A High Mobility Group Protein Binds to Long CAG Repeat Tracts and Establishes Their Chromatin Organization in Saccharomyces cerevisiae*

Long CAG repeat tracts cause human hereditary neurodegenerative diseases and have a propensity to expand during parental passage. Unusual physical properties of CAG repeat tracts are thought to contribute to their instability. We investigated whether their unusual properties alter the organization of CAG repeat tract chromatin. We report that CAG repeat tracts, embedded in yeast chromosomes, have a noncanonical chromatin organization. Digestion of chromatin with the restriction enzyme Fnu4HI reveals hypersensitive sites occurring ∼125 bp apart in the repeat tract. To determine whether a non-histone protein establishes this pattern, we performed a yeast one-hybrid screen using CAG repeat tracts embedded in front of two reporter genes. The screen identified the high mobility group box protein Hmo1. Chromatin immunoprecipitation of epitope-tagged Hmo1 selectively precipitates CAG repeat tracts DNAs that range from 26 to 126 repeat units. Moreover, deletion of HMO1 drastically alters the Fnu4HI digestion pattern of CAG repeat chromatin. These results show that Hmo1 binds to CAG repeat tracts in vivo and establish the basis of their novel chromatin organization.

Long CAG repeat tracts cause human hereditary neurodegenerative diseases and have a propensity to expand during parental passage. Unusual physical properties of CAG repeat tracts are thought to contribute to their instability. We investigated whether their unusual properties alter the organization of CAG repeat tract chromatin. We report that CAG repeat tracts, embedded in yeast chromosomes, have a noncanonical chromatin organization. Digestion of chromatin with the restriction enzyme Fnu4HI reveals hypersensitive sites occurring ϳ125 bp apart in the repeat tract. To determine whether a non-histone protein establishes this pattern, we performed a yeast one-hybrid screen using CAG repeat tracts embedded in front of two reporter genes. The screen identified the high mobility group box protein Hmo1. Chromatin immunoprecipitation of epitope-tagged Hmo1 selectively precipitates CAG repeat tracts DNAs that range from 26 to 126 repeat units. Moreover, deletion of HMO1 drastically alters the Fnu4HI digestion pattern of CAG repeat chromatin. These results show that Hmo1 binds to CAG repeat tracts in vivo and establish the basis of their novel chromatin organization.
Long CAG repeat tracts cause more than 10 human hereditary neurodegenerative diseases, including Huntington disease, myotonic dystrophy, and a number of hereditary ataxias (1). Disease-causing repeat tracts are dynamic mutations that frequently expand during parental passage. The length of CAG repeat tracts correlate with their instability and with the severity and early onset of the disease. The underlying cause of CAG repeat tract instability and propensity to expand may be their ability to assume unusual DNA structures. First, single strands of CAG and its complement CTG can undergo intramolecular base pairing to form hairpin structures, and this is thought to interfere with their replication when the parental strands are unwound (2)(3)(4). Second, double strand CAG repeat tracts may also have novel physical properties. For example, duplex CAG tracts have unusually rapid electrophoretic mobility that is proposed to result from a high degree of flexibility (5)(6)(7).
The physical properties of duplex CAG repeat tracts may be important for how CAG repeat tracts are packaged in chromatin. Studies carried out with purified histone cores show that they bind preferentially to CAG repeat tracts in comparison to random sequence DNA (8 -12). Curiously, repeat tracts as small as 6 units bind octamer cores, and the affinity of binding does not increase with tract length (9). These results have been taken to mean that CAG repeat tracts may position nucleosomes (9,10), and this is supported by a study showing increased DNase I protection of sequences adjacent to long repeat tracts in cultured cells (13).
In this study we have revisited the issue of chromatin organization of CAG repeat tracts by attempting to directly assess nucleosome positioning in vivo. In doing so we have discovered that repeat tracts are bound by a high mobility group (HMG) 2 box protein that forms the basis of their unusual chromatin organization.

