Identification and Mutation of Phosphorylation Sites in a Linker Histone

Linker histone phosphorylation has been suggested to play roles in both chromosome condensation and transcriptional regulation. In the ciliated protozoan Tetrahymena, in contrast to many eukaryotes, histone H1 of macronuclei is highly phosphorylated during interphase. Macronuclei divide amitotically without overt chromosome condensation in this organism, suggesting that requirements for phosphorylation of macronuclear H1 may be limited to transcriptional regulation. Here we report the major sites of phosphorylation of macronuclear H1 in Tetrahymena thermophila. Five phosphorylation sites, present in a single cluster, were identified by sequencing 32P-labeled peptides isolated from tryptic peptide maps. Phosphothreonine was detected within two TPVK motifs and one TPTK motif that resemble established p34 cdc2 kinase consensus sequences. Phosphoserine was detected at two non-proline-directed sites that do not resemble known kinase consensus sequences. Phosphorylation at the two noncanonical sites appears to be hierarchical because it was observed only when a nearby p34 cdc2 site was also phosphorylated. Cells expressing macronuclear H1 containing alanine substitutions at all five of these phosphorylation sites were viable even though macronuclear H1 phosphorylation was abolished. These data suggest that the five sites identified comprise the entire collection of sites utilized by Tetrahymena and demonstrate that phosphorylation of macronuclear H1, like the protein itself, is not essential for viability in Tetrahymena.

In eukaryotes, the association of DNA with histones in nucleosomes and the folding of nucleosomal filaments within chromatin can restrict the accessibility of DNA sequences to factors required for gene expression and DNA replication (1, 2). Crystallographic analyses have led to detailed appreciation of the arrangement of core histones within nucleosomes and the histone-histone and histone-DNA interactions responsible for the stability of nucleosomal structure (3,4). Many aspects of nucleosome structure are expected to be common to all eukaryotes because core histones are among the most highly conserved proteins known (1, 2). However, the structure and occurrence of proteins, collectively referred to as H1 or linker histones that bind the outer surface of nucleosomes and portions of the linker DNA extending between adjacent nucleosomes, are more variable (1, 2, 5, 6).
Data from recent biochemical and molecular genetic analyses have firmly established that chromatin structure plays a fundamental role in the regulation of gene expression. Acetylation of conserved lysine residues within the amino termini of core histones, for example, is a major pathway for modulating transcriptional activity (7,8). In contrast, the function of linker histones and their various post-translational modifications in vivo remain unclear (5,6,9). Previous notions that H1 acts globally to repress transcription (10,11) and that H1 phosphorylation is involved in mitotic chromosome condensation (9) are in contrast to evidence that linker histones can affect transcription positively or negatively in gene-specific fashion in vivo (12,13) and that linker histones are not required for mitotic chromosome condensation in vitro (14) or in vivo (15). Furthermore, recent observations suggest a role for H1 phosphorylation in transcriptional regulation (16,17).
Tetrahymena thermophila is a model organism that offers several advantages for investigations of the function and metabolism of linker histones. Like other ciliated protozoa, vegetative Tetrahymena contain two nuclei, a transcriptionally inactive, germ-line micronucleus, which divides mitotically, and a transcriptionally active, somatic macronucleus, which divides amitotically (18). Macronuclear H1 shares the overall amino acid composition and acid solubility characteristics of eukaryotic H1, in general, but lacks the central globular domain found in H1 of metazoans (6,19,20). In contrast to the multiplicity of H1 genes found in most eukaryotes, the Tetrahymena haploid genome contains only one copy of the macronuclear H1 gene (20), facilitating methods of genetic transformation we have utilized previously to demonstrate that although macronuclear H1 is not required for viability, it is required for normal compaction of interphase chromatin (15) and can function as a positive or negative regulator of expression in a gene-specific fashion in vivo (12). Like H1 of multicellular eukaryotes, macronuclear H1 is a substrate for p34 cdc2 in vitro, and we have presented evidence previously that suggests that this H1 is phosphorylated by a homolog of Cdc2 in vivo (21,22). However, unlike many eukaryotes in which high levels of H1 phosphorylation occur mainly during mitosis, the majority of macronuclear H1 molecules are highly phosphorylated in unsynchronized, growing cultures of Tetrahymena (19,(21)(22)(23). Given the absence of mitotic chromosome condensation in the amitotically dividing macronucleus (18), these differences likely reflect the high percentage of the macronuclear genome that is transcribed (24). Thus, in Tetrahymena macronuclei, potential relationships between macronuclear H1 phosphorylation and chromatin transcription can be investigated apart from effects related to the hyperphosphorylation of linker histones at mitosis that occurs in other eukaryotes. In pursuit of this goal, we have mapped the major sites in macronuclear H1 that are phosphorylated in vivo in T. thermophila to enable a molecular genetic approach employing mutagenesis of sites to investigate the function of macronuclear H1 phosphorylation.

