INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The highly conserved heterochromatin protein 1 (HP1) class of chromobox genes (HP1) encode structural adapters whose probable role is to assemble a variety of macromolecular complexes in chromatin (30). The possible functions of these complexes are wide-ranging and include roles in transcriptional repression (12, 36, 54, 55), transgene silencing (17, 26), chromosome segregation (14, 31), recruitment of silent genes to heterochromatin (7, 54), localization of heterochromatin to the nuclear periphery (67), and sex chromosome inactivation during mammalian spermatogenesis (44).

The swi6 gene in Schizosaccharomyces pombe is a nonessential gene that is required for the recombination-suppression and silencing which encompasses the mat2-K-mat3 region (33). Cloning of the gene showed that swi6 is a member of the HP1 class of chromobox genes (38), suggesting that the recombination-suppression and silencing are due to the packaging of the mat2-K-mat3 region into a heterochromatin-like complex that renders the region inaccessible to the transcriptional and recombination machinery (38, 63). Other trans-acting factors that are required for repression at the silent loci include rik1, clr1, clr2, clr3, clr4, and clr6 (34, 64). rik1, clr1, and clr4 are thought to encode structural components of the heterochromatin-like complex, while clr3 and clr6 share considerable homology with histone deacetylases (20). Along with the silent mating-type loci, swi6p is also involved in silencing at the fission yeast centromeres and telomeres (14, 47) and plays a role in chromosome segregation at anaphase (14).

HP1 proteins are characterized by the possession of both a classical chromodomain (CD) and a chromo shadow domain (CSD) (2) linked by a variable intervening region (IVR) or "hinge" (16). In addition, a stretch of acidic amino acids immediately precedes the CD of HP1 proteins (see Fig. 1A). The solution structure of the CD from the murine HP1-like heterochromatin-associated protein, M31 (also known as mHP1 and MOD1) (58, 65), has been elucidated by nuclear magnetic resonance (NMR) and shown to consist of an N-terminal three-stranded anti-parallel -sheet which packs against a C-terminal -helix (4). The most highly conserved residues are contained within the hydrophobic core of the CD structure, and most of these conserved residues are to be found at the bottom of a hydrophobic groove on the surface of the -pleated sheets under the -helix.

The precise role of CD proteins in the assembly of heterochromatin-like complexes is unknown, although a series of experiments involving deletions and a few natural and experimentally induced point mutations within the archetypal Drosophila chromodomain proteins, HP1 and Polycomb (Pc), have shown that the CD is crucial for function (18, 41, 50, 51). Moreover, chromodomain-swap experiments have shown that CD-containing proteins are likely to be directed to specific sites within the genome through a CD-CD interaction (50). Little is known about the stoichiometry of CD proteins in heterochromatin complexes, although a model of heterochromatin assembly, the mass-action model (62), based on the sensitivity of variegating phenotypes to changes in the dosage of structural components of heterochromatin, such as HP1, suggests that oligomers of such structural components may be incorporated into heterochromatin(-like) complexes.

In this study we identify the functional domains of swi6p and demonstrate that the M31 CD can functionally replace the CD of swi6p, showing that classical CD function is conserved from yeasts to humans. Given this conservation, we have undertaken a mutational analysis of the conserved amino acids in the swi6p CD. Using three functional assays, that measure silencing at the mating-type loci and chromosome segregation, we show that the hydrophobic amino acids which lie in the core of the CD are the most important for proper function. We also show that HP1 proteins can self-associate and can form oligomers in vitro. We suggest that it is as oligomers that HP1 proteins are incorporated into heterochromatin complexes that silence gene activity.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of strains used. Three crosses were required to produce the strain used for the chromosome loss rate assay. Random spore analysis was used throughout (42). The wild-type (wt) S. pombe strain SP11 (h ade6-704 ura4D-18 leu1-32) (11) was crossed with swi6 null strain AL91 (h90 swi6::ura4⁺ ura4-D18 ade6-M216) (38) to obtain AL91L (h90 Swi6::ura4⁺ ura4-D18 leu1-32 ade6-704), which produced a strain that contained both the swi6 null mutation and the ade6-704 mutation. Second, strain CN2 (h⁺ leu1-32 ade6-704) carrying the minichromosome Ch10 (sup3-5) (60) was crossed with SP11 to obtain strain SPCN3 (h⁺ leu1-32 ura4-D18 ade6-704/Ch10::sup3-5). SPCN3 contains both the ade6-704 mutation and its suppressor, sup3-5, which is carried on the minichromosome, Ch10. Finally, the swi6 null mutation was introduced into SPCN3 by crossing SPCN3 with AL91L to obtain strain CN91L (h90 leu1-32 swi6::ura4-D18 ade6-704/Ch10::sup3-5). CN91L was used to determine the minichromosome loss rate (MCLR) associated with each of the swi6 mutations used in this study (see below). Deletion of swi6 in CN91L was confirmed by PCR and Southern blotting (42).

