The Ubiquitin-Dependent Targeting Pathway in Saccharomyces cerevisiae Plays a Critical Role in Multiple Chromatin Assembly Regulatory Steps

Corresponding author: Troy A. A. Harkness, College of Medicine, University of Saskatchewan, B313 Health Sciences Bldg., 107 Wiggins Rd., Saskatoon, Saskatchewan S7N 5E5, Canada., troy.harkness{at}usask.ca (E-mail)

Communicating editor: F. WINSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In a screen designed to isolate Saccharomyces cerevisiae strains defective for in vitro chromatin assembly, two temperature-sensitive (ts) mutants were obtained: rmc1 and rmc3 (remodeling of chromatin). Cloning of RMC1 and RMC3 revealed a broad role for the ubiquitin-dependent targeting cascade as the ubiquitin-protein ligases (E3s), the anaphase promoting complex (APC; RMC1 encodes APC5) and Rsp5p, respectively, were identified. Genetic studies linked the rmc1/apc5 chromatin assembly defect to APC function: rmc1/apc5 genetically interacted with apc9, apc10, and cdc26 mutants. Furthermore, phenotypes associated with the rmc1/apc5 allele were consistent with defects in chromatin metabolism and in APC function: (i) UV sensitivity, (ii) plasmid loss, (iii) accumulation of G2/M cells, and (iv) suppression of the ts defect by growth on glucose-free media and by expression of ubiquitin. On the other hand, the multifunctional E3, Rsp5p, was shown to be required for both in vitro and in vivo chromatin assembly, as well as for the proper transcriptional and translational control of at least histone H3. The finding that the distinctly different E3 enzymes, APC and Rsp5p, both play roles in regulating chromatin assembly highlight the depth of the regulatory networks at play. The significance of these findings will be discussed.

ALL life depends on exquisite mechanisms evolved by cells to decode, replicate, and segregate the information stored within chromosomes. The core structure of the chromosome, chromatin, is extremely dynamic and is crucial for the highly regulated temporal and morphological changes that chromosomes undergo through every round of growth. Chromatin is made up of 147 bp of DNA wrapped around an octameric protein complex containing two copies of each of the four histones, H2A, H2B, H3, and H4 (WOLFFE 1992 ; VERREAULT 2000 ). A range of evolutionarily conserved proteins function to deposit histones onto DNA and then to modify the DNA-bound histones to facilitate transcription, DNA repair, chromosome compaction, and cell-cycle progression (ITO et al. 1997 ; ADAMS and KAMAKAKA 1999 ; HASSAN et al. 2001 ; JENUWEIN and ALLIS 2001 ). Post-translational modification of histones through acetylation, phosphorylation, methylation, and ubiquitination is known to have far-reaching effects on many of these activities (GRUNSTEIN 1997 ; SAUVE et al. 1999 ; DE LA BARRE et al. 2000 ; REA et al. 2000 ; ROBZYK et al. 2000 ). However, very little is known regarding how these intracellular post-translational activities are regulated or how extracellular signals that control these processes are conveyed.

Modification of histones by ubiquitin is an important step for cell-cycle progression (ROBZYK et al. 2000 ). Ubiquitin is a small protein that is notorious for marking proteins that are to be degraded, whether via the proteosome (TYERS and JORGENSEN 2000 ) or via the vacuole (ROTIN et al. 2000 ). The attachment of ubiquitin to the -amino group of specific lysine residues in target molecules is catalyzed by up to three classes of enzymes: a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin-protein ligase (E3; CIECHANOVER 1998 ). Ubiquitin plays a key role in ordered cell-cycle progression as many proteins that block or activate crucial steps in this progression are degraded in a ubiquitin-dependent manner (KOEPP et al. 1999 ). However, several ubiquitin-targeted proteins, such as Met4p and Spt23p, are exceptions, as their ubiquitination does not result in degradation (HOPPE et al. 2000 ; KAISER et al.. 2000 ). Ubiquitination of Met4p negatively regulates its transcriptional activity without altering its stability (KAISER et al.. 2000 ). Alternatively, ubiquitination activates Spt23p through a proteolytic processing step that cleaves the C-terminal transmembrane domain of the inactive endoplasmic reticulum membrane-bound precursor, releasing an active transcription factor (HOPPE et al. 2000 ).

Two previous studies have linked ubiquitin to chromatin. In the first study, mutations to the SAGA (Spt/Ada/Gcn5 acetyltransferase) subunit, Spt3p, were suppressed when combined with a mutant version of the ubiquitin-protein ligase Rsp5p (B. BERG, A. HAPPEL and F. WINSTON, cited in HUIBREGTSE et al. 1995 ). SAGA is a large protein complex, which, among other functions, influences the activity of the histone acetyltransferase Gcn5p (EISENMANN et al. 1992 ; ROBERTS and WINSTON 1997 ). In the second study, ubiquitination of histone H2B by the ubiquitin-conjugating enzyme Rad6p was shown to be essential for proper progression through mitosis and meiosis (ROBZYK et al. 2000 ). How ubiquitin-mediated processes modulate chromatin structure through SAGA and Rad6p remains to be elucidated.

The results presented here shed light on possible intracellular mechanisms regulating chromatin assembly that involve the ubiquitin-dependent protein targeting pathway. In this report, we describe the characterization of two mutants obtained in a screen of yeast whole-cell extracts for those exhibiting in vitro chromatin assembly defects. Cloning of the genes responsible for the mutant phenotypes revealed the requirement for two ubiquitin-protein ligases (E3s), the anaphase promoting complex (APC) and Rsp5p, in the regulation of chromatin assembly. The dramatic structural and functional differences between APC and Rsp5p suggest distinct molecular mechanisms are at work to modulate chromatin assembly throughout the yeast life cycle.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains and methods:
Table 1 lists the yeast strains used. A364a was the parental strain of the temperature-sensitive (ts) library (HARTWELL 1967 ) from which the founding rmc (remodeling of chromatin) mutant, YTH335, was isolated. All strains derived from YTH335 were crossed to S288c or W303 wild-type strains 8–10 times.

The APC5::HIS3 (YTH1061) strain was constructed by PCR-based homologous integration. Primers were designed to produce a DNA fragment whereby the HIS3 gene, including 200 upstream nucleotides, could be integrated 100 bp downstream of APC5. The 5' HIS3 primer contained nucleotides -50 to -100 downstream of APC5, followed by 18 nucleotides derived from a sequence 200 nucleotides upstream of HIS3. The 3' HIS3 primer contained the last 18 nucleotides of HIS3 followed by nucleotides -100 to -150 downstream of APC5. Homologous integration of a fragment generated by amplifying YTH3 genomic DNA resulted in insertion of HIS3 100 bp downstream of the APC5 stop codon, as confirmed by PCR.

The protein A IgG-binding domain (PA) was fused to the 3' end of Apc5p in the rmc1 and wild-type strains using similar PCR-based techniques. The 5' primer was generated from the final 70 nucleotides of APC5, minus the stop codon, fused to 21 nucleotides of the PA epitope, including the factor X cleavage site. The 3' primer fused 21 nucleotides of the Schizosaccharomyces pombe his5⁺ promoter to 70 nucleotides derived from a sequence 100 nucleotides downstream of APC5. DNA from pBXAHis5⁺ (a generous gift from J. Aitchison), which encodes PA and S. pombe his5⁺, was used to amplify a fragment containing PA and his5⁺ flanked by APC5 sequences, which was transformed directly into YTH6 cells. Since we could not recover APC5-PA::his5⁺ transformants, the PCR-generated fragment was transformed into a wild-type strain carrying the rmc1 complementing plasmid pTH20 (see Table 2). This resulted in fusion of PA to the plasmid-borne APC5, creating pTH40. From this plasmid, larger regions of APC5 homology flanking PA and his5⁺ could be generated by PCR. Using 500 nucleotides of homology on each end of the fragment, the 3' end of endogenous APC5 in both wild-type and rmc1 cells was fused to DNA encoding PA, generating YTH1115 and YTH1117. All positives were identified and confirmed by PCR. The apc5^CA-PA::his5⁺ strain was used in all subsequent crossing experiments.