EXPERIMENTAL PROCEDURES
Cell Culture and Strains Employed-T. thermophila strains CU427 (Mpr/Mpr [6 mp-s VI]) and CU428 (Chx/Chx-[cy-S]VII) were grown in 1% enriched proteose peptone (SPP) as described previously (25). The creation of strain ⌬1-5 will be described in detail elsewhere. 1 Briefly, site-directed mutagenesis was employed to convert the five phosphorylation sites to alanines in a construct containing the HHO1 gene coding region as well as 5Ј-and 3Ј-flanking sequences, with a single copy of the neo2 cassette (26) containing the drug resistance marker neo inserted within the HHO1 gene 3Ј-flanking sequence. Complete replacement of the wild type HHO1 gene in polyploid macronuclei was obtained by homologous recombination following biolistic transformation and selection of transformants with paromomycin.
Preparation of Macronuclear H1-Incubation of Tetrahymena at 4°C rapidly induces dephosphorylation of macronuclear H1. 2 Accordingly, except where noted, cells were collected by centrifugation at room temperature (1000 ϫ g for 5 min) prior to disruption and extraction at 4°C as described below. Macronuclei were isolated as described previously (27) using 0.1 mM p-chloromercuriphenylsulfonic acid in the homogenization and wash buffers to inhibit phosphatases (28). Macronuclei were extracted with 0.2 M H 2 SO 4 , and crude macronuclear H1 was prepared by selective precipitation of other proteins with 5.4% (w/v final) perchloric acid (PCA). 3 Crude macronuclear H1 was recovered from the PCA-soluble fraction by precipitation with trichloroacetic acid (20% w/v final), and these precipitates were washed once with acetone containing 0.1% (v/v) HCl and twice with acetone. Alternatively, crude macronuclear H1 was prepared by direct extraction of cells in 5.4% (w/v) PCA. Cells collected by centrifugation were washed twice at room temperature in 10 mM Tris-HCl, pH 7.4, prior to sonication in 5 volumes of ice-cold 5.4% (w/v) PCA. Four to six 10-s bursts of sonication (Branson model 250 with semi-micro tip, output ϭ 5, continuous duty) were required to disrupt the majority of cells. Sonicates were incubated in an ice water bath between bursts so that the temperature did not exceed 6°C. Sonicates were clarified by centrifugation at 10,000 ϫ g for 20 min, and crude macronuclear H1 was recovered from the supernatant by trichloroacetic acid precipitation. Purified macronuclear H1 was prepared from extracts of isolated macronuclei or whole cells by reversephase HPLC (RP-HPLC) using a C8 column (Aquapore RP-300, 4.6 mm x 250 mm, Perkin-Elmer Applied Biosystems) and a gradient of acetonitrile (1% (v/v) per min) in 0.1% (v/v) trifluoroacetic acid. Macronuclear H1 was recovered from RP-HPLC eluates by solvent evaporation.
Phosphorylation Site Mapping-32 P-Labeled macronuclear H1 (1-10 mg/ml) in 50 mM NH 4 HCO 3 was digested overnight (typically 14 h) at 37°C with tosylphenylalanyl chloromethyl ketone-treated trypsin (Sigma T-8642) using a 1:10 (w/w) trypsin:H1 ratio. The next morning, digests were supplemented with 50% more trypsin and incubated for 6 h. Excess NH 4 HCO 3 was removed by repeated drying under vacuum, and digests were analyzed on 40% (w/v) acrylamide gels (resolving segment 10 cm long) at alkaline pH (31). Gels were dried onto chromatography paper (Whatman No. 3MM) immediately following electrophoresis and exposed to autoradiographic film (Kodak X-Omat). Using autoradiograms as templates, phosphopeptide bands of interest were excised from the dried gels, and most of the backing paper was removed using a razor blade. Each phosphopeptide band was then cut into approximately 3 mm x 3 mm pieces, transferred to a 15-ml polypropylene centrifuge tube, and extracted with 5-10 volumes of distilled water for 4 -6 h at 4°C using a laboratory rotator. Following sedimentation of gel pieces by centrifugation, the supernatant was removed and the extraction was repeated overnight. The supernatants from both extractions were pooled and lyophilized or used directly for phosphopeptide purification by RP-HPLC.