S. pombe media. Media were essentially as described earlier (42). For chromosome loss tests, Edinburgh minimal medium (EMM) plates were supplemented with 0.15 the normal amount of adenine (12 mg/liter).

Site-directed mutagenesis and construction of yeast expression plasmids. swi6 wt cDNA was cloned to pGEX (Promega) vector by reverse transcription-PCR and confirmed by sequencing. The fission yeast expression vector pREP81 (40) was used for expression of the wt and mutant swi6p proteins in AL91L and CN91L strains. This vector utilizes the thiamine-repressible yeast nmt promoter (39) with a mutated TATA box and gives one level of expression in transformed null cells equivalent to the swi6p level in the wt strain SP557. For nuclear localization experiments, the higher-level expression vector pREP1 was also employed. All swi6 mutants were generated by PCR and PCR extension (27). pGEX-swi6 wild type was used as a template. Taq polymerase was purchased from Boehringer.

(i) Primers used to generate point mutations in swi6p. For all the primers listed below, the position in the swi6 or M31 cDNA sequence, in relation to the A of the initiation AUG codon, of the 3' nucleotide is indicated. For all point mutations the external primers were the 5'-ApaI primer GAAGGGCCCAATGAAGAAAGGAGGTGTTCG²⁰ and the 3'-BamHI primer CACGGATCCATTATTCATTTTCACGGA⁹⁷⁰. The inner primers for each point mutation were, for, E74-D79E80 to A74-80, the 5'-3' primer GGAGGAGCAGCAGCAGCAGCG-GCTGCATATGTTGTA²⁴⁹ and the 3'-5' primer TGCTGCTGCTGCTCCTCCTTCTTCTTC²⁰⁷; for E74-D79E80 to R74-80, the 5'-3' primer (GGAGACGGCGAAGATATGTTGTAGAAAAG²⁵⁵ and the 3'-5' primer TTCGCCGTCTCCTTCTTCTTCCTCCTTCT²⁰⁶; for the NTW(113-115) deletion, the 5'-3' primer AGTGATAGTTCAGAAGCCGATT³⁶¹ and the 3'-5' primer TGAACTATCACTGGGATCGTC³²²; for W115 to G, the 5'-3' primer ATAATACAGGGAGTTCAGAAGCCG³⁵⁸ and the 3'-5' primer CCCTGTATTGTCACTGGGATCGTC³²²; for Y100Y107 to CC, the 5'-3' primer TTGAAATGGGAAGGTTRTGAC³²⁴ and the 3'-5' primer TCCCCATTTCAAAAGGYATCC²⁹⁵; for W293 to G, the 5'-3' primer ACTKGGRAGAACGGTGCAATAT⁸⁹⁵ and the 3'-5' primer GTTCTYCCMAGTCAGGTAAATT⁸⁶⁵; for K103W104 to VV, the 5'-3' primer TTGGTAGTGGAAGGTTATGACGAT³²⁷ and the 3'-5' primer TTCCACTACCAAAAG-GTATTCATAGC²⁹⁰; for E84 to F, the 5'-3' primer TTTAAGGTTTTAAAACACCGT²⁷⁰ and the 3'-5' primer TAAAACCTTAAATACAACATATTC³⁰⁰; for L101L102 to NN, the 5'-3' primer TACAACAACAAATGGGAAGGTTATGAC³²⁴ and the 3'-5' primer TTTGTTGTTGTATTCATAGCCTCCAC²⁸⁴; for M91A92 to VV, the 5'-3' primer GTCTAGTAAGAAAAGGTGGAGGC²⁹¹ and the 3'-5' primer CCTTTTCTTACTAGACGGTGTTTT²⁶⁰; for K103 to Q, the 5'-3' primer TTGCAATGGGAAGGTTATGACGAT³²⁷ and the 3'-5' primer TTCCCATTGCAAAAGGTAT-TCATAGC²⁹⁰; and for M91A92 to QG, the 5'-3' primer CCGTCAGGGGAGAAAAGGTGGAGGC²⁹¹ and the 3'-5' primer TTTTCTGGGCTGACGGTGTTTTA²⁵⁹.