Media were prepared according to AUSUBEL et al. 1995 . Genetic techniques were performed as described (GUTHRIE and FINK 1991 ). Escherichia coli strains JM109 and DH10B were used to propagate DNA plasmids. DNA manipulations such as restriction enzyme digests, DNA purification, yeast and E. coli transformation, and yeast genomic DNA preparation have been described previously (AUSUBEL et al. 1995 ).

Growth rates were determined by inoculating 5 ml of YPD with 1 x 10⁴ cells from a fresh culture followed by incubation at various temperatures. Aliquots were removed at the indicated times and the cell density was determined at an optical density (OD) of 600 nm. Viability curves were generated by first diluting an overnight culture grown at 24° to an OD₆₀₀ of 0.1 followed by incubation at 37°. At various times, aliquots were removed and the cell density was determined. Equal numbers of cells were then plated on YPD plates and incubated at 24°. The total number of cells that grew after the 37° treatment was then compared to untreated cells. Typical results from a minimum of three repeats are presented for viability and growth curves. Chromatin assembly was assayed by assessing the ability of yeast whole-cell extracts to support a plasmid supercoiling reaction, as previously described (SCHULTZ et al. 1997 ).

Northern analysis:
Total RNA was extracted from yeast using the hot phenolic extraction method (Current Protocols). For Northern analysis, 30 µg of RNA was resolved by denaturing formaldehyde gel electrophoresis and transferred by capillary elution to nylon membranes (GeneScreen Plus, DuPont-New England Nuclear). Following elution the membranes were UV-crosslinked at 254 nm (1200 mW/cm²) using a Stratalinker 1800 (Stratagene, La Jolla, CA) to facilitate reprobing of the membranes. Histone H3 DNA was excised from topoH3 by EcoRI digestion and ACT1 DNA was excised from a pBlue Script-based plasmid by BamHI/HindIII digestion. The DNA probes were labeled with -³²P by the random priming method following the manufacturer's instructions (Boehringer Mannheim, Indianapolis). Membranes were prehybridized at 42° for 1 hr in buffer (UltraHyb, Ambion, Austin, TX) and hybridized with the DNA probes (1 x 10⁷ cpm) overnight at 42° in a rotary hybridization oven (VWR Canlab). Following hybridization, the membranes were washed with 2x SSPE (0.3 M NaCl; 20 mM NaH₂PO₄ · H₂O; 2 mM EDTA, pH 7.4), 0.1% SDS at 42° for 30 min with two changes of buffer, followed by two washes with 1x SSPE, 0.1% SDS for 30 min. The probed membranes were exposed to X-ray film (Kodak XAR) at -80° for 1 hr.

Western analysis:
Protein lysates were prepared from yeast using a Mini bead beater (BioSpec, Bartlesville, OK) according to the manufacturer's instructions. For Western analysis, lysates were subjected to Bradford protein assays and equivalent amounts of protein were resolved by SDS-PAGE and verified by Coomassie Blue staining. Following electrophoresis, proteins were transblotted onto nitrocellulose membranes. The membranes were blocked with 5% reagent-grade fat-free skim milk powder (Bio-Rad Laboratories, Richmond, CA) in PBS containing 0.1% (v/v) Tween 20. Primary and secondary antibody incubations and subsequent washes were carried out using this same buffer. Primary antisera to post-translationally unmodified histone H3 and actin were obtained from Upstate Biotechnology (Lake Placid, NY) and Sigma (St. Louis), respectively. Secondary horseradish-peroxidase-conjugated antibodies were purchased from Upstate Biotechnology. A chemiluminescent detection system (Perkin-Elmer Life Sciences) was used for identification of secondary antibody.

Mutant isolation:
Yeast whole-cell extracts were prepared using coffee mills (SCHULTZ et al. 1997 ) from 100 ts yeast strains (HARTWELL 1967 ) grown in rich media at room temperature to an OD₆₀₀ of 2–3, prior to culture at 37° for 2 hr. Screening using a plasmid supercoiling assay (SCHULTZ et al. 1997 ), with reactions performed at 30°, resulted in the identification of one strain yielding an assembly-deficient extract (YTH335). Variations in supercoiling efficiencies were sometimes observed between independently prepared extracts of the same strain and between independent reactions carried out with the same extracts. Therefore at least three extracts from each potential mutant were tested two to three times. An extract was deemed to be deficient if, in all three trials, the ratio of intermediate to supercoiled products indicated an accumulation of intermediate topoisomers (compared to wild type).

Genetic manipulation of mutants:
rmc mutants were backcrossed 8–10 times to S288c and W303 wild-type strains to obtain isogenic wild-type and mutant partners. In initial crosses, extracts were prepared from at least two complete tetrads to follow the segregation of the chromatin assembly defect. Backcrossing quickly revealed that YTH335 contained three mutations that influenced chromatin assembly: rmc1, rmc2, and rmc3.

Plasmid construction:
All plasmids used in this study are listed in Table 2. APC5 was cloned by internally deleting a 20-kb fragment isolated from a single copy screen that complemented the YTH239 ts phenotype. From 50,000 transformants that grew at 30°, 31 were capable of growth at 37°. Plasmids, referred to as class I, were isolated from 15 transformants that grew at wild-type levels at 37°. APC5 was present in each of these plasmids. Plasmids isolated from the remaining 16 transformants, referred to as class II, restored growth at 37° in rmc1 cells to variable, but suboptimal, levels of growth. Sequencing demonstrated that these plasmids carried 12 different inserts, likely representing single copy suppressors of the rmc1 phenotype. Preliminary subcloning experiments have revealed that PRP22, HSL7, RVS161, and RPT23 are responsible for the suboptimal suppression in four of the plasmids (V. RAMASWAMY, M. C. SCHULTZ and T. A. A. HARKNESS, unpublished data).

To clone the entire APC5 open reading frame (ORF), PCR primers designed to amplify from 300 bp upstream of APC5 to the stop codon were utilized. SalI and KpnI sites were added to the 5' and 3' primers, respectively. Correct clones were digested with SalI and KpnI and ligated into SalI-KpnI-digested YCplac111, creating pTH101. The rmc1 allele was cloned in a similar manner, yielding plasmid pTH102. All PCR products, amplified from YTH6 and YTH239, were subcloned into the pCR2.1-TOPO vector and sequenced.

RMC3 was cloned by internally deleting 8- and 11-kb overlapping fragments from plasmids pTH62 and pTH64, isolated from a single copy screen that complemented the YTH125 ts phenotype. In that screen, 48 colonies grew at 37° out of 50,000 tested, and 22 of the 48 isolated were found to contain four versions of the same overlapping sequence (see Table 2). The 20-kb fragment contained nine ORFs: YER123–YER131 on chromosome V. The smallest clones capable of complementation, pTH62 and pTH65, contained the entire RMC3/RSP5/YER125w gene. To clone RSP5 in isolation, PCR primers were designed to amplify from 300 bp upstream of RSP5 to the stop codon, with SalI and KpnI sites added to the 5' and 3' primers, respectively. All PCR products, amplified from YTH3 genomic DNA, were subcloned into the pCR2.1-TOPO vector and sequenced on both strands. Correct clones were digested with SalI and KpnI and ligated into SalI-KpnI-digested YCplac111, creating pTH210. The rmc3 allele was cloned in a similar manner, yielding plasmid pTH211. In this case, genomic DNA was isolated from YTH125 and used as a template in PCR reactions.