Phosphopeptides were recovered in salt-free form by RP-HPLC. Lyophilisates were dissolved in 0.1% (v/v) trifluoroacetic acid, and all samples were adjusted to pH 2.5 with trifluoroacetic acid prior to injection onto a C18 column (Vydac 218TP52, The Separations Group). After washing with 0.1% (v/v) trifluoroacetic acid to return the base line to the preinjection level, the column was eluted with a shallow gradient of acetonitrile in 0.1% (v/v) trifluoroacetic acid. Peptide elution was monitored at 214 nm and by liquid scintillation counting. Following solvent evaporation under vacuum, phosphopeptides were analyzed by amino-terminal sequencing.
Macronuclear H1 or phosphopeptides were coupled to Immobilon-AA (Millipore) according to the manufacturer's instructions prior to sequencing. Sequencing was performed on an Applied Biosystems 477A system using standard techniques except that 90% (v/v) methanol containing 15 l of 85% (w/v) phosphoric acid/100 ml was employed for residue extraction because this gave better recovery of phosphorylated derivatives. An in-line split diverted a portion of each sequencing cycle to a fraction collector to determine 32 P content by liquid scintillation counting while the remaining portion was employed for residue identification by RP-HPLC.

RESULTS AND DISCUSSION
Macronuclear H1 prepared from Tetrahymena labeled with [ 32 P]orthophosphate during vegetative growth was resolved into five distinct bands upon electrophoresis in polyacrylamide gels containing acetic acid and urea (AU-PAGE) and in gels containing SDS (SDS-PAGE) (Fig. 1A). Essentially all the electrophoretic heterogeneity associated with macronuclear H1 in this species appears to be due to phosphorylation because it is abolished by incubating extracts with alkaline phosphatase prior to electrophoresis (19,21,23,32). Careful alignment of gels and corresponding autoradiograms established that the FIG. 1. T. thermophila macronuclear H1 is phosphorylated at multiple sites in vivo. A, macronuclear H1 prepared from cultures labeled with [ 32 P]orthophosphate in vivo was electrophoresed in polyacrylamide gels containing acetic acid and urea or SDS. Only the portions of the stained gels containing macronuclear H1 and corresponding autoradiograms are shown. Arrowheads denote the position of nonphosphorylated macronuclear H1 in each gel system. B, 32 P-labeled macronuclear H1 was digested extensively with trypsin, and the resulting peptides were resolved on polyacrylamide gels at alkaline pH. Nine peptide species (labeled 1-9 on the left) were consistently observed (lanes marked digest). Aliquots of the eight peptide fractions recovered from bands excised from preparative gels were resolved in the same gel system to assess their purity prior to microsequencing (lanes 1-7 and 9). The recovery of peptide 8 was insufficient for subsequent analysis. Only the autoradiogram is shown. fastest migrating species in each gel system represented nonphosphorylated macronuclear H1 (arrowheads, Fig. 1A). The observation that the AU-PAGE mobility of H1 proteins is inversely related to the degree of phosphorylation (33) suggests that the most slowly migrating form in AU-PAGE represented tetraphosphorylated macronuclear H1.
Although 6 serine ϩ threonine residues occur within the first 11 residues of this H1 (Fig. 3), phosphorylation was not detected between residues 1 and 12 when macronuclear H1 prepared from cultures grown overnight in the presence of [ 32 P]orthophosphate was sequenced (data not shown). Due to the absence of potential sites between residues 12 and 30, we inferred that no phosphorylation occurs within the first 30 residues of this protein. We have shown previously that all phosphorylation in this species occurs in the amino-terminal fragment liberated upon CNBr cleavage of the methionine residue at position 69 2 (21, 23) (Fig. 3). Together, these data suggest that all sites of phosphorylation can be localized to the segment extending from Thr-31 to Thr-68. Eleven potential sites of phosphorylation, comprising 2 serine and 9 threonine residues, together with numerous lysine residues are found within this 37-residue segment.