(ii) Primers used for generation of deletion mutants. The primers for the generation of deletion mutants were: for the C260-258 deletion, the 3'-5' primer CGGGATCCTAGCTAGCCGTCAGCTCTCTGTTGTC⁷⁶⁰; for the M164-235 deletion, the 5'-3' primer TCTCGTCCTAGCAATGTTACTCC⁷²² and the 3'-5' primer GCTAGGACGAGA-CTTTGGTGAAG⁴⁷³; for the N1-73 deletion, the 5'-3' primer GACGGGCCCATGG-AAGAAGAAGAAGAGGAT¹³⁶; for the N1-135 deletion, the 5'-3' primer GCGGGCCCATATGATAGCTAGCGGAGGAAGACCAGAACC⁴²²; for the C 313-328 deletion, the 3'-5' BglII primer GGAGATCTTACTGAGGACATTTTTTATTG⁹¹⁸; and for the M141-252 deletion, the 5'-3' primer GAAGACCAGAACCGGACAACAGAGAG⁷⁷¹ and the 3'-5' primer GGTTCTGGTCTTCCTC⁴⁰⁷.

(iii) swi6p-M31 domain-swap primers. The Swi6p-M31 domain-swap primers were the M31 cDNA 5'-3' XhoI primer GCACTCGAGATGGGGAAAAAGCAAAAC¹⁸, the M31 3'-5' BamHI primer GTGGATCCGAAGGCTGTGGGTTGTGG, the M31 Chromobox I 3'-5' NheI primer ATCAGAGCTAGCCCATTTCATCAGGAA⁴⁰⁰, and the M31 Chromobox II 5'-3' NheI primer AAGGCTAGCAAAGAAGAGTCAGAAAAG²³⁸.

Transformation of S. pombe, iodine staining, and calculation of percentage of normal asci. The pREP81 expression plasmids containing different swi6 mutants were transformed into S. pombe protoplasts according to standard procedures (5). The transformants were incubated at 30°C on Ura and Leu EMM plates containing 1.2 M sorbitol and became visible after 3 to 4 days.

Nuclear localization and subnuclear distribution of Swi6 wt and mutants. The RSGFP4 (9) plasmid which encodes the red-shift green fluorescent protein (GFP; 714 bp, 238 amino acids) was fused to swi6p wt and mutant proteins in order to investigate nuclear localization of the proteins. The RSGFP4 open reading frame was amplified by PCR and inserted into pREP1-swi6 constructs. For N-terminal GFP fusions, i.e., GFP-swi6p wt and mutants, the RSGFP4 was amplified by PCR using the two following primers: GFP5'-3' XhoI primer (CCGCTCGAGATGAGTAAAGGAGAAGAAC) and GFP3'-5' ApaI primer I (TCGGGGCCCTTGTATAGTTCATCCATG). For construction of GFP-N1-135 fusion, the primer was the GFP3'-5' ApaI primer II (GAGGGCCCATTTGTATAGTTCATCCAT). For construction of the GFP-C260-328 fusion, the primer used was the GFP5'-3' NheI primer (ATAGCTAGCATGAGTAAAGGAGAAGAAC). For construction of the swi6wt-GFP (swi6-CG) fusion, the primer used was the swi6 3'-5' XhoI-BamHI primer (GCGGATCCCTCGAGCTCATTTTCACGGAACG⁹⁶⁸). For construction of GFP-N1-138, the primer used was the swi6 NLS 3'-5' XhoI primer (CGTGCTCGAGTTCATCAGTTTTAC⁴⁸⁸). For the GFP-236-328 fusion, the primer used was the swi6 L6 5'-3' ApaI primer (GTGGGCCCTAGCAATGTTACTC⁷²¹). For the GFP-M31 cDNA fusion, the primer used was the GFP3'-5' SalI primer (CGTCGACTTTGTATAGTTCATCCATG). For the construction of the GFP-IVR fusion, the primers used were the IVR 5'-3' primer (GCGGGCCCAGGAGGAAGACCAGAAC⁴²¹) and the 260-328 deletion primer described above. Two further primers were used to verify the sequence at the junctions of swi6p and GFP: GFP69 3'-5' primer (ATCACCATCTAATTCAACAAG) and GFP556 5'-3' primer (AACTAGCAGACCATTATCAAC).

MCLR assay. We used the CN91L strain for the MCLR assay (see above for construction). The CN91L strain (h90 leu1-32 swi6::ura4-D18 ade6-704/Ch10::sup3-5) contains the swi6 null mutation, the ade6-704 mutation, and its suppressor, sup3-5, which is carried on the minichromosome, Ch10. When CN91L cells are cultured on EMM-0.15 M adenine plates, CN91L cells that retain the minichromosome give rise to white colonies, while loss of the minichromosomes gives rise to red colonies (3, 60). We used this MCLR assay to measure the activity of the swi6 mutants generated in this study.