To clone the Rsp5p Cys Ala₇₇₇ active site mutant allele into the YCplac111 plasmid, an internal fragment was generated by PCR from pYESRSP5^C-A (a gift from J. Huibregtse) that contained the conserved Cys Ala₇₇₇ mutation. This fragment was first subcloned into the pCR2.1-TOPO vector and then into pTH210 using restriction sites within RSP5 to generate pTH212. Histone H3 was cloned by amplifying the H3 gene from genomic DNA followed by cloning of the PCR fragment into the pCR2.1-TOPO vector. All other plasmids were obtained from the sources listed (Table 2).

Sequencing of the apc5^CA and rsp5^P-S alleles:
Sequencing of plasmid inserts was performed using primers flanking the BamHI site within the tetracycline resistance gene in the plasmids YCp50 and YEp13. The sequence obtained was compared to the yeast genome database using the BLAST network search algorithm (ALTSCHUL et al. 1990 ).

Sequencing of the rmc1 APC5 allele was completed as follows. DNA was obtained by PCR amplification of genomic DNA isolated from YTH239 and YTH6. The primers were designed to amplify each half of the gene independently. The sequence of the primers was N-half, 5'-GCTAGGATCCAGTAAGTATGGTCCATTG-3' and 5'-AAGTATCTCGGAGCTCTAAA-3' and C-half, 5'-TTTAGAGCTCCGAGATACTT-3' and 5'-GCTAGGTACCTGCAGTTACATCTGAACATCCCT-3'. The amplified DNA products were purified using QIAGEN (Chatsworth, CA) columns. The results were confirmed by sequencing N-terminal PCR products of the mutant and wild-type genes generated using the N-half, 3' primer and the 5' primer (5'-GCTAGTCGACTGTCCCAGGCCGACTTC-3') designed to amplify 300 bp upstream of the APC5 start site.

A PCR-based strategy also was followed to obtain sequence from the rmc3 allele. DNA was obtained by PCR amplification of genomic DNA isolated from YTH125 and YTH3. Nine primers were designed to amplify and sequence the entire RSP5 gene.

Phenotypic analysis: Ultraviolet light irradiation:
To test for UV light sensitivity, the various isogenic strains were transformed with YCplac111 to leucine auxotrophy and grown at permissive temperatures overnight. The next day, cells were diluted to 1 x 10⁶ cells/ml and grown for an additional 24 hr at 37°. Cell concentrations were determined and a dilution series was plated on leucine-deficient media. Since rmc3 cells were found to have a survival rate of 1 x 10^-4 at 37° (Fig 1E), 10⁵, 10⁶, and 10⁷ cells were routinely plated following this treatment, compared to 10¹, 10², and 10³ cells for wild-type controls. The plates were then exposed to an increasing dosage of UV irradiation (0, 37, 72, and 108 J/m²). The plates were covered and grown in the dark for 3 days at room temperature prior to colony counting. Percentage viability was calculated as (number of colonies on exposed plates/number of colonies on unexposed plates) x 100. UV light sensitivity was the same for wild-type and mutant cells at the permissive temperature and when the cells were grown in YPD media (data not shown). The viability curves shown are a typical result from at least three experiments.

View larger version (57K): In this window In a new window Download PPT slide   Figure 1. Growth characteristics and chromatin assembly profiles of rmc mutants. (A) rmc1 cells are temperature sensitive. Isogenic yeast strains, YTH3 (WT), YTH457 (rmc1), YTH461 (rmc2), and YTH455 (rmc1 rmc2), were grown on YPD agar plates at either 30° or 37° for 3–5 days. (B) rmc1 cells are chromatin assembly defective. The same strains as above were processed for chromatin assembly reactions. The four strains were grown at 30° prior to a 2-hr shift to 37°. Extracts were then prepared and plasmid supercoiling was assayed using a 32P-labeled circular, relaxed plasmid. Increasing amounts of extract protein were assayed in the reactions. The products of the supercoiling assay were separated by agarose gel electrophoresis, after which the gel was dried and exposed to film. Inefficient chromatin assembly was observed by the accumulation of intermediate topoisomers (inter.) compared to fully supercoiled plasmids (sup.). The majority of the input plasmid was open, relaxed circular DNA (O, R). (C) rmc3 mutants are temperature sensitive. The isogenic strains YTH3, 457, and YTH520 (rmc3) were grown overnight at 24° in YPD media. The next day the cultures were diluted back to an OD600 of 0.05 and incubated at the indicated temperatures. Samples were removed at the indicated times and the cell densities were determined. Experiments were repeated at least three times with typical curves shown. (D) rmc3 cells are defective for in vitro chromatin assembly. Extracts were prepared from isogenic wild-type and rmc3 strains (YTH3 and YTH520). Twenty-five and 50 µg of extract protein were incubated with a closed, circular, relaxed plasmid for 60 min at 30° as described in B. (E) rmc3 mutants rapidly die at 37°. YTH3, 455, and 520 were grown overnight at 24°. The next day the cells were diluted back to OD600 of 0.1 and incubated at 37°. At the indicated times, samples were removed, the cell concentrations were determined, and a dilution series was plated onto YPD plates to determine viability. A representative curve from a minimum of three experiments is shown.

FACS scan analysis:

Cells were grown overnight at 30° and then diluted to an OD₆₀₀ of 0.1. Next, the cells were grown at 30° or 37° for an additional 4–6 hr prior to preparation for fluorescence-activated cell sorter (FACS) analysis according to EPSTEIN and CROSS 1992 .

Plasmid maintenance:

The different strains, transformed with YCplac111 (CEN-ARS) or YEplac181 (2µ), were grown at room temperature in liquid SD-leu. The cultures were diluted to 1 x 10⁶ cells/ml in fresh SD-leu and grown for an additional 24 hr at either room temperature or 37°. Cells then were plated on nonselective media (YPD) and grown at 30° until colonies formed. As above, 10³, 10⁴, and 10⁵ rmc3 cells were plated following growth at 37°. Colonies were then replica plated to selective plates lacking leucine and grown for an additional 3 days. Plasmid loss was calculated as 1 - (number of colonies on selective plates/number of colonies on nonselective plates) x 100. The results are expressed as the average of at least two separate experiments.

In vivo chromatin assembly:

Total genomic DNA was isolated from asynchronously growing strains according to KIM et al. 1988 . Genomic DNA was separated by slow electrophoresis in the dark through a 0.8% agarose gel in the presence of 2.5 µg/ml freshly prepared chloroquine, transferred to nitrocellulose membrane, and hybridized to a ³²P-labeled yeast 2µ gene fragment, REP1, generated by PCR. The membrane was exposed to Kodak X-ray film at -80°.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of mutants that yield chromatin-assembly-deficient extracts:
Our screen for chromatin assembly mutants exploited a yeast whole-cell extract in which in vitro nucleosome reconstitution is physiologically relevant (LING et al. 1996 ; SCHULTZ et al. 1997 ; MA et al. 1998 ). Micrococcal nuclease digestion analysis of the reconstituted chromatin revealed the formation of correctly spaced nucleosomal arrays (SCHULTZ et al. 1997 ), and this activity was most active during mitosis (ALTHEIM and SCHULTZ 1999 ). A mutant hunt based upon the use of this system has uncovered a novel chromatin assembly regulatory system defined by two ubiquitin-protein ligases, the APC and Rsp5p.