MATERIALS AND METHODS
Yeast Strains-All yeast strains in this study were derived from strains SSL204 (14) or PJ69-4 (15). Strain CAG325-126 was generated by one-step gene replacement of trp1-101 in SSL204 with an ARS1-TRP1 fragment in which a CAG repeat tract of 126 repeat units was placed into the NheI site. The repeat tract is oriented so that the strand containing CAG is the lagging strand template when replication begins at ARS1. Derivatives of CAG325-126 with shorter tract lengths were generated by taking advantage of the propensity of yeast to contract long CAG repeat tracts. HMO1 was deleted from strain SSL204 and its derivatives by one-step gene replacement using amplified PCR products of the KanMX4 cassette knock-out of HMO1 (Open Biosystems YSC 1021). Strains CAG101 and SSL520 msh2⌬ were previously described (16,17). Strain Y1H was derived from strain PJ69-4 (15) by substituting the upstream-activating elements of the GAL promoter regions of ADE2 and lacZ with a CAG repeat tract of 130 repeat units by two-step gene replacement.
Nuclei Isolation and Fnu4HI Digestion-Yeast nuclei were isolated based on the method described by Almer et al. (18). Nuclei prepared by this method were washed in Fnu4HI digestion buffer (20 mM Tris-Ac (pH 7.9), 50 mM KOAc, 10 mM MgAc, 1 mM dithiothreitol), centrifuged, and suspended in Fnu4HI digestion buffer (ϳ5 ml/g cell). Nuclei in 0.2 ml were digested with 5 or 10 units of Fnu4HI at 37°C for 10 min. Reactions were stopped by addition of the same volume of stop solution (2% SDS, 50 mM EDTA) followed by incubation with 2 mg/ml proteinase K at 65°C for 2 h. DNA was purified by phenol/chloroform extraction and ethanol precipitation. DNA was hydrated in 50 l of TE buffer (10 mM Tris (pH 8.0), 0.1 mM EDTA).
Indirect End Labeling-Purified DNA was digested with EcoRI or EcoRV for the repeat tract embedded at ARS1-TRP1, BanI for the repeat tract located within ADE2, and NheI or AvaII for the SPC25 sequence. Fragments were subjected to electrophoresis through a 1% agarose gel and transferred to MagnaGraph nylon membrane (Osmonics) by capillary blotting. DNA fragments were hybridized by standard methods with appropriate probes and detected using a PhosphorImager (Fuji).
To determine the distance between Fnu4HI digestion sites, we first determined the length of the repeat tract by PCR and electrophoresis of the PCR products through a DNA sequencing gel (supplemental Fig. 1). After Fnu4HI digestion of the chromatin, electrophoresis, and indirect end labeling, the lengths of the observed fragments were calculated using the ImageGage program (Fuji) (supplemental Fig. 2).

Long CAG Repeat Tracts Have a Noncanonical Chromatin
Organization-To investigate the chromatin organization of long CAG repeat tracts, we first generated strain CAG325-126 by embedding a CAG tract of 126 repeat units (378 bp) adjacent to ARS1 in Saccharomyces cerevisiae chromosome IV (Fig. 1A). We performed indirect end labeling (IEL) by digesting nuclei with an endonuclease, purifying the DNA, and cutting with a restriction enzyme that flanks the repeat tract. For our analyses we used two probes labeled RI and RV that flank the repeat tract on either side (Fig. 1A). Although micrococcal nuclease has been used for many nucleosome phasing studies, CAG repeat tracts do not digest well with micrococcal nuclease, and this endonuclease could not be used in our studies (supplemental Fig. 3). Instead, we employed the restriction enzyme Fnu4HI that recognizes the pentanucleotide GCNGC. Fnu4HI efficiently digests the CAG repeat tract in naked DNA and does not appear to cleave preferentially at any site within the tract (Fig. 1, B and C). Digestion of chromatin with Fnu4HI reveals four hypersensitive sites, two at the ends of the repeat tract ( Fig. 1, B and C, sites a and d) and two within the repeat tract ( Fig. 1, B and C, sites b and c). The hypersensitivity is striking because DNA bound by core histone octamers is refractory to restriction enzyme digestion (24,25). We calculated that these sites are ϳ125 bp (Ϯ15 bp) apart. The 125-bp sepa-ration is distinctly shorter than the 165-bp repeat distance of canonical yeast nucleosomes (26,27). A shorter tract of 78 repeat units (234 bp) in strain CAG325-78 yields a pattern with hypersensitive sites 110 bp (Ϯ10 bp) apart, two near the edges of the repeat tract and one inside the tract (Fig. 1D). (The hypersensitive site at the distal end of the repeat tract was confirmed by probing from the opposite direction.) To confirm that this unusual pattern of digestion is not