Production of monoclonal antibodies. The swi6 coding sequence was cloned into pET25b (Novagen) in frame with a C-terminal herpes simplex virus (HSV)-His₆ tag. The resulting swi6-HSV-His tag fusion protein was expressed and purified according to standard protocols (Novagen). Monoclonal antibodies were raised in a female F344 rat using the Y3Ag1.2.3 fusion partner (19). Hybridomas were screened by enzyme-linked immunosorbent assay using swi6p-His tag as the antigen. One monoclonal antibody, MAC391, that specifically recognized swi6p was used in this study.

Western blotting. For Western blots of swi6 wt and mutants, yeast transformants were cultured in 20 ml of EMM selection medium overnight at 30°C (optical density of ca. 0.2 to 0.3). Cells were collected by centrifugation, washed once with phosphate-buffered saline, and resuspended at a concentration of 10⁹ cells/ml in homogenization buffer (10 mM Tris, pH 7.5; protease inhibitor cocktail [Boehringer]; 1 mM phenylmethylsulfonyl fluoride). The cells were lysed by vortexing vigorously for 1 to 2 min with an equal volume of glass beads, followed by incubation for 5 to 10 min on ice. This was repeated five to eight times. Lysates were centrifuged at 5,000 rpm for 5 min, and the supernatants were transferred to fresh tubes. Then, 40 µg of total protein from each sample was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) prior to electroblotting onto nitrocellulose filters. Blots were probed with monoclonal antibody MAC391 at a concentration of 1 µg/ml, and antibody binding was detected using a horseradish peroxidase (HRP)-coupled sheep anti-rat second-stage antiserum (Sigma) by using enhanced chemiluminescence (ECL; Amersham).

Gel overlay. Yeast cell lysates (40 µg of protein) from AL91L, SP556, and SP557 were separated by SDS-10% PAGE and electroblotted onto nitrocellulose. After blocking overnight at 4°C, the blot was overlaid with swi6-HSV-His tag fusion protein at a concentration of 10 µg/ml at room temperature for 2 h. The wt Swi6-Swi6-HSV-His tag interaction was detected with mouse anti-HSV antibody, followed by HRP-conjugated anti-mouse immunoglobulin G (Sigma) and visualized by ECL (Amersham). As a positive control, MAC391 was used to show the presence of wt swi6p in the blot. As a negative control, the anti-HSV was used directly without prior overlay with swi6-HSV-His tag.

Purification and gel filtration of HSV-His tag proteins. Swi6 and M31 were each expressed in the Escherichia coli BL21 with an N-terminal His tag and purified on a nickel affinity column (Novagen). M31 testes was expressed from pET25b with a C-terminal HSV-His tag and purified similarly. For swi6p, 100 µg of purified fusion protein in was loaded onto a Superose 12 HR26/60 gel filtration column (Pharmacia) in 1.0 ml of bicarbonate buffer (0.1 M, pH 8.0, containing 0.5 M NaCl) and eluted with same buffer at 0.5 ml/min. For M31 and M31 testes, protein was loaded onto a Superdex 75 HR26/60 in 50 mM Tris (pH 7.5)-100 mM NaCl-2 mM EDTA at 1 ml/min and eluted with same buffer at 5 ml/min. Molecular mass markers from 12 to 150 kDa (Pharmacia) were used to generate a standard molecular mass curve, and the swi6p, M31 and M31testes molecular masses were calculated by using that curve.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Assays used to measure swi6p function. We tested the ability of a series of swi6 deletion and point mutants to complement the phenotype of a swi6 null mutant strain using three different assays. First, the sporulation phenotype of all strains was determined by exposure to iodine vapors, which detects a starch-like material that accumulates in spores (6). The amount of staining reflects the sporulation frequency, roughly indicating the efficiency of mating-type switching to the opposite allele. Accordingly, the expression of plasmid-borne wt swi6 product (swi6p) in the null mutant strain gave rise to a black phenotype, while the null mutant, transformed with the expression vector alone, gave rise to much lighter, mottled colonies. Second, the sporulation of each strain was also measured microscopically (more than 1,000 cells scored) and is presented as the percentage of cells forming normal zygotic asci: haploid cells that simultaneously express all four mating-type proteins form azygotic spores (32), which can be readily identified by microscopic examination (Table 1). Finally, mitotic stability assays (3) were undertaken to measure the ability of swi6 mutants to maintain a linear minichromosome, Ch10 (sup3-5), during colony formation (48, 60).

Functional domains of the swi6p protein. A series of swi6p mutations containing specific deletions were constructed (Fig. 1B). These were then expressed at wt levels (Fig. 1C) on a swi6 null background to test their ability to complement the mutant phenotype (Table 1).