Whole-cell extracts were prepared from 100 strains selected at random from a ts yeast library (HARTWELL 1967 ). The strains in this collection typically harbor multiple mutations (R. STERNGLANZ, M. WINEY and M. YAFFE, personal communications), increasing the likelihood that a mutant of interest will be isolated. Extracts were analyzed by DNA supercoiling for their ability to assemble chromatin. One strain, YTH335, was found to reproducibly yield assembly-defective extracts (data not shown). To characterize the genetic nature of this phenotype, isogenic wild-type and mutant partners were generated by backcrossing YTH335 with laboratory wild-type strains 8–10 times. During this process three recessive complementation groups, referred to as rmc1, rmc2, and rmc3, were defined. The ts phenotype was followed through the backcrosses, and the resultant rmc1, rmc1 rmc2, and rmc3 strains remained defective for in vitro chromatin assembly (Fig 1B and Fig D). The lack of assembly activity was not due simply to an accumulation of an inhibitor in the mutant extracts because mixing wild-type with rmc1 and rmc3 extracts restored assembly activity (data not shown). Although we were surprised to recover three mutations that affect chromatin assembly in a single strain, others have previously recovered multiple mutations affecting a pathway of interest from one strain. MITSUZAWA 1993 , for example, from a pool of random mutants identified a strain bearing lesions in three genes involved in RAS signaling, pde1, pde2, and cyr1.

The rmc1 mutation conferred slow growth at 37° (Fig 1A and Fig C) and yielded assembly-defective extracts as judged by DNA supercoiling [ Fig 1B; compare the signal corresponding to highly supercoiled (sup.) vs. intermediate (inter.) topoisomers]. Extracts were determined to be defective if, in all cases, the mutants accumulated intermediate topoisomers, despite the variation that was sometimes observed (compare rmc1 lanes in Fig 1B with those in Fig 2B). The rmc1 rmc2 double mutant displayed accentuated ts and assembly phenotypes compared to the rmc1 single mutant (Fig 1A and Fig B). Although the rmc2 mutation exacerbated the temperature sensitivity of rmc1 cells, cells expressing only the rmc2 allele grew at 37° and exhibited wild-type assembly activity (Fig 1A and Fig B). While the molecular basis of this effect remains unknown, the accentuation of rmc1 phenotypes in an rmc2 background evidently is not due to decreased viability of rmc1 rmc2 cells at 37° (Fig 1E). The rmc2 mutation may modify rmc1 phenotypes through a mechanism not directly related to assembly. Cloning of RMC2 will be required to resolve this issue.

View larger version (70K): In this window In a new window Download PPT slide   Figure 2. The yeast gene APC5 restores growth at 37° and chromatin assembly defects in rmc1 cells. (A) YTH1155 (apc5CA-PA::his5+) was transformed with the CEN-URA3-based vectors YCp50 or pTH54 (APC5). The transformants were grown on SD-ura plates at either 30° or 37° for 3–5 days. Following the incubation the plates were scanned. (B) Extracts were prepared from YTH1155 cells transformed with YCp50 or pTH54 and assayed for chromatin assembly as described in Figure 1B. Untransformed YTH3 cells were included to compare efficiency of chromatin assembly.

The rmc3 allele also conferred a ts growth phenotype and produced chromatin-assembly-defective extracts (Fig 1C and Fig D). In contrast to rmc1 cells, which continue to grow at 37°, although at a reduced rate, rmc3 cells could not recover from exposure to 37° (Fig 1E). However, neither rmc1 nor rmc3 cells arrested at any one point in the cell cycle at 37° (data not shown). Rather, rmc3 cells, in particular, undergo lysis at 37° (data not shown). Therefore, in vivo defects associated with a mitotic chromatin assembly activity do not necessarily manifest into cell-cycle-specific phenotypes.

RMC1 encodes Apc5p, a subunit of the anaphase promoting complex:
RMC1 was identified by complementation of the rmc1 rmc2 ts phenotype using a low-copy library of yeast genomic DNA. As described above, the ts phenotype of rmc1 cells is exacerbated in an rmc2 background, thus enabling easier selection of positive colonies at 37°. From 50,000 transformants, 15 restored growth to wild-type levels in rmc1 and rmc1 rmc2 strains at 37°. Sequencing revealed that all 15 plasmids contained overlapping fragments of the same region of the genome (YOR241w/MET7 to YOR254c/SEC63 on chromosome XV). A minimal fragment of 2699 bp (pTH54) was subcloned, which rescued both the conditional growth phenotype and the assembly defect of rmc1 and rmc1 rmc2 cells (data not shown). Sequencing revealed that this insert encoded the first 660 amino acids, of a total of 686 amino acids, from the Apc5p subunit of the APC (YU et al. 1998 ; ZACHARIAE et al. 1998 ). Like the C-terminally truncated fragment, full-length APC5 also rescued the ts growth and chromatin assembly defects of rmc1 (Fig 2A and Fig B) and rmc1 rmc2 (data not shown).

To investigate the possibility that the rmc1 phenotype was due to a mutation in a gene other than APC5, HIS3 was integrated immediately downstream of endogenous APC5 in YTH5. A strain containing the correct integration (YTH1061), as confirmed by PCR, was then crossed to an isogenic his3 rmc1 strain. Diploids were selected and sporulated and the resultant tetrads were dissected and incubated directly at 30° or 37°. The growth characteristics of the tetrads are shown in Fig 3A. At 30°, the HIS3 marker segregated 2:2 in all 15 tetrads, confirming that HIS3 integrated at only the APC5 locus. Independently dissected tetrads grown directly at 37°, on the other hand, had a markedly different growth pattern. In these tetrads, only two spores in each of the 15 tetrads grew. This result verified that, in the rmc1 strain, a single mutation conferred the ts phenotype. When these tetrads were allowed to recover at room temperature, the rmc1 spores eventually grew (data not shown). This is consistent with the results shown in Fig 1A and Fig C, which demonstrates that growth at 37° does not kill rmc1 cells. Furthermore, all 30 spores that grew at 37° expressed the wild-type APC5::HIS3 allele, whereas the rmc1 allele segregated with the ts phenotype in all 15 tetrads. This experiment demonstrates that the rmc1 allele encodes a mutation in the Apc5p protein that renders the cells ts and chromatin assembly defective in vitro.

View larger version (42K): In this window In a new window Download PPT slide   Figure 3. A mutation to APC5 in rmc1 cells is responsible for the ts growth and chromatin assembly defects. (A) Segregation of the ts phenotype and the rmc1 allele. YTH1061 (APC5::HIS3) was crossed with YTH1106 (rmc1 his3) to determine if the HIS3 marker and growth at 37° cosegregated. Fifteen tetrads were dissected for growth at 30° and 37° on CM media. The plates were grown for 3 days and then replica plated to SD-his plates. The SD-his plates were then incubated at the appropriate temperatures to monitor ts growth and growth of cells expressing wild-type APC5. (B) Sequence analysis revealed that the mutation in rmc1 cells was within APC5. Two base pairs were deleted in the codon encoding F12, resulting in the generation of 18 additional nonspecific amino acids followed by a stop codon. (C) The apc5CA allele confers a ts phenotype. Cells lacking a genomic copy of the essential APC5 gene were maintained with a plasmid expressing CEN-URA3-APC5. These cells were then transformed with CEN-LEU2-based plasmids expressing either wild-type- or rmc1-derived APC5. Subsequent plasmid-shuffling experiments by growth on 5-FOA resulted in the loss of the URA3-based plasmids and maintenance of the LEU2-based plasmids. These strains, now expressing the different LEU2-based versions of APC5 as the sole source of APC5, were tested for growth at 30° and 37° by spotting 10-fold serial dilutions of cells. The plates were grown for 3–5 days and then scanned.

To determine the molecular nature of the defect, the APC5 gene was cloned from isogenic wild-type and rmc1 strains and sequenced. This analysis revealed that the rmc1 APC5 allele contained an AT deletion 12 amino acids into the ORF that created an in-frame stop codon 18 amino acids downstream of the deletion (Fig 3B). The rmc1 allele identified in this analysis will henceforth be referred to as apc5^CA to reflect the role that Apc5p evidently plays in chromatin assembly. Since apc5 strains are inviable (YU et al. 1998 ; ZACHARIAE et al. 1998 ), the apc5^CA allele likely produces an N-terminally truncated protein initiated at an internal methionine. This conclusion was supported by the following results of two experiments: (i) a plasmid expressing the first 78 bp of APC5 did not complement apc5^CA (data not shown) and (ii) an apc5 strain (kindly provided by A. Page and P. Hieter) expressing an apc5^CA LEU2 plasmid, generated by the loss of the complementing APC5 URA3 plasmid by growth on fluoroorotic acid (FOA), was ts (Fig 3C).

