34To test whether the accumulation of HP1α was sufficient to disrupt kinetochore structure, we fused tetR-EYFP to HP1α, thereby targeting the heterochromatin protein by a mechanism independent of core histone modifications. Strikingly, this directed targeting of tetR-EYFP-HP1α to the alphoidtetO repeats was sufficient to disrupt kinetochore structure, as determined by the loss of CENP-C staining (Figure 8B). When cells were scored for the loss of CENP-C staining, the tTS and HP1 were found to yield a comparable effect (Figure 8C). Targeting of HP1α to the alphoidtetO array did not block replication of the HAC, as detected by the presence of two paired sister chromatids (Figure 8B), but did destabilize the HAC (S.C., Jan Bergmann, M.N., Marcella Cervantes, V.L., H.M., and W.C.E., unpublished data). All these results suggest that HP1α recruitment may be sufficient to account for the action of the tTS and that the nucleation of heterochromatin within the array is sufficient to inactivate the kinetochore.

35We have developed a novel human artificial chromosome (HAC) to manipulate for the first time, to our knowledge, the epigenetic status of a single human kinetochore. The HAC was based on a synthetic repeated DNA sequence, the alphoidtetO dimer. One part of this dimer is a natural human alphoid monomer from chromosome 17 containing a CENP-B box. The other has no known counterpart in the natural human genome, and contains a 42 bp tetracycline operator (tetO) in place of the CENP-B box.

36The kinetochore of the alphoidtetO HAC is functionally indistinguishable from the kinetochore of endogenous chromosomes. It binds several key markers for centromeric chromatin, including CENP-A, CENP-C, and CENP-H, as well as H3K4me2. It apparently also contains some normal heterochromatin associated with H3K9me3 and INCENP. Most importantly, the alphoidtetO HAC exhibits normal behavior in mitosis and normal stability in long-term culture.

37The tetO sequences within the active kinetochore of the alphoidtetO HAC remain accessible to tetracycline repressor (tetR). Furthermore, targeting of mRFP or EYFP into the kinetochore as tetR fusions had at most a minimal effect on HAC stability. However, targeting of the tTA transcriptional activator or tTS transcriptional repressor inactivated the HAC kinetochore. Thus, this HAC is, to our knowledge, the first chromosome outside of budding yeast with a conditional centromere.

38Targeting of the tTA to the alphoidtetO array inactivates the kinetochore in some cells. The VP16 transactivation domain of the tTA can directly interact with several general transcription factors including the TATA-binding protein (TBP), TFIIB, and the SAGA histone acetylase complex in vivo. tTA inactivation of the kinetochore could involve either transcriptional activation, or simply induction of a transcriptionally competent open chromatin structure around its binding site (Figure 8E).

39The tTA appeared to destabilize the HAC in some cells but not in others, resulting in a less potent effect overall on HAC stability than the tTS. We have considered two explanations for this. First, CENP-A chromatin may be able to resist transcriptional activation by the tTA. Alternatively, CENP-A chromatin may tolerate a degree of chromatin “opening” while retaining kinetochore activity. The presence of actively transcribed genes within rice centromeres and human neocentromeres, plus the detection of low level transcription of alphoid sequences in humans (Figure 4E) or in DT40 cells with a human chromosome, suggest that in many cases, functional kinetochores can contain “open” chromatin. Possibly, low levels of transcription are tolerated within kinetochore chromatin, however high level transcription triggered by a stochastic event occurring in a subset of cells expressing the tTA may disrupt kinetochore function. It has long been known that bombardment with transcription from a strong promoter can inactivate a yeast kinetochore.

40The tTS was a significantly stronger disruptor of kinetochore function than the tTA (Figure 8E). The tTS can recruit the scaffolding protein KAP-1, which in turn recruits chromatin modification factors including Mi-2α— a core component of the NuRD histone deacetylase complex, SETDB1—the histone H3 Lys9 selective methyltransferase, and heterochromatin protein 1 isoforms. In controls, a tTS point mutant unable to interact with KAP-1 did not destabilize the alphoidtetO HAC, suggesting that recruitment of KAP-1 can lead to inactivation of the alphoidtetO kinetochore. We will report elsewhere a detailed analysis of the role of the various KAP-1 domains in kinetochore inactivation (S.C., Jan Bergmann, M.N., Marcella Cervantes, V.L., H.M., and W.C.E., unpublished data).

41tTS disruption of the kinetochore is unlikely to be a passive byproduct of the recruitment of bulky chromatin remodeling complexes onto the alphoidtetO array. We have shown that a variety of proteins, including tTSmut can be recruited into the alphoidtetO kinetochore without having any detectable effect on its function. Furthermore, it is important to note that the tTA also recruits a different spectrum of chromatin remodeling complexes into the kinetochore, yet has a much milder effect on kinetochore function.

42It appears that the tTS inactivates the alphoidtetO kinetochore by nucleating the formation of classical heterochromatin containing H3K9me3 and HP1. In a previous study, binding of a KAP-1 fusion protein to a targeted transgene reduced levels of histone H3K9 acetylation and H3K4 methylation, and increased H3K9me3 levels at the site. Here, we found that tTS binding decreased H3K4 methylation, increased H3K9me3 levels, and induced HP1 accumulation within the alphoidtetO kinetochore. This heterochromatin-like state spread onto the flanking marker gene, which had previously been in an open, actively transcribing state. In fact, we found that targeting of HP1 directly as a fusion to tetR-EYFP was sufficient to inactivate the kinetochore. This rules out the necessity of other activities of the tTS or KAP-1 in kinetochore inactivation.