To prepare phosphopeptides for microsequencing, 32 P-labeled macronuclear H1 was digested extensively with trypsin, and the resulting peptides were resolved on polyacrylamide gels at alkaline pH (31). Following alignment of the autoradiograms and corresponding gels, bands of interest were excised from the dried gels and the phosphopeptides eluted by extracting excised gel pieces with water. Phosphopeptides were recovered from these eluates in salt-free form using RP-HPLC. This procedure enabled the recovery of eight phosphopeptides, purified to apparent homogeneity, representing all the major species and all but one of the minor species detected in the original digest (Fig. 1B). Single peaks of radioactivity were detected for all eight peptides when RP-HPLC eluates were monitored by liquid scintillation counting, suggesting that each band in the original gels was homogeneous (data not shown).
Each of these eight phosphopeptides was analyzed by microsequencing. Plots of the 32 P cpm released (y axis) against the major amino acid released in each sequencing cycle (x axis) unequivocally identified 2 serine and 3 threonine residues as sites of phosphorylation (Fig. 2). Remarkably, the eight phosphopeptides represented only three peptide "families" in which either the same sequence was phosphorylated to different extents (e.g. phosphopeptides 1 and 7, phosphopeptides 3 and 6) or the same phosphorylation site was contained within both a limit tryptic peptide and a slightly longer peptide resulting from incomplete digestion (e.g. phosphopeptides 2 and 4, phosphopeptides 5 and 9, phosphopeptides 6 and 7). Three phosphorylation sites, Ser-42, Ser-44, and Thr-46, were identified during the sequencing of phosphopeptides 1, 3, 6, and 7, which are all related to the limit tryptic peptide spanning residues 40 -49. Phosphorylation at Thr-34 was detected during sequencing of the peptides spanning residues 30 -37 (phosphopeptide 4) and residues 29 -39 (phosphopeptide 2). Phosphorylation at Thr-53 was detected during sequencing of the peptides spanning residues 51-56 (phosphopeptide 9) and residues 51-60 (phosphopeptide 5). These data are summarized and compared with findings for T. pyriformis (34) in Fig. 3. Although the analysis of macronuclear H1 phosphorylation in T. pyriformis did not provide unambiguous localization of phosphorylation to single residues in every instance, comparison of the two macronuclear H1 sequences suggests that three of the T. thermophila sites identified here, Thr-34, Ser-44, and Thr-46, are conserved as Thr-35, Ser-46, and Thr-48, respectively, in T. pyriformis (Fig. 3). Given these similarities, it is surprising that these species differ with regard to phosphorylation in their carboxyl-terminal domains.
Based on microsequencing of 32 P-labeled tryptic peptides, phosphorylation at five sites accounts for all of the major species resolved on the phosphopeptide maps, suggesting that these five sites comprise the entire collection of sites utilized in vivo. To test this hypothesis, we analyzed macronuclear H1 phosphorylation in a Tetrahymena strain, ⌬1-5, expressing macronuclear H1 in which alanine was substituted for Ser/Thr at the identified phosphorylation sites (Fig. 4A). A detailed description of the creation, molecular characterization, and phenotypic analysis of this strain will be given elsewhere. 1 Initial analyses of macronuclear H1 from growing ⌬1-5 cells on SDS gels revealed normal amounts of macronuclear H1 completely lacking the electrophoretic heterogeneity associated with macronuclear H1 from growing wild type cells, suggesting that phosphorylation of ⌬1-5 macronuclear H1 did not occur in vivo or occurred at a limited number of sites in a manner that was not sufficient to retard migration of the protein in SDS gels. To determine whether ⌬1-5 macronuclear H1 was phosphorylated to any extent in vivo, we labeled ⌬1-5 and wild type cells in parallel with [ 32 P]orthophosphate during growth overnight, prepared H1 from whole cell extracts, and analyzed it by autoradiography of SDS gels. Even though phosphorylation was readily detected in macronuclear H1 from wild type cells,  Fig. 1B were covalently coupled to a sequencing support and analyzed by amino-terminal microsequencing. The results for each peptide are shown by a graph (labeled 1-7 and 9 in boxes) in which 32 P cpm released in each cycle are plotted. The major residue recovered in each cycle is also identified. Phosphorylated residues are underlined. Numbers below the first and last residues of each phosphopeptide indicate their position within the macronuclear H1 sequence.