The apc5^CA allele confers phenotypes consistent with defects in chromatin metabolism and APC function:
To further characterize apc5^CA phenotypes, in vivo chromatin metabolism was assessed. For this analysis, apc5^CA strains were assayed for sensitivity to UV irradiation, which is observed in strains harboring mutations to the yeast chromatin assembly factor-I (CAF-I; KAUFMAN et al. 1997 ). Reminiscent of cells expressing CAF-I mutations, apc5^CA cells were found to be slightly UV sensitive and this phenotype was exacerbated in the rmc2 background (Fig 4A).

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. The apc5CA allele confers chromatin and APC-associated defects. (A) apc5CA and apc5CA rmc2 cells are sensitive to UV irradiation. Wild-type, apc5CA, rmc2, and apc5CA rmc2 strains (same as above) carrying the vector YCplac111 were cultured in liquid media at 37°, plated at the permissive temperature, and immediately exposed to the indicated doses of UV irradiation. A dilution series was plated out and grown at room temperature in the dark for 5 days. (B) apc5CA cells accumulate in G2/M. Wild-type (YTH6) and apc5CA (YTH457) cells were grown overnight in YPD, diluted back the next morning, and grown for an additional 24 hr at either 30° or 37°. The next morning, the cells were processed for FACS analysis. (C) The apc5CA allele confers an increased rate of CEN plasmid loss. Plasmid loss was assessed at the permissive and restrictive temperatures in wild-type and apc5CA cells carrying plasmids YCplac111 (CEN based) or YEplac181 (2µ based).

The APC is required for the targeted degradation of proteins involved in cell-cycle progression through mitosis (MORGAN 1999 ; PETERS 1999 ; ZACHARIAE and NASMYTH 1999 ). We reasoned that one or more of the APC-dependent steps might be perturbed in apc5^CA cells if the apc5^CA mutation indeed impaired APC function. To address this, we initially performed a FACS scan analysis, which revealed that, compared to wild type, apc5^CA strains contained a greater proportion of cells with a 2n DNA content, whether grown at 30° or following a shift to 37° (Fig 4B). This observation is consistent with a defect in cell-cycle progression following S phase. Furthermore, since assembly activity peaks during M phase in wild-type cells (ALTHEIM and SCHULTZ 1999 ), the assembly defect in apc5^CA cells is most likely direct.

Previous reports observed that mutations to several APC subunits (cdc16-1 and cdc23-1) or overexpression of histones in nonstoichiometric ratios resulted in a chromosome loss phenotype (HARTWELL and SMITH 1985 ; MEEKS-WAGNER and HARTWELL 1986 ). This common phenotype shared by cells harboring APC and histone alterations led us to test whether the apc5^CA allele also conferred this defect. Our experiments illustrate that apc5^CA cells are also compromised in their ability to retain CEN-based plasmids at the restrictive temperature (Fig 4C). When considered with the FACS scan results, the apc5^CA plasmid loss phenotype is most consistent with a plasmid segregation defect in mitosis. In conclusion, our phenotypic results suggest a function for APC in the maintenance of chromosome structure and integrity during mitosis.

apc5^CA genetically interacts with APC mutants:
If the apc5^CA allele does indeed compromise the function of APC, then the apc5^CA mutation would be predicted to worsen the phenotype of other APC mutants. We tested this hypothesis by crossing apc5^CA strains with APC mutant strains apc9, apc10, and cdc26. Fig 5 demonstrates that genetic interactions were observed between apc5^CA and all three APC mutants tested. At 30°, apc5^CA apc9 mutants were readily isolated (Fig 5A), indicating that cells do not require Apc9p and a fully functional Apc5p at permissive temperatures. However, only the double mutants were inviable at 37° on glucose media, compared to apc9 cells, which were not ts, and apc5^CA cells, which would eventually grow at 37°. This phenotype was observed in a total of 18 complete tetrads from two successive crosses. Interestingly, at 37°, but on media containing glycerol as the sole carbon source, apc5^CA apc9 cells grew at a rate observed to be faster than that of wild-type, apc5^CA, and apc9 cells, which all grew at an equivalent rate (Fig 5B). Note that growth of apc5^CA cells on glycerol suppressed the ts defect. This is reminiscent of observations that some APC mutants are suppressed by growth on carbon sources other than glucose (IRNIGER et al. 2000 ).

View larger version (52K): In this window In a new window Download PPT slide   Figure 5. apc5CA genetically interacts with other APC mutants. (A) YTH1155 was crossed with YTH1030 (apc9::kanMX4) and sporulated. Tetrads were dissected and grown on CM media at 30°. The spores were replica plated onto the plates shown to follow the segregation and phenotype of the different mutations. Open arrows are used to highlight selected double mutants. (B) The apc5CA and apc5CA apc9 ts phenotypes are suppressed on glycerol at 37°. The growth of apc5CA apc9 cells was compared to WT, apc9, and apc5CA cells on YPGlycerol plates at 37°. No differences were observed when the cells were grown at 30°. (C and D) YTH1033 (apc10::kanMX4) and YTH1032 (cdc26::kanMX4) were crossed to YTH1155 and treated as in A. Open arrows highlight selected cdc26 apc5CA mutants.

At 30°, apc5^CA apc10 cells were not readily isolated from 16 tetrads analyzed. Fig 5C shows a representative set of tetrads. After an additional week of growth, three apc5^CA apc10 colonies eventually grew. Similar results were observed when apc5^CA and cdc26 strains were crossed (Fig 5D). Occasionally, slow-growing colonies were observed in the 15 tetrads analyzed and, in all five cases, these colonies were apc5^CA cdc26. The isolated apc5^CA apc10 and apc5^CA cdc26 mutants were again crossed to wild-type strains and the same segregation patterns were observed (data not shown). The synthetic interactions observed between apc5^CA, apc10, and cdc26 suggest that Apc5p may share redundant functions with Apc10p and Cdc26p. However, it remains formally possible that our results reflect an interaction between unrelated phenotypes. Taken together, our genetic and phenotypic data suggest that the apc5^CA mutation leads to compromised APC function and aberrant regulation of APC-dependent chromatin assembly.

RMC3 encodes Rsp5p, a HECT domain-containing ubiquitin-protein ligase:
Complementation of the rmc3 ts phenotype using a yeast genomic CEN-based library identified RSP5 (Fig 6). RSP5 is an essential gene that encodes the Rsp5p protein-ubiquitin ligase, an E3 activity unrelated to APC. Unlike other protein-ubiquitin ligases in yeast, some of which are large multi-subunit complexes (such as APC and SCF; Skp1p/Cdc53p/F-box protein; DESHAIES 1999 ), Rsp5p is a classical single polypeptide E3, containing the C-terminal catalytic HECT domain (HUIBREGTSE et al. 1995 ). The identification of RSP5 and APC5 in our screen demonstrates the link between Ub-dependent protein-targeting systems and chromatin assembly, an association as likely to be conserved throughout evolution as are the molecular components.