View larger version (35K): [in this window] [in a new window]   FIG. 1.   Activity of swi6p N-terminal, C-terminal, and internal deletion mutants. (A) Organization of swi6p. The black box represents the glutamic acid region. (B) swi6p deletion mutants. The activities of each mutant in sporulation assays and minichromosome loss experiments (see Table 1) are summarized on the right. Examples of iodine-stained colonies for each transformant are shown below. (C) Western blot showing expression of wt and deletion mutant swi6p. Except for lanes 1 and 2, all lanes correspond to swi6 deletion strain AL91L transformed with the described REP81-swi6p deletion mutants. Lanes: 1, strain AL91L; 2 and 7, strain SP557 (wt); 3, N1-135; 4, N1-73; 5, M164-235; 6, C313-328; 8, wt swi6p. The blot was probed with monoclonal antibody MAC391.

View larger version (34K): [in this window] [in a new window]   FIG. 2.   Nuclear and heterochromatin localization activities within swi6p. (A) GFP-swi6p constructs used in this study. The shaded ellipse represents the red-shifted GFP molecule fused to either the N-terminal or C-terminal end of wt swi6p or swi6p deletion mutants. N, nuclear distribution; C, cytoplasmic localization; ND, not determined. The iodine-staining phenotype of the N- and C-terminal GFP fusions is shown below. (B) Predicted nuclear localization motifs within the swi6p amino acid sequence. Longer square brackets, SV40 NLS-like seven-residue motifs; shorter brackets, four-residue motifs. (C) Summary of nuclear and heterochromatin binding data.

View larger version (76K): [in this window] [in a new window]   FIG. 3.   Subnuclear localization of swi6p-GFP fusion proteins in swi6 null cells. (A, C, E, G, I, and K) Phase-contrast images. (B, D, F, H, J, and L) GFP signal. Panels A and B show that the swi6-NG fusion localizes to specific subnuclear sites within the nucleus. Panels C and D show that swi6-CG fusion localizes to specific subnuclear sites with the nucleus. Panels E and F show that GFP-C260-328 fusion localizes to specific subnuclear sites with the nucleus. Panels G and H show that the GFP-N1-135 fusion localizes to the nucleus but does not localize to specific subnuclear sites within the nucleus. Panels I and J and show that the GFP-M164-235 fusion localizes to specific subnuclear sites with the nucleus. Panels K and L show that the GFP-IVR localizes to the nucleus but not to specific subnuclear sites.

The heterochromatin-binding domain of swi6p is in the N-terminal region. As a measure of heterochromatin binding, we compared the distribution of the swi6p-GFP constructs with the pattern observed for endogenous swi6p. swi6p localizes to subnuclear spots within interphase nuclei (14), which represent the blocks of heterochromatin associated with the chromosomes and can vary in number from one to three depending on associations that lead to larger chromocenters. In an initial experiment, we observed that both GFP-swi6p (swi6-NG) and swi6p-GFP (swi6-CG) (Fig. 2A and Fig. 3A to D) localized to the endogenous swi6p spots, confirming that fusion of swi6p to GFP per se does not disrupt subnuclear localization.

View larger version (134K): [in this window] [in a new window]   FIG. 4.   Localization of GFP-N1-135 in wt and swi6 null cells. (A and C) GFP signal. (B and D) Phase-contrast images. The scale bar in panel A represents 5 µM. Panels A and B show expression of the GFP-N1-135 fusion on a wt background. The fusion protein localizes to bright heterochromatic foci. Panels C and D show expression of the GFP-N1-135 fusion on the null mutant background. The fusion protein is distributed evenly throughout the nucleus.

A strong NLS resides in the IVR of swi6p. The observation that deletion mutants N1-135 and C260-328 both localize to the nucleus suggests that the IVR, which they share, contains a nuclear localization signal (NLS). This was confirmed directly by the nuclear localization of a further GFP construct, GFP-IVR (GFP-136-259) (Fig. 2A and Fig. 3K and L). The nuclear targeting function was more precisely defined by the GFP-M164-235 deletion, which also localizes to the nucleus (Fig. 2A and Fig. 3I and J), indicating that the NLS in the IVR resides either in the interval from positions 136 to 163 or in the interval from positions 236 to 259. Interestingly, the computer program PSORTII (28, 45) revealed that both of these regions contain matches to the four- and seven-residue nuclear localization motifs of the simian virus 40 (SV40) type (10). As shown in Fig. 2B, two overlapping seven- and four-residue motifs fall within the segment from positions 136 to 163 of swi6p, a finding consistent with this region having a strong nuclear localization function. In addition, one seven- and one overlapping four-residue pattern also exist between residues 241 and 247. We searched the rest of swi6p for other NLSs by introducing a GFP fusion in which the IVR was deleted (GFP-M141-252). We found that the GFP-M141-252 localized mainly, though not exclusively, to the nucleus (Fig. 2A; data not shown) supporting the existence of an additional, weaker NLS. An N-terminal construct, GFP-N1-143 (Fig. 2A; data not shown) localized to the cytoplasm, suggesting that this second nuclear localization activity lies C-terminal to residue 252. No matches to the consensus NLS sequences were found in this C-terminal region.