phosphorylation was not detected when an equivalent amount of macronuclear H1 from ⌬1-5 cells was analyzed in parallel, even after lengthy autoradiographic exposures (Fig. 4B). Because macronuclear H1 from the wild type cells (but not that from ⌬1-5 cells) was prepared using a method that enriched less phosphorylated forms, we conclude that macronuclear H1 is not detectably phosphorylated in growing ⌬1-5 cells under the conditions employed here.
Taken together, these data suggest that the five sites iden-tified by microsequencing represent the complete set of macronuclear H1 phosphorylation sites utilized by T. thermophila. Alternatively, a hierarchy of sites in macronuclear H1 may exist such that the abolition of phosphorylation (or possibly the alanine substitution itself) at one or more of the sites mutated in ⌬1-5 macronuclear H1 precludes phosphorylation at unidentified secondary sites. It should be noted that our analyses of phosphopeptides 1, 3, 6, and 7 indicate that, in wild type cells, phosphorylation occurs hierarchically, first at Thr-46, then at Ser-42, and last at Ser-44, in vivo (Fig. 2). Whether this hierarchy extends to other sites is not known at present. Regardless of whether we have identified all of the sites of phosphorylation, it is clear that eliminating the 5 sites identified here results in cells completely lacking detectable phosphorylation of macronuclear H1. Our findings that macronuclear H1 phosphorylation is not essential and that ⌬1-5 strain cells are viable without marked alterations in growth, 1 suggest either that macronuclear H1 phosphorylation does not have major, global effects on transcription or that mechanisms exist that are capable of compensating for the absence of macronuclear H1 phosphorylation in ⌬1-5 cells. Further analyses of the ⌬1-5 strain, in comparison with wild type Tetrahymena and to strains constructed to mimic constitutive phosphorylation or dephosphorylation of all or some of the five sites identified here, should define the function of macronuclear H1 phosphorylation in vivo.  (35) is shown aligned with the protein sequence determined for T. pyriformis macronuclear H1 (lower) (34). Gaps introduced to maximize homology are indicated (Ϫ), and identical residues are noted (*). No phosphorylation of residues 1-11 in the T. thermophila sequence was detected by direct microsequencing of intact, purified, highly phosphorylated macronuclear H1 labeled in vivo. The five sites identified by microsequencing are indicated (P). Residues in T. pyriformis macronuclear H1 that were found to be phosphorylated or potentially phosphorylated in chemical assays of tryptic peptides (34) are underlined.
FIG. 4. Mutagenesis of identified sites prevents phosphorylation of macronuclear H1 in T. thermophila. A, a segment (residues 31-60) of macronuclear H1 from wild type Tetrahymena that contains the five phosphorylation sites (P1-P5) identified by microsequencing in Fig. 2 is shown (upper). Three sites that resemble the consensus sequence described for Cdc2 kinase (36) are underlined. The equivalent segment from Tetrahymena strain ⌬1-5 is shown (lower). This strain was transformed with an HHO1 gene mutagenized in the codons for Thr-34, Ser-42, Ser-44, Thr-46, and Thr-53 so that all copies of macronuclear H1 expressed by ⌬1-5 cells contain alanine at these positions as indicated (*). B, the ability of ⌬1-5 cells to phosphorylate the mutated macronuclear H1 in vivo was assessed by analysis of extracts from 32 P-labeled cells on SDS gels. Wild type (lanes wt) and ⌬1-5 cells were grown overnight in media containing [ 32 P]orthophosphate. Prior to extraction with 5.4% (w/v) PCA, wild type cells, but not ⌬1-5 cells, were cooled briefly to induce partial dephosphorylation. Equivalent amounts (approximately 5 g) of macronuclear H1 from each strain was analyzed on a polyacrylamide gel containing SDS. The macronuclear H1 region of the stained gel (left) and the corresponding autoradiogram (right) are shown.