specific to the chromosomal locus, we repeated the analysis with a CAG repeat tract placed at another locus. This was important because the placement of the CAG repeat tract in strain CAG325 is adjacent to ARS1, and this replication origin is known to establish a pattern of phased nucleosomes that might affect the chromatin organization of the adjacent CAG repeat tract (28,29). Consequently, we performed IEL analysis with chromatin isolated from strain CAG101 in which a CAG repeat tract is . Chromatin was digested with increasing amounts of Fnu4HI followed by DNA purification, and the DNA was then digested with either EcoRI (B) or EcoRV (C) and subjected to electrophoresis and blotting, and the blot was hybridized with either the RI probe (B) or the RV probe (C ). The position of the CAG repeat tract in the blot is marked with a thin line. An additional band at the bottom of B is from an Fnu4HI site located outside of the CAG repeat tract. D, Fnu4HI-hypersensitive sites in the chromatin of a CAG repeat tract of 78 repeat units in strain CAG325-78. Southern blot analysis was performed as described above with EcoRI digestion and the RI probe. This blot was also probed with the RV probe to confirm the presence of a hypersensitive site at the distal end of the repeat tract. E, IEL analysis of chromatin of a CAG repeat tract of 78 repeat units embedded at aro2::ADE2 locus in chromosome VII of strain CAG101. Fnu4HI digestion and Southern blot analysis were performed as described above except that purified DNA was digested with BanI and probed with probe BI (not shown). embedded within ADE2 and placed at ARO2 in chromosome VII (17). In this strain the CAG repeat tract of 78 repeat units is located greater than 1 kb from a replication origin (30). Digestion of chromatin from this strain with Fnu4HI generates three hypersensitive sites. Two are close to the edges of the repeat tract and the other is ϳ110 bp (Ϯ10 bp) from the first site (Fig. 1E). The CAG repeat chromatin at the aro2::ADE2 locus has an almost identical pattern of Fnu4HI-hypersensitive sites as the pattern generated by the repeat tract of 78 repeat units adjacent to ARS1 (Fig. 1D). These results show that the Fnu4HI digestion pattern is not locus-specific and that CAG repeat tracts are not likely to be condensed by canonical nucleosomes composed of histone octamer cores.
A Yeast One-hybrid Screen Identifies Hmo1 as a CAG Repeat Tractbinding Protein-Based on the noncanonical endonuclease digestion pattern, we speculated that proteins other than (or in addition to) core histones bind, either directly or indirectly, to CAG repeat tracts and contribute to the unusual chromatin organization. To identify specific proteins, we performed a yeast one-hybrid screen (31). For this study, we modified a standard yeast two-hybrid strain (labeled Y2H) by substituting a CAG repeat tract of 130 repeat units for the upstream-activating elements of the GAL promoter in front of two reporter genes, ADE2 and lacZ ( Fig. 2A) (15). This one-hybrid strain (labeled Y1H) was transformed with yeast genomic libraries Y2H-C1, Y2H-C2, and Y2H-C3 previously prepared for two-hybrid analyses in which yeast genes were fused to the GAL activation domain (AD) (15). Of 2 ϫ 10 5 transformants, 72 were both Ade2 ϩ and positive for ␤-galactosidase activity. Upon retesting, the clones that gave the strongest signals contained the entire sequence of HMO1 fused to the GAL activation domain by a bridge of 10 amino acid residues. HMO1 encodes a high mobility group box protein in yeast that binds to single and double strand DNA (32,33).
To confirm that reporter expression results from Hmo1 association with the CAG repeat tract, and not to cryptic promoters near the reporter genes, we transformed the HMO1-AD plasmid into the original yeast two-hybrid strain, containing the GAL promoters in front of the two reporter genes. If reporter expression in Y1H results from the Hmo1-AD fusion protein binding to the CAG repeat tract, neither reporter gene should be expressed in the Y2H strain. Indeed, Y2H transformed with the HMO1-AD plasmid fails to grow, whereas the Y1H strain grows when transformed with the same plasmid (Fig. 2B). The vector containing the GAL AD alone does not support growth of either strain (Fig. 2B). Both Y1H and Y2H transformed with either plasmid grow on leucine-deficient media as expected (Fig. 2C). These results support the conclusion that Hmo1 promotes reporter expression in the Y1H strain by binding directly to the CAG repeat tracts or to proteins that themselves bind directly.