CD function is conserved from yeasts to humans. We have previously shown, using zooblots, that the CD motif is conserved in a variety of animal and plant species (58). We have directly tested the functional conservation of the CD through a domain-swap experiment in which the swi6p CD was replaced by the M31 CD (M31NT-swi6; Fig. 5A). When expressed at levels similar to that of endogenous swi6p (Fig. 5B), the chimeric protein could almost fully complement the swi6 null mutation (approximately 70 and 80% of the wt activity in asci formation and MCLR assays, respectively; Fig. 5 and Table 1). By contrast, when the CSD of swi6p was substituted by the M31 CSD (swi6-M31CT; Fig. 5A) only residual activity was obtained (Table 1). Expression of full-length M31 also failed to complement the null phenotype (Fig. 5; Table 1), although GFP-M31 was found to localize to heterochromatin (data not shown). These data indicate that classical CD function is likely to be conserved across species, while the CSD appears to retain a species-specific function.

View larger version (43K): [in this window] [in a new window]   FIG. 5.   swi6p-M31 domain swap constructs. (A) Construction of domain-swap mutations. The activities of the wt and domain-swap constructs are summarized on the right, and the iodine-staining phenotypes of the mutants are shown below. (B) Western blot showing expression of wt and domain-swap swi6 mutants. Except for the wt swi6 strain SP557 (lane 1), all lanes correspond to swi6 deletion strain AL91L transformed with the following pREP81-constructs: lane 2, M31NT-swi6; lane 3, swi6-M31CT; and lane 4, wild-type M31. swi6p was detected with monoclonal antibody MAC391; swi6-M31CT and M31 were detected with anti-M31 monoclonal antibody MAC353 (65).

Modeling the swi6p CD on the three-dimensional structure of the M31 CD. Given the functional equivalence of the M31 and swi6p CDs and their high degree of sequence similarity, it is likely that they adopt a similar three-dimensional conformation. The solution structure of the M31 CD was recently solved and consists of a three-stranded anti-parallel -sheet folded against a C-terminal -helix (4) (see Fig. 6A and B). Residues contained in the -strands of the M31 CD are particularly well conserved with the equivalent residues in swi6p (73% identity and 89% similar residues; Fig. 6A). In contrast, residues corresponding to the turns between the -strands are less similar, and it is here that gaps must be introduced into the M31 sequence to maintain the alignment with swi6p (Fig. 6A). Given the above findings, we modeled the swi6p CD on the M31 CD NMR structure using the SWISS-MODEL and Swiss PDB-Viewer programs (23, 24) (Fig. 6B to D). The primary difference between the modeled swi6p structure and the M31 NMR structure is, as anticipated, in the turns (T₁ to T₃), which are longer in swi6p (Fig. 6A). We also note that the amino acids in swi6p that immediately follow the -pleated sheet favor the formation of an -helix, like M31, despite several differences between the swi6p and M31 sequences in this region.

View larger version (73K): [in this window] [in a new window]   FIG. 6.   Site-directed mutagenesis of the swi6p CD. (A) Alignment of the CDs of M31 and swi6p showing the secondary structure arrangement above. Up arrows, residues contributing to the hydrophobic core of the M31 structure (those with asterisks line the hydrophobic cleft on the surface of the -sheet) (4). Down arrows, site-directed mutants generated in this study (residues shown below represent substitutions). A white character on a black background indicates a mutation that retains activity; inactive substitutions are indicated by black letters. Colored circles, key for amino acid side chains shown in panel D. (B) M31 CD (residues 20 to 72) and modeled swi6p CD (residues 80 to 136) three-dimensional structure shown superimposed in "ribbon" format. In the turns T1 to T3 the swi6p chain is shown in green. (C) Modeled swi6p CD showing the side chains of residues where equivalent mutation in HP1 or Pc affects gene silencing. (D) swi6p mutations employed in this study. In the righthand part of the figure the CD has been rotated through 45° about a vertical axis. (E) Western blot showing expression of wt and site-directed mutants of swi6p. Except for wt swi6 strain SP557 (lane 1), all lanes correspond to swi6 deletion strain AL91L transformed with the following pREP81-constructs: lane 2, E74-E80R; lane 3, K103W104-VV; lane 4, N113-W115; lane 5, W115-G; lane 6, Y100Y107-CC; lane 7, AL91 (swi6 deletion strain); lane 8, E94-F; lane 9, K103-Q; lane 10, M91A92-VV; lane 11, L101L102-NN; lane 12, M91A92-QG; lane 13, W239-G; lane 14, E74-E80A; 15, wt swi6p.