View larger version (67K): In this window In a new window Download PPT slide   Figure 6. A mutation in RSP5 is responsible for the ts and assembly defects in rmc3. (A) A section of the HECT domain indicating the residues mutated in the rsp5 alleles used in this study. (B) The mutation in rmc3 is allelic to mdp1-1, an unrelated mutation in the RSP5 HECT domain (a generous gift from T. Zoladek). Diploids containing the recessive rmc3 and mdp1-1 alleles were generated and tested for growth at 37°. Growth was compared to diploids prepared by crossing WT with rmc3 and mdp1-1 strains. (C) The ts growth defect associated with rmc3 cells is due to a mutation in RSP5. Diploids containing rmc3 and mdp1-1 were sporulated and the tetrads were dissected. The plates were grown directly at 30° and 37°. (D) The mdp1-1 mutation confers a chromatin assembly defect. Extracts were prepared from the mdp1-1 and the isogenic wild-type strain and assayed for in vitro chromatin assembly. (E) RSP5 complements the rmc3 chromatin assembly defect. The rmc3 strain, YTH520, was transformed with the plasmids indicated. The empty vector controls, YEplac181 and YCplac111, are 2µ- and CEN-based, respectively. YCprsp5C-A expresses Rsp5p with the conserved Cys Ala777 mutation. YCprsp5P-S expresses the Pro Ser772 mutation identified in the rmc3 strain. The transformants were grown at 30° and 37° on SD-leu for 3 days prior to scanning. (F) RSP5 complements the rmc3 chromatin assembly defect. Extracts were prepared from YTH520 cells transformed with YCpRSP5, YCprsp5P-S, or YCprsp5C-A and assayed for chromatin assembly.

Many documented rsp5^ts alleles contain mutations within the C-terminal HECT domain (ZOLADEK et al. 1997 ; FISK and YAFFE 1999 ; WANG et al. 1999 ). The mutant allele recovered in our screen was sequenced and was also found to contain a mutation in the HECT domain at nucleotide 2314, creating a Pro Ser₇₇₂ substitution five amino acids away from the E3 active site cysteine (C₇₇₇; Fig 6A). The rmc3 allele will henceforth be referred to as rsp5^P-S. As found for other rsp5^ts mutant strains, the rsp5^P-S allele also conferred sensitivity to canavanine, growth at elevated temperatures, and acute heat shock (data not shown and Fig 1C and Fig E; KANDA 1996 ).

To demonstrate that the Pro Ser₇₇₂ mutation was responsible for the observed rmc3 phenotype and not an uncharacterized genomic mutation, the rmc3 strain was crossed to a strain expressing an independently isolated rsp5^ts allele, mdp1-1 (ZOLADEK et al. 1997 ). The resultant mdp1-1/rmc3 diploids were ts compared to mdp1-1/wild-type and rmc3/wild-type diploids, which grew at 37°, suggesting that the mutation in the rmc3 strain was allelic to mdp1-1 (Fig 6B). The mdp1-1/rmc3 diploids were then sporulated and the tetrads dissected. Each tetrad produced four spores that were ts, clearly demonstrating that the ts phenotype observed in the rmc3 strain is due to the rsp5^P-S allele (Fig 6C). Henceforth, the rmc3 strain will be referred to as rsp5^CA to reflect its role in chromatin assembly. Furthermore, extracts prepared from mdp1-1 cells were also chromatin assembly defective in vitro (Fig 6D). Taken together, these results demonstrate that the mutation in rmc3 is in RSP5, which strongly supports the hypothesis that Rsp5p plays a crucial role in chromatin assembly.

To confirm the above results, the RSP5 and rsp5^P-S alleles were cloned from isogenic wild-type and rsp5^CA strains into a YCplac111-based plasmid. If the Pro Ser₇₇₂ mutation was responsible for the observed phenotype, then rsp5^CA cells transformed with the rsp5^P-S plasmid should remain ts and chromatin assembly defective. Fig 6E and Fig F, demonstrates that only those strains containing the wild-type version of RSP5 were capable of growth at 37°, indicating that the rsp5^P-S mutation is necessary and sufficient to cause the ts and assembly-defective phenotypes.

The active site cysteine in Rsp5p, at position 777 (see Fig 6A), is required for the Rsp5p catalytic activity and viability, and this activity can be completely abolished by creating a Cys Ala₇₇₇ mutation (HUIBREGTSE et al. 1995 ; SPRINGAEL et al. 1999 ). To test whether C₇₇₇-dependent ubiquitination was required for chromatin assembly, rather than the Rsp5p motifs involved in substrate recognition or localization (the WW and C2 domains, respectively), the Cys Ala₇₇₇ mutation was cloned into the YCpRSP5 plasmid to produce YCprsp5^C-A. As predicted, the ubiquitin-dependent activity was required, because rsp5^CA cells harboring YCprsp5^C-A remained both ts and chromatin assembly defective (Fig 6E and Fig F). A similar analysis was used to demonstrate a requirement for the Rsp5p ubiquitin-dependent activity in mitochondrial inheritance (FISK and YAFFE 1999 ).

Suppression of ts and in vivo chromatin assembly defects with ubiquitin:
The active site cysteine of Rsp5p is required for chromatin assembly (Fig 6F; compare topoisomers observed in RSP5 with rsp5^C-A lanes). Therefore, it is likely that the role of Rsp5p in the regulation of chromatin assembly is ubiquitination of target proteins rather than a scaffolding or structural role. Ubiquitin (P_CUP1-Ub; ELLISON and HOCHSTRASSER 1991 ) was provided to rsp5^CA cells at high levels to potentially overcome the HECT domain mutation. Consistent with a role for ubiquitin in Rsp5p function, only those cells expressing ubiquitin at high levels were capable of significant growth at the nonpermissive temperature (Fig 7A). This is similar to previous findings where overexpression of ubiquitin complemented growth defects in rsp5 (ZOLADEK et al. 1997 ; FISK and YAFFE 1999 ). However, although we cannot conclude that suppression of the ts defect by ubiquitin is directly related to assembly, the mechanistic defects in rsp5^CA and apc5^CA cells appear to be similar because expression of ubiquitin, in the absence of copper, suppressed the apc5^CA rmc2 ts defect (Fig 7B). Furthermore, uninduced expression of RSP5, but not rsp5^C-A, also suppressed the growth defect of apc5^CA rmc2 cells (Fig 7C), strengthening the hypothesis that the apc5^CA allele confers a defect in a key ubiquitination activity. Therefore, our results suggest the likely scenario that chromatin assembly requires the ubiquitination of some protein(s) and that at least two branches of the ubiquitin-dependent targeting cascade are involved.

View larger version (61K): In this window In a new window Download PPT slide   Figure 7. Overexpression of ubiquitin suppresses the rsp5CA and apc 5CA ts phenotypes. (A) YTH520 was transformed with an empty vector (YEp24) or a plasmid expressing ubiquitin (UBI4) from the PCUP1 promoter (pES7; a generous gift from M. Ellison). The cells were grown at 30° and 37° in the presence CuSO4 (0.5 µM). (B) YTH455 cells (apc5CA rmc2) were treated as in A. In this case, CuSO4 induction was not required to suppress the apc5CA ts phenotype. (C) YTH455 was transformed with plasmids expressing RSP5 or rsp5C-A from the PGAL promoter (pYES-RSP5 and pYES-rsp5C-A; kindly supplied by J. Huibregtse). YEplac195 was used as a control. The transformants were serially diluted and spotted onto galactose (Gal)- or glucose (Glu)-containing media to induce or repress expression of RSP5 or rsp5C-A. The plates were then incubated at 30° and 37°.