CAG Repeat Tracts Can Be Precipitated by Antibodies Directed against Epitope-tagged Hmo1-To confirm the interaction of Hmo1 with CAG repeat tracts, we carried out ChIP (Fig. 3). We tagged Hmo1 with three FLAG epitopes at the C-terminal end of HMO1 (Hmo1-FL) and expressed the tagged protein in an hmo1 deletion strain (CAG325-126 hmo1⌬). Hmo1-FL is functional because its expression rescues the retarded growth phenotype of the hmo1 deletion strain (data not shown (32)). When we subjected Hmo1-FL-expressing cells that had been treated with formaldehyde to ChIP using a FLAG-specific antibody, CAG repeat DNA was precipitated (Fig. 3B). To show the specificity of the precipitation and the affinity of the interaction, we tested two other genomic sequences. We used a portion of ADE2 that has GC content of 44.5% and a portion of SPC25 that has a 56% GC content. Neither sequence precipitates as efficiently as CAG repeat DNA. These comparisons are justified because we carried out PCRs using two dilutions of template DNA and a limited number of PCR cycles such that the amount of PCR product is proportional to the amount of template precipitated by the antibody. Additionally, when wild-type cells lacking Hmo1-FL were subjected to ChIP, none of the fragments precipitated, showing the necessity and specificity of the antibody precipitation.
The affinity of the interaction between Hmo1 and the CAG repeat tract is further evidenced by coprecipitation from cells that had not been  To detect sequences by PCR, template DNA was diluted to 0.025% (1ϫ) or 0.05% (2ϫ) of the starting DNA for input (input) and 2.5% (1ϫ) or 5% (2ϫ) of starting DNA for the precipitated samples (ChIP and mock). Twice as much PCR product of CAG repeat DNA was subjected to agarose gel electrophoresis than PCR products of SPC25 and ADE2. B, formaldehyde cross-linked chromatin and native chromatin were subjected to ChIP analysis. PCR conditions were the same as described for A. C, formaldehyde cross-linked chromatin prepared from an hmo1⌬/hmo1⌬ diploid. One copy of chromosome IV carries a repeat tract of 55 repeat units and one carries a 497-bp portion of ADE2 (see A and B) at the same NheI site adjacent to TRP1 ARS1. Primers originate in the TRP1 ARS1 sequences common to both chromosomes. D, formaldehyde cross-linked chromatin prepared from a strain that had both a CAG repeat tract adjacent to TRP1-ARS1 and a repeat tract embedded in ADE2 were subjected to ChIP. treated with the cross-linking agent formaldehyde (Fig. 3B). This noncross-linked chromatin was prepared in a solution with a moderate salt concentration (140 mM NaCl) and a high detergent concentration (1% Triton X-100) that should disrupt weak interactions. The ability to immunoprecipitate the CAG repeat tract without the need to cross-link the protein to the DNA indicates the high affinity that Hmo1 has for the CAG repeat tract.
To investigate further the affinity of the association, we immunoprecipitated Hmo1-FL from a hmo1⌬/hmo1⌬ diploid strain in which one copy of chromosome IV bears a CAG repeat tract of 55 repeat units within the NheI site adjacent to ARS1, and the other copy of chromosome IV harbors a 497-bp segment from ADE2 cloned into the same NheI site. A larger percentage of the chromosomal DNA containing the repeat tract is precipitated than the chromosomal DNA containing the ADE2 fragment (Fig. 3C).
Another indication of the affinity of Hmo1 is the ability to precipitate repeat tracts at different chromosomal locations and to short repeat tracts. The one-hybrid studies show that it binds either directly or indirectly to long repeat tracts embedded in the GAL2 and GAL7 promoter sequences in front of the ADE2 and lacZ reporters, respectively. In addition, precipitation of Hmo1 from chromatin prepared from a strain with a repeat tract of 78 repeat units located in the ARO2 locus on yeast chromosome VII also brings down the repeat tract (Fig. 3D). This shows that chromosomal context does not affect the affinity of Hmo1 for CAG repeat tracts.
To examine the issue of repeat tract length on Hmo1 affinity, we noted that two of our chromatin preparations from CAG325-126 contained a substantial portion of repeat tracts that had contracted during the growth of the culture, in one case to 15 repeat units and in the other to 30 repeat units (Fig. 4A). Neither appears to precipitate as well as the 126-repeat unit tract. Because long tracts are not as good PCR templates as shorter tracts and because antibody precipitation might favor longer tracts that have multiple copies of Hmo1 associated with them, we carried out a more systematic study with repeat tracts of 54, 35, and 26 repeat units (Fig. 4B). All three repeat tracts precipitate, showing that Hmo1 binds to tracts that comprise as few as 78 bp, shorter than the length of DNA wrapped around a histone octamer core. Thus, Hmo1 likely binds to tracts of different lengths and locations.