Mutational analysis of the swi6p CD. Using the information obtained from the modeled swi6p CD, we undertook a mutational analysis of the conserved residues within and adjacent to the CD. The effects of the site-directed mutants on swi6p activity were measured by using the three functional assays described above (Table 1). Our initial experiment addressed the function of the stretch of negatively charged amino acids immediately adjacent to the CD (Fig. 1A). Previous studies with Drosophila HP1 (50) have shown that three of the six glutamic acid residues can be replaced by alanine, with no effect on function in vivo (E78 to E80 to A, using swi6p numbering; Table 2). However, we found that a more extensive replacement of all seven residues with either alanine (E74 to E80 to A) or arginine (E74 to E80 to R), abolished the ability of swi6p to complement the swi6 null phenotype (Tables 1 and 2).

swi6p self-association. When bacterially expressed swi6 fused to an N-terminal His tag (His-swi6p) was subjected to gel filtration chromatography under native conditions, the majority of the protein eluted in a peak corresponding to an apparent molecular mass of 164 kDa (Fig. 7A). This is almost exactly four times the predicted monomer molecular mass of His-swi6p (40 kDa). Similarly, when purified His-M31 was loaded on a gel filtration column the protein eluted with an apparent molecular mass of approximately 96 kDa (Fig. 7B), again a value almost exactly four times the predicted monomer molecular mass of the His-tagged fusion protein (26 kDa). We also determined the molecular mass of a naturally occurring isoform of M31, M31testes, which lacks the C-terminal 85 amino acid residues of full-length M31 (49). M31testes-HSV-His₆ eluted with a molecular mass of approximately 28 kDa (Fig. 7B), twice the predicted monomer mass (14 kDa). Dimerization of M31testes was also observed using electrospray ionization mass spectroscopy, in which the major molecular weight species of 28,083.36 ± 2.33 was observed in H₂O (pH 7.0). In dilute formic acid (denaturing buffer) the major M31testes species had a molecular weight of 14,042 ± 1.29, which is the monomer.

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Swi6p quaternary structure. (A) Superose 12 HR26/60 chromatogram of swi6p-HSV-His tag. Elution of markers is shown on the horizontal axis. (B) Superdex 75 HR26/60 FPLC profiles of M31-HSV-His tag and M31testes-HSV-His tag. (C) Gel overlay experiment. The cell lysates indicated were separated by SDS-PAGE and blotted onto a nitrocellulose membrane. Lanes 6 to 10 were incubated with HSV epitope-tagged swi6p before the membrane was probed with the antibodies shown (see Materials and Methods). AL91 and T2 are swi6 deletion strains, SP556 and SP557 are swi6 wt strains. Lane 10 contains purified recombinant swi6-HSV-His protein.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have undertaken a structure-function analysis of swi6p, an HP1-like protein in S. pombe (38). Like Drosophila HP1 (13) and M31 in mice (17, 65), swi6p is localized predominantly within heterochromatin domains which, in S. pombe, are found at the centromeres, silent mating-type loci, and telomeres. Swi6p is involved in silencing at these loci (14, 33) and, in addition, cells lacking swi6p exhibit an increased rate of chromosome loss (3), suggesting that swi6p, like HP1 in Drosophila (31), is required for efficient chromosome segregation. Thus, HP1-like proteins appear to be functionally as well as structurally conserved between fission yeast and metozoans.

Conservation of CD function. The close sequence similarity and functional conservation of the CDs of M31 and swi6p prompted us to model the three-dimensional structure of the swi6p CD on the recently solved M31 structure. We then used the modeled swi6p structure to complete a detailed structure-function analysis of the CD using the advantages of the yeast system. In particular, we have extended initial findings using Drosophila Pc and HP1, which identified residues within the CD that are important for function (41). Previously, mutations in three hydrophobic residues have been shown to disrupt function in Drosophila HP1 or Pc: Y24-F(HP1), V26-M (corresponding to the su(var)2-5⁰² mutation) (50), and I31-F (corresponding to the pc¹⁰⁶ mutation) (41). The V26-M (HP1) and I31-F (Pc) mutations map to residues V83 and V86 in the swi6p CD (Fig. 6C), both of which contribute to the hydrophobic core of modeled domain. HP1 Y24-F maps to the swi6p residue Y81, which sits at the edge of the hydrophobic groove formed between the -sheet and the -helix (Fig. 6C). While substitutions of amino acids in the hydrophobic core of the domain might be expected to affect its folding, the loss of function associated with the Y24-F might instead interfere with interactions of the CD with other proteins, as proposed by Ball et al. (4).