The relevance of the in vitro chromatin assembly defect in rsp5^CA cells could be assessed by testing whether similar defects in chromatin assembly were observed in vivo. An in vivo assay that demonstrated the histone H2B- and H4-dependent assembly of the endogenous 2µ plasmid was subsequently tested (HAN et al. 1987 ; KIM et al. 1988 ). Fig 8A shows that 2µ plasmids isolated from the rsp5^CA strain grown at 37° were indeed assembly defective, compared to a wild-type control (compare the accumulation of intermediate topoisomers observed in lane 4 with lane 3 in Fig 8A). Comparable results were also observed when the 2µ plasmid was isolated from rsp5^CA cells grown at room temperature, illustrating that the in vivo defect was not a direct result of cell death (data not shown). These results are strikingly similar to the defective 2µ assembly observed when histones H3 and H4 are depleted (compare lanes 1 and 2 with lanes 3 and 4 in Fig 8A; HAN et al. 1987 ). As in restoration of the 37° growth defect, expression of RSP5 and Ub (from CEN or P_CUP1 plasmids, respectively) restored wild-type in vivo assembly to the isolated 2µ plasmids (Fig 8A, lanes 5 and 6). These results represent the link between ubiquitination and chromatin assembly that was lacking above and therefore strongly support the hypothesis that the Rsp5p ubiquitination activity is required for both in vitro and in vivo chromatin assembly.

View larger version (95K): In this window In a new window Download PPT slide   Figure 8. rsp5CA cells are defective for in vivo 2µ nucleosome assembly and histone metabolism. (A) Total genomic DNA was isolated from YTH3, YTH520, and RMY102 (H3/H4) cells transformed with the indicated plasmids. The growth conditions required to express the indicated plasmids are shown. Efficiency of nucleosome deposition was assessed by agarose gel electrophoresis in the presence of 2.5 µg/ml chloroquine. Following electrophoresis, the DNA was transferred to nitrocellulose membrane and hybridized with 32P-labeled REP1 DNA. (B) Multicopy expression of histones H2A, H2B, H3, and H4 suppresses the in vivo chromatin assembly defect in rsp5CA cells. YTH519 (rsp5CA, isogenic to W303 cells) was transformed with a 2µ vector expressing H2A, H2B, H3, and H4 from their endogenous promoters and grown at 37° prior to isolation of 2µ plasmids. Assembly of the 2µ plasmid was compared to untransformed YTH1 (WT, W303) and YTH519 cells. (C) rsp5CA and apc5CA cells express altered levels of histone H3. Crude whole-cell extracts were prepared from the isogenic W303 strains, YTH1335 (WT) and YTH521 (rsp5CA), and the isogenic S288c strains, YTH6 and YTH1155 (apc5CA), by bead-beating and the proteins were separated by SDS-PAGE. The proteins were transferred to nitrocellulose and immunoblotted with antibodies against native histone H3 and actin as a control. (D) Histone H3 transcript levels in rsp5CA are reduced at the restrictive temperature. Total RNA was isolated from the strains used in C and 30 µg was used for Northern analysis. DNA fragments encoding histone H3 and actin were labeled with 32P and used to detect H3 and actin transcripts.

Histone H3 levels are reduced in rsp5^CA cells:
To test whether increased abundance of histones could compensate for the defect in rsp5^CA in vivo assembly, we transformed the rsp5^CA strain with a 2µ plasmid expressing all four histones from their endogenous promoters (a generous gift from K. Robinson). Multicopy expression of all four histones in rsp5^CA cells was sufficient to suppress the in vivo chromatin assembly defect (Fig 8B). This experiment suggests that one or all histones may be altered, either transcriptionally or post-translationally, in such a way that they are not available for assembly in rsp5^CA cells.

To determine whether histone levels were indeed altered in rsp5^CA cells, we analyzed the relative abundance of native histone H3 in isogenic wild-type and rsp5^CA strains. Crude extracts were prepared from the isogenic strains, grown either at room temperature or at room temperature followed by a shift to 37° for 4 hr. The proteins from these extracts were analyzed by Western blotting using antibodies against post-translationally unmodified H3 and against actin as a control (Fig 8C). The results demonstrate that the unmodified levels of H3 are indeed reduced in rsp5^CA cells. As a comparison, we determined the relative abundance of unmodified H3 in apc5^CA cells. The results show very little change in H3 levels at room temperature in apc5^CA cells (Fig 8C). At 37°, however, there is a modest increase in H3 protein levels. A Northern analysis of H3 transcripts in apc5^CA cells demonstrates that the altered H3 protein levels are not due to aberrant transcription (Fig 8D). On the other hand, H3 mRNA in rsp5^CA cells is reduced approximately twofold in cells grown at 37°. Therefore, H3 transcription and translation is affected in rsp5^CA cells.

In summary, our analysis of yeast cells harboring a defective mitotic-specific chromatin assembly activity has identified a complex regulatory network in which components of the ubiquitin-dependent targeting pathway play a crucial role. The highly dissimilar E3s, Rsp5p and the APC, may ultimately effect chromatin assembly by playing antagonistic roles in at least histone H3 metabolism.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The results presented in this report define a broad role for the ubiquitin-dependent protein-targeting pathway in regulating chromatin assembly. Two unrelated yeast mutants identified in a screen for yeast whole-cell extracts deficient for in vitro chromatin assembly harbored lesions in two different genes encoding ubiquitin-protein ligases (E3s). The E3s identified here, APC (RMC1 encodes Apc5p) and Rsp5p/Rmc3p, are required for vastly different cellular processes, giving us the first glimpse of the extreme dynamics involved in properly regulating the deposition of histones onto DNA.

RMC1 encodes Apc5p, a subunit of the APC:
The observations that demonstrate that RMC1 encodes Apc5p include: (i) the mutation in rmc1 cells was in APC5; (ii) the rmc1 ts phenotype and the APC5::HIS3 allele segregated away from each other in 15 tested tetrads; and (iii) apc5^CA expressed from a CEN plasmid rendered apc5 cells ts. We also showed that Rmc1p/Apc5p is a subunit of APC genetically and biochemically by (i) demonstrating genetic interactions between apc5^CA and apc9, apc10, and cdc26, including enhanced phenotypes for apc5^CA apc9 cells on glucose- and glycerol-containing media at 37°, and by (ii) immunoprecipitating GST-Apc5p from cells coexpressing Apc11-HA_3x using antibodies against hemagglutinin (data not shown; the Apc11-HA_3x plasmid was a generous gift from T. Hunter; LEVERSON et al. 2000 ). And finally, apc5^CA cells were found to express APC and in vivo chromatin metabolism defects on the basis of the following experiments: (i) apc5^CA cells were defective in segregating CEN-based plasmids; (ii) the apc5^CA defect was suppressed by overexpression of genes encoding ubiquitin and a second E3, Rsp5p; (iii) apc5^CA cells accumulated with replicated DNA; and (iv) apc5^CA cells were sensitive to UV irradiation. In summary, we have identified a novel ts allele of APC5, a previously uncharacterized subunit of the yeast APC, that has allowed us to link regulation of chromatin assembly with progression through mitosis and ubiquitin-dependent protein modifications.

Cells expressing the apc5^CA allele do not behave, in most respects, like previously characterized APC ts mutants. For example, apc2, apc10, cdc16, cdc23, cdc26, and cdc27 mutants all arrest at the restrictive temperature at metaphase as large-budded cells that cannot degrade Clb2 p or Pds1p (ZACHARIAE et al. 1996 ; KRAMER et al. 1998 ; YAMASHITA et al. 1999 ; GOH et al. 2000 ). apc5^CA cells do not express these phenotypes, nor do cdc16, cdc23, cdc26, and apc9 mutants exhibit chromatin assembly defects (T. A. A. HARKNESS, V. RAMASWAMY and M. C. SCHULTZ, unpublished results). Our genetic analysis, demonstrating that apc5^CA interacted with apc9, apc10, and cdc26, was therefore necessary to show that apc5^CA cells were indeed defective for some aspect of APC function. Taken together, the results suggest that Apc5p forms a subcomplex with some other APC component(s) to ubiquitinate a target protein involved in chromatin assembly. Apc5p likely shares a common function with other APC subunits, since apc5 cells experience a lethal arrest at the metaphase/anaphase transition, similar to other essential APC subunits (ZACHARIAE et al. 1998 ; ZACHARIAE and NASMYTH 1999 ). This supports the idea that the chromatin assembly and general APC functions for Apc5p are separable. This type of scenario has been proposed for Drosophila APC5 (BENTLEY et al. 2002 ). In that study, dAPC5 was found to play a role in late cell-cycle events, but additional observations suggested a role in which dAPC5 controlled regulatory subfunctions of APC. We have preliminary data supporting the hypothesis that Apc5p functions with a subset of APC components to regulate chromatin assembly, as cells lacking Apc10p are also defective for in vitro chromatin assembly (V. RAMASWAMY, T. A. A. HARKNESS and M. C. SCHULTZ, unpublished results). The significance of this observation is highlighted by the recent genetic survey of APC mutants in Caenorhabditis elegans, demonstrating that mothers injected with apc-5 or apc-10 dsRNAs do not express the same meiotic phenotypes as other tested APC mutants (DAVIS et al. 2002 ).