Deletion of HMO1 Alters the Chromatin Structure of CAG Repeat Tracts-Both the yeast one-hybrid screen and the ChIP analysis provide clear evidence that Hmo1 interacts with CAG repeat tracts. To determine whether Hmo1 is responsible for the unusual chromatin organization of long CAG repeat tracts that we detected by Fnu4HI digestion, we performed IEL on the CAG repeat chromatin from the derivative of CAG325-126 in which HMO1 had been deleted (Fig. 5A). The pattern of Fnu4HI-hypersensitive sites in the CAG repeat tract of the chromatin from the hmo1 deletion strain is strikingly different from that of the HMO1 strain (Fig. 5, A and B). Instead of two strong hypersensitive sites within the interior of the repeat tract, only one weak site appears. Furthermore, this interior site in the hmo1⌬ chromatin is not positioned at the location of either of the two sites in the HMO1 chromatin. This is best seen when the downstream RI probe is used for IEL ( Fig. 1A and Fig.  5A). The distance between the interior site and the sites at the borders is ϳ165 bp (Ϯ15 bp), which is consistent with the repeat distance of canonical yeast nucleosomes. Whether this altered pattern results from canonical histone octamer cores will require further examination. Whatever the chromatin state in the absence of Hmo1, the change in the Fnu4HI digestion pattern caused by its absence strongly supports its presence in CAG repeat chromatin.
To substantiate the specificity of the disruption of the Fnu4HI digestion pattern by deletion of HMO1, we carried out two additional control studies. First, we examined whether deletion of another potential CAG repeat binding protein would produce the same disruption. Human MSH2 has been shown to bind in vitro to a substrate generated by denaturing a CAG repeat tract and permitting it to reanneal (34). Consequently, we investigated CAG repeat chromatin in a derivative of CAG325-126 in which MSH2 had been deleted. CAG repeat tract chromatin isolated from the msh2 deletion strain was digested with Fnu4HI and analyzed using IEL analysis (Fig. 5C). Deletion of MSH2 does not alter the pattern of Fnu4HI-hypersensitive sites in CAG repeat chroma-  tin. This result suggests that not all potential CAG repeat binding proteins affect the chromatin organization of the repeat tracts in vivo.
Second, we investigated whether deletion of HMO1 affects the Fnu4HI sensitivity of chromatin from a GC-rich sequence. The 514-bp segment of SPC25 previously used in the ChIP analyses (Fig. 3, B and C) was used for these analyses. In addition to having a GC content of 56%, it also possesses 11 Fnu4HI recognition sites (Fig. 6A). Digestion of naked SPC25 DNA with Fnu4HI generates a digestion pattern without any strong positional or sequence preference (Fig. 6, B and C). Digestion of the SPC25 chromatin from the HMO1 strain shows that sites a-c are well protected; sites i-k are hypersensitive, and sites d-h have an intermediary sensitivity to Fnu4HI digestion (Fig. 6, B and C). The overall pattern of protection, hypersensitivity, and intermediary sensitivity is preserved in the hmo1⌬ mutant. Differences in relative sensitivity of sites in the region containing sites d-h between wild-type and mutant chromatin are apparent. This region is the most GC-rich (62%), and these small differences might reflect a weak interaction of the sequence with Hmo1 that is also evident in the weak signal generated in formaldehyde-treated samples of the ChIP analyses (Fig. 3, A and B). The similarities between wild-type and hmo1⌬ mutant SPC25 chromatin (Fig. 6, B and C), in contrast to the gross differences in CAG repeat chromatin (Fig. 5, A and B), argue that Hmo1 does not contribute as substantially to the chromatin organization of the GC-rich SPC25 segment as it does to the CAG repeat tract chromatin.

DISCUSSION
We have shown that an HMG protein binds to a CAG repeat tract embedded in a yeast chromosome. This demonstration is a striking example of the utility of HMG proteins in chromatin organization. Hmo1 is one of 10 HMG proteins in yeast. Although we do not know whether other yeast HMG proteins can bind to the repeat tract in the absence of Hmo1, the nuclease digestion pattern of the repeat tract chromatin in the hmo1⌬ strain suggests that in the absence of Hmo1 the repeat tract has canonical histone octamer cores bound to it. Yeasts, as all other eukaryotes, possess the most highly conserved proteins in nature, the core histones, and our studies show that yeast cells have a protein that can bind preferentially to CAG repeat tracts, seemingly displacing some or all of the core histones.