Functional organization of HP1 proteins. Our deletion analysis has revealed the functional organization of swi6p and has identified constraints on the organization of the molecule. Residues N-terminal of the acidic amino acid stretch are largely dispensable, but the CD and an intact CSD are necessary for activity (Fig. 1B; Table 1). Shortening of the IVR is compatible with activity, but some spacing between the CD and the CSD is required. Another constraint concerns the stretch of acidic residues adjacent to the N terminus of the CD. This typically consists of six or seven glutamic acid residues and is a characteristic of HP1 proteins (30). Negatively charged residues seem to be required at this point in the molecule, since substitution of the entire acidic stretch with alanine or lysine in swi6p results in a loss of activity (Tables 1 and 2). A smaller substitution of three of the residues with alanine in Drosophila HP1 does not seriously affect activity (50). The function of this stretch of negatively charged amino acids is unknown, although it has been suggested that this region might interact with the lysine-rich histone tails (58), whose positive charge can be regulated by acetylation (22).

Self-association of HP1 proteins. The notion that heterochromatin consists of a macromolecular protein complex is based on the characterization of variegating position effects (21, 25, 52). In a prescient mathematical model for the assembly of heterochromatin complexes, Tartof et al. suggested that oligomers of certain components would be incorporated into the complex (61). Consistent with the model, the data presented suggest that HP1 proteins can form oligomers through self-association. (i) Under nondenaturing conditions, swi6p and M31 both eluted from fast-performance liquid chromatography (FPLC) gel filtration columns with a molecular weight four times that of the respective monomers (Fig. 7A and B). (ii) The N-terminal 100 amino acid residues of M31 behave as a dimer as seen by mass spectroscopy under nondenaturing conditions and in gel filtration experiments (Fig. 7B). (iii) M31 has been shown to self-associate in yeast two-hybrid screens (35). (iv) Self-association of swi6p in vitro can also be demonstrated in gel overlay experiments (Fig. 7C).

View larger version (32K): [in this window] [in a new window]   FIG. 8.   Possible models for incorporation of swi6p and M31 into heterochromatin complexes. (A) Modular structure of swi6p. The CD and CSD are shown as a hatched rectangle or ellipse, respectively, and the hinge region or IVS joining them is shown as a black line. The acidic region adjacent to the CD is also indicated. (B) One possible oligomeric configuration of swi6p involving dimerization of monomers through the CSD and the association of dimers to form a tetramer. (C) Model for the organization of swi6p-containing heterochromatin-like complexes. The stoichiometry and spatial relationship of components within the complex are unknown. The thick gray line running through the complex represents nucleosomes deacetylated (arrows) by clr3p and clr6p. (D) A model for the recruitment by KRAB zinc-finger proteins of M31 into local heterochromatin-like complex. The stoichiometry and spatial relationship of components within this complex are also unknown.

Heterochromatin-like complexes as a conserved mechanism of gene regulation. We suggest that it is as oligomers that HP1 proteins are incorporated into heterochromatin-like complexes. Accordingly, we have modeled the incorporation of a swi6p oligomer into a heterochromatin-like complex that represses the donor mating-type loci (Fig. 8C). We envisage the swi6p oligomer as part of a core complex, that is, a repeating unit, similar to the repeating SIR complex in budding yeast that can assemble chromosomal DNA into a repressed domain (22). Oligomerization of swi6p is likely to be driven by the high local concentrations within heterochromatic foci (Fig. 3A and B). This may be analogous to the way that spectrin oligomers (43) are thought to be assembled at the red blood cell membrane from spectrin dimers and tetramers (37), a process that is promoted by high local concentrations and interactions with other proteins, e.g., ankyrin (43). Likewise, oligomerization of swi6p may be promoted by interaction with other constituents of the core complex, such as clr1p, clr4p, and rik1 (Fig. 8C): clr4p and rik1p are required for incorporation of swi6p into the heterochromatin complexes at the centromeres, telomeres, and the mat2-K-mat3 loci (15). Two further proteins that are required for repression at the silent mating-type loci, clr3p and clr6p, share considerable homology with histone deacetylases, suggesting that silencing of the donor loci involves histone deacetylation (20). This model for silencing of the donor mating-type loci is similar to that proposed for KRAB-ZFP (Krüppel-associated box-zinc finger protein)-mediated repression in mammals (Fig. 8D) (54) in which the KAP-1 corepressor recruits a heterochromatin-like complex through a KAP-1-CSD interaction (54). The recruited complex includes SUV39H1 (a human homologue of clr4p), which is immunoprecipitated as part of the same chromatin fraction as M31 (1). Recent work also suggests that assembly of a heterochromatin-like complex by KAP-1 is associated with histone deacetylation (46). This finding is consistent with the observation that CHD3 protein, which is a component of the NuRD deacetylase complex, interacts with KAP-1 (54).