43Although it was long taken for granted that centromeres were heterochromatic, recent cytological observations have suggested that this might not be the case. In addition, it has been observed that overexpressed CENP-A does not incorporate into heterochromatin, and in Drosophila, heterochromatin can block neocentromere formation. The experiments shown here demonstrate for the first time, to our knowledge, that targeted nucleation of heterochromatin within the kinetochore abolishes the binding of the centromere specific histone H3, CENP-A, and inactivates a fully functional kinetochore. Although the detailed mechanism remains to be determined, we have considered three ways in which heterochromatin formation could evict CENP-A from the alphoidtetO array. First, the conformation of chromatin fibers containing HP1 might be inconsistent with the relatively rigid structure proposed for CENP-A-containing chromatin. In other studies, we have found that formation of H3K9me3 chromatin prevented CENP-A assembly on an ectopically integrated cluster of short alphoid DNA arrays, provided that CENP-B is present.

44Alternatively, tTS binding leads to a rapid drop in the levels on the alphoidtetO array of H3K4me2, which has recently been described as a marker for centromere chromatin. This drop, which is balanced by a temporary increase of H3K9me3 levels, appears to slightly precede the loss of CENP-A from the alphoidtetO array. This raises the possibility that H3K4me2, possibly together with other core histone modifications, could have an essential role in helping to maintain the structure of centromere chromatin.

45Finally, heterochromatin assembly might inactivate the alphoidtetO kinetochore by suppressing the (low) level of endogenous transcription of the array. Although evidence for RNAi involvement in vertebrate chromosome segregation is scant, RNAi has a clear role in establishment of the flanking heterochromatin at S. pombe kinetochores.

46Until now, the chromatin configuration of active kinetochores has been the subject of careful observation and intense speculation, but has remained inaccessible to direct targeted manipulation. In addition to providing a clear demonstration that experimentally induced heterochromatin within the centromere is incompatible with kinetochore activity, our studies reveal that the remarkable dynamism of the chromatin within the kinetochore of the alphoidtetO HAC should allow the systematic manipulation of the “histone code” within the kinetochore and definition of the full epigenetic signature of centromere chromatin.

47Standard procedures, including descriptions of cell lines, the BAC transfection protocol, analysis of de novo HAC formation analyses by fluorescence in situ hybridization (FISH), construction of tetR-fusion proteins expression vectors, Chromatin immunoprecipitation (ChIP), real-time PCR, and indirect immunofluorescence are discussed in the Supplemental Experimental Procedures.

48The alphoidtetO monomer sequence was assembled by coligation of five double-stranded oligomers and cloned into pBluescript using the XhoI/SalI sites. A naturally occurring alphoid monomer (168 bp) containing a consensus CENP-B box was isolated from p17H8. The 2.7 kb higher-order repeat was digested with BsmI and the appropriate DNA fragment was isolated after agarose gel electrophoresis. The fragment was blunt-ended with T4 polymerase and cloned into the SmaI site of pBluescript. After amplification of the monomer by PCR to introduce SalI and XhoI sites, it was inserted into pBluescript next to the tetO monomer.

49Subsequent cycles of digestion and cloning the XhoI/SalI site of pBluescript yielded p3.5α, containing 10 alphoidtetO dimer fragments (Figure 1A). The extension of the tetO dimer alphoid repeat was carried out by rolling-circle amplification (RCA) for the circularized template of the alphoidtetO 10-mer excised from p3.5α, using ϕ29 DNA polymerase. Resulting RCA products were cloned into a targeting vector in yeast by transformation-associated homologous recombination (TAR) according to previously described methods. The targeting vector (RSA/SAT43) contains YAC (HIS3, CEN6, ARSH4) and BAC (Cm, ori F) cassettes as well as a mammalian selectable marker (bsr).

50Purified genomic DNA from the yeast clones was electroporated into Escherichia coli cells (DH10B, Invitrogen). The insert size of the synthetic alphoidtetO DNA array was analyzed by PFGE (CHEF, Bio-Rad) after NotI digestion of BAC DNA.

51AB2.2.18.21 cells expressing tetR, tTA, and tTS fusion proteins were constructed by retroviral vector gene expressing system. Infected cells were treated by geneticin and/or 1 μg/ml of doxycycline for 30 d (and 7 or 14 d for tTS expressing cell). The genomic DNAs of each cell lines were purified using DNeasy Tissue Kit (QIAGEN) according to the manufacturer's instructions. Purified genomic DNAs were sonicated (Bioruptor sonicator, Cosmo Bio, Japan) for 3 min. One hundred nanograms of genomic DNA was used for analysis by real-time PCR with primers tet-1 and tet-3 using the iCyclerIQ™ MultiColor Real Time PCR Systems (Bio-Rad). The following primer sets were used: 5SDNA-F1 and 5SDNA-R1 for 5S ribosomal DNA, and tet-1 and tet-3 for the alphoidtetO dimer.

52Cells were transfected with Lipofectamine 2000 (Invitrogen) at 90% confluence and after 24 hr puromycin was added to the medium at a concentration of 1.5 μg/ml. After 24–30 hr the medium was changed and the cells incubated for 4 d before being selected again with the same concentration of Puromycin for 24–30 hr. After the second selection, cells were incubated in fresh medium for 5 d, plated on poly-lysine-coated slides and processed for fluorescence in situ hybridization (FISH) using a BAC-based probe. The slides, fixed with Carnoy's fixative, were left to age overnight. The probe was denatured for 5 min at 95°C and added to the slides, which were incubated at 72°C for 2 min before overnight incubation at 39°C. After washes with 0.1 × SSC (20 × SSC: 3 M NaCl, 0.3 M NaCitrate) at 65°C followed by a wash with 4 × SSC + 0.1% Tween20 at RT, slides were incubated (with intervening washes), successively, with FITC-avidin, biotinylated-anti-avidin, and FITC-avidin. Slides were mounted with VectaShield.