A need for controlled chromatin assembly during mitosis:
The identification of APC as a positive regulator of chromatin assembly implies that chromatin assembly activity plays an important role during mitosis. The segregation of chromosomes during mitosis is a highly complex event that may involve multiple steps. The precise role chromatin assembly plays during mitosis remains elusive although several models can be envisioned. For example, as cells exit mitosis they prepare themselves for a new round of growth by transcribing the genes dedicated to DNA replication. A defect in chromatin maintenance could lead to untimely, or inhibition of, transcription of these genes. We observed that apc5^CA cells accumulate with replicated DNA, perhaps due to a delay in exiting mitosis. Alternatively, chromatin assembly could aid in packaging of the genome as the cells prepare themselves for mitosis. A defect in this process could lead to incompletely condensed chromosomes and a propensity to lose chromosomes. We have also observed that apc5^CA cells are unable to properly segregate CEN-based plasmids during mitosis. A third role for chromatin assembly during mitosis could entail a protective mechanism. Perhaps the APC triggers some chromatin assembly machinery to scan the genome for damaged chromatin as chromosomes are pulled apart or when the cells are exposed to DNA-damaging agents. A defect in this process could cause DNA to be more exposed or to lose the ability to withstand the stresses involved in chromosome segregation. Phenotypically, this could be expressed as sensitivity to UV, for example, and perhaps through increased chromosome loss, both of which are observed in apc5^CA cells. We cannot discard the possibility, however, that aspects of all these processes occur as we observe phenotypes consistent with all of the defects described above.

RMC3 encodes the E3 Rsp5p:
Strikingly, cloning and sequencing of RMC3 revealed that we had identified a second E3 enzyme. The results from six key experiments demonstrate conclusively that Rsp5p/Rmc3p is crucial for regulating chromatin assembly: (i) a mutant strain, rmc3, harboring a ts allele of RSP5 is defective for in vitro and in vivo chromatin assembly; (ii) sequencing of RSP5 from rmc3 cells revealed a Pro Ser₇₇₂ mutation in the HECT domain of Rsp5p; (iii) plasmids expressing wild-type RSP5, but not the Pro Ser₇₇₂ mutation, complement the ts growth defect and the in vitro and in vivo chromatin assembly defects in rmc3 cells; (iv) single copy expression of the rsp5^C-A active site mutant did not suppress rmc3 phenotypes; (v) overexpression of ubiquitin suppressed the ts growth defect and the in vivo chromatin assembly defect; and (vi) an unrelated mutation in the HECT domain, Pro Thr₇₈₄ (mdp1-1), conferred an in vitro chromatin assembly defect and was shown to be allelic to rmc3. Furthermore, Rsp5p was linked to chromatin because the in vivo chromatin assembly defect in rmc3 cells was suppressed by multicopy expression of all histones. We confirmed that this result reflected altered histone metabolism since at least histone H3 mRNA and protein levels were found to be reduced. Taken together, our results show that the Rsp5p ubiquitin-dependent catalytic activity is required for chromatin assembly.

A plasma membrane protein is an unlikely effector of chromatin assembly:
The localization of Rsp5p to subcellular compartments depends on the N-terminal C2 and C-terminal HECT domains, but not on the internal WW domains (GAJEWSKA et al. 2001 ; WANG et al. 2001 ). Previously published reports have shown that cells expressing the HECT domain mutations, rsp5^C-A, rsp5-1 (a ts allele encoding a Leu Ser₇₃₃ mutation; WANG et al. 1999 ), or rsp5^C6 (lacks the Rsp5p C-terminal six amino acids; HUIBREGTSE et al. 1995 ), all lack perivacuolar localization, but retain plasma membrane localization (WANG et al. 2001 ). Similar mislocalization of Rsp5p in sla2/end4-1 cells, which have a defect in the internalization step of endocytosis (WANG et al. 2001 ), suggests that a functional endocytic pathway is required for Rsp5p perivacuolar localization. Because of its similar mutation site and phenotype to rsp5-1 (Leu Ser₇₃₃), Rsp5p in rmc3 cells (Pro Ser₇₇₂) is most likely localized solely to the plasma membrane. Therefore, it is likely that the subcellular localization of Rsp5p to sites other than the plasma membrane is a crucial requirement for its chromatin-assembly-specific function. A direct test of this hypothesis would be to determine if sla2/end4-1 cells, which lack Rsp5p perivacuolar localization, are defective for in vitro chromatin assembly.

In vitro vs. in vivo chromatin assembly:
We observed differences between the in vivo and in vitro assembly activities and therefore must consider that they encompass, at least in part, separate chromatin assembly activities. Our previous and present work has established that the in vitro chromatin assembly activity is most active during mitosis (ALTHEIM and SCHULTZ 1999 ; Fig 2). In contrast, our in vivo experiments utilizing 2µ plasmids isolated from cells at various stages in the cell cycle indicate that the in vivo activity is not tightly linked to cell-cycle control (T. G. ARNASON and T. A. A. HARKNESS, unpublished results). Therefore, on the basis of our preliminary data, the in vitro and in vivo activities appear to be separate and distinct. The observation that Rsp5p is required for both activities (Fig 6 and Fig 8) implies a broad role for the Rsp5 ubiquitination activity in more than one chromatin assembly pathway. It will be of interest to compare and contrast the molecular events involved in the two assembly activities.

The Rsp5p-dependent chromatin assembly pathway ultimately effects histone metabolism:
Our results demonstrating altered histone H3 mRNA and protein levels in rsp5^CA cells indicate that the pathway in which Rsp5p functions could ultimately impinge upon histone metabolism. However, activities that control histone post-translational modifications, stability, or assembly into chromatin may also be impaired. The possibility that the assembly defect in rsp5^CA cells is strictly due to a lack of at least histone H3 is consistent with studies demonstrating that loss of H3 or H4 results in in vivo and in vitro chromatin assembly defects (KIM et al. 1988 ; SCHULTZ et al. 1997 ). It is also intriguing that apc5^CA cells appear to have a modest increase in H3 protein levels at the restrictive temperature. It is tempting to speculate that the APC and Rsp5p have antagonistic effects on similar pathways that control at least H3 metabolism.

Regulation of chromatin assembly requires a complex network of cellular interactions:
Rsp5p and the APC are distinctly different E3 enzymes, both structurally and functionally. The APC is a large protein complex with Apc11p the proposed catalytic subunit and, as a complex, is responsible for controlled progression through mitosis. On the other hand, Rsp5p is a single polypeptide E3 that encodes the hallmark HECT catalytic domain. Unlike APC, which is specifically required for progression through mitosis, Rsp5p has been shown to be involved in many cellular activities, including endocytosis, plasma membrane protein turnover, mitochondrial biogenesis, recombination, transcription, signal transduction, and resistance to anesthetics (WATANABE et al. 1995 ; IMHOF and MCDONNELL 1996 ; Z