The formal possibility exits that Hmo1 does not bind directly to the DNA but is tightly bound to histones or other tightly bound proteins that are bound to the CAG repeat tract. This possibility seems remote for a number of reasons. First, Hmo1 has two DNA binding domains and binds to both single strand and double strand DNA (33). Second, Hmo1 has never been found to interact with the core histones. Third, we precipitated the CAG repeat tract using antibody directed against the epitope attached to Hmo1 without benefit of formaldehyde treatment. Fourth, the observation that the CAG repeat chromatin becomes more refractory to Fnu4HI digestion when HMO1 is deleted means that either it binds directly or drastically alters the configuration of the proteins to which it binds (Fig. 5, A and B). Although the question of direct binding is important, it does not diminish the conclusion that Hmo1 strongly associates with CAG repeat chromatin and establishes its sensitivity to Fnu4HI digestion.
The widespread presence of HMG box proteins in nature means that our results should have broader implications. Hmo1 has both a B box and an A box and is similar to HMGB1/2 in humans (32,33). Some HMG proteins exhibit sequence or structure specificity including HMGB1 and -2 that bind to poly(CA) in a four-stranded complex and (GGA/TCC) 11 in a triple-stranded complex (35,36). More notably, the HMG box protein SRY binds tightly to a CA repeat tract and weakly to a CAG repeat tract (37). Thus, HMG box proteins might organize CAG repeat tract chromatin in humans.
Although studies using purified core histones have shown that they preferentially reconstitute on CAG repeat tract DNA (8 -12), some studies indicate that the binding may be anomalous. In particular, the studies by Godde and Wolfe (9) show that only six repeat units are needed to effect binding and that the affinity of binding does not increase with repeat length, at least for repeat tracts that can accommodate a single octamer core. Our studies have been done in vivo, where core histones must compete with other DNA-binding proteins. Whether Hmo1 substitutes for core histones or augments their binding in the formation CAG repeat chromatin is uncertain. HMG proteins are thought either to bind adjacent to core histone octamers or to the DNA wrapped around the cores (38,39). When HMG proteins bind adjacent to histone octamer cores, they increase the amount of DNA protected from nuclease digestion (40,41). In the case of CAG repeat tract chromatin, the length of DNA protected in the chromatin is shorter than that protected by normal nucleosomes, suggesting that they are unlikely to serve as linkers to octamer cores. We look to the chromatin organization of the ribosomal DNA promoter in yeast as a possible model for the organization of the CAG repeat chromatin. This promoter has H3 and H4 bound to it but appears devoid of H2A and H2B (42). Further work has shown that Hmo1 is needed for RNA polymerase I transcription of ribosomal DNA suggesting that Hmo1 may have replaced H2A and H2B (43). Thus, we present a model in which Hmo1 replaces histones H2A and H2B in the complex of proteins that condenses the CAG repeat tract (Fig. 7). Possibly, Hmo1 serves as a linker between tetramer cores of histones H3 and H4 that have been shown in reconstitution studies to provide protection of 120 bp of DNA (44).  A potential model showing that Hmo1 may bind adjacent to a tetramer of core histones H3 and H4. Our working model shows that H2A and H2B are displaced from CAG repeat tract chromatin by Hmo1. This is consistent with the evidence that the yeast ribosomal DNA promoter appears devoid of H2A and H2B (42) and that the necessity of Hmo1 for promotion (43) suggests its binding to the abbreviated cores.
Based on the correlation in the difference in affinity of core histones for pure and interrupted repeat tracts with the difference in their respective genetic stability, the postulate has been made that chromatin organization contributes to CAG repeat tract instability (8). We tested whether the alteration in the chromatin organization that we observed in the hmo1⌬ strain affects repeat tract stability in our tester strain (17). We did not find greater instability in the hmo1⌬ strain than in our wild-type strain. 3 The negative result probably reflects the fact that chromatin proteins are displaced from the DNA as the replication fork passes and that the excess DNA synthesis needed to expand repeat tracts is the fault of enzymes involved in the dynamic processes of replication, particularly those that are involved in Okazaki fragment joining (45-49). By having shown that CAG repeat tract chromatin is based on the binding of an HMG box protein, we will be curious to learn whether all, or a subset, core histones bind with Hmo1 on long CAG repeat tracts.