Sequence Analysis of Acetylation and Methylation in Two Histone H3 Variants of Alfalfa*

Analysis of acetylation in the two histone H3 variants of alfalfa by acid/urea/Triton-polyacrylamide gel electrophoresis has established that the minor variant H3.2 has a 2-fold higher level of acetylation than the major variant H3.1. Purification and sequence analysis of both variants showed sequence identity across the complete amino-terminal domain, which contains the 6 modified lysines 4, 8, 14, 18, 23, and 27. The two proteins have different distributions for acetylation: mono-, di-, and tri-methylation. The higher level of acetylation of H3.2 was confirmed in a wider pattern across all 6 lysines. Lysine modification levels varied for all sites in both proteins between 5 and 95%, with combinations of one to four types of modification co-existing at each residue. Additional sequence analysis of the H3.1 and H3.2 proteins and of tryptic core peptides established that the two histones differ only in residues 31, 41, 87, and 90. This indicates that major histone H3.1 is the product of the major alfalfa histone H3 gene and makes it likely that H3.2 is the product of the minor H3 gene, known from a partial cDNA clone. The variant-specific differences in lysine modifications in protein domains with identical primary structures suggest that the pattern and level of lysine modifications may be directed by the distinct chromatin environments of the two histone H3 variants.


Analysis
of acetylation in the two histone H3 variants of alfalfa by acid/urea/Triton-polyacrylamide gel electrophoresis has established that the minor variant H3.2 has a a-fold higher level of acetylation than the major variant H3.1. Purification and sequence analysis of both variants showed sequence identity across the complete amino-terminal domain, which contains the 6 modified lysines 4, 8, 14, 18, 23, and 27. The two proteins have different distributions for acetyla-Con: mono-, di-, and tri-methylation.
The higher level of acetylation of H3.2 was confirmed in a wider pattern across all 6 lysines. Lysine modification levels varied for all sites in both proteins between 5 and 95%, with combinations of one to four types of modification coexisting at each residue. Additional sequence analysis of the H3.1 and H3.2 proteins and of tryptic core peptides established that the two histones differ only in residues 31, 41, 87, and 90. This indicates that major histone H3.1 is the product of the major alfalfa histone H3 gene and makes it likely that H3.2 is the product of the minor H3 gene, known from a partial cDNA clone. The variant-specific differences in lysine modifications in protein domains with identical primary structures suggest that the pattern and level of lysine modifications may be directed by the distinct chromatin environments of the two histone H3 variants.
Study of the two histone H3 variants of alfalfa (Me&ago satiua) has revealed that the major variant, histone H3.1, representing two-thirds of the histone H3 protein, has a steady-state acetylation that is only half of that seen in the minor variant, histone H3.2 (Waterborg et al., 1987(Waterborg et al., , 1989. The correlation between high levels of histone acetylation and transcriptional activation of chromatin (Matthews and Waterborg, 1985;Hebbes et al., 1988;Pfeffer et al., 1988Pfeffer et al., ,1989Turner and Fellows, 1989) suggests that histone H3.2 plays a specific role in active chromatin in a way similar to other histone variant proteins with a specific location in active chromatin (Allis et al., 1986;Donahue et al., 1986). The current model for gene activation within chromatin envisions that specific gene regulatory proteins induce a change in the conformation of chromatin. This change increases the accessibility of the core histones for dynamic acetylation and results in higher levels of histone acetylation. The increased acetylation levels appear a necessary but not * This work was supported by National Science Foundation Grant DCB-8896292.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C.
Section 1734 solely to indicate this fact. sufficient requirement to allow RNA polymerases to further unfold the chromatin for gene transcription (for reviews, see Matthews and Waterborg, 1985;Garrard, 1987, 1988;Pederson and Simpson, 1988;Vidali et al., 1988). This model assumes that differences in histone sequence, observed as histone variants, play no role. However, this model cannot explain observed preferential localization of some histone variants in active chromatin (Allis et al., 1986;Donahue et al., 1986;Jasinskiene et al., 1985Jasinskiene et al., , 1988 nor their exclusion from it (Halleck and Gurley, 1982). Moreover, it predicts that all histone variants of a certain histone type will have, on average, identical levels of histone acetylation, whereas the levels of histone H3 acetylation in the two H3 variants of alfalfa (Waterborg et al., 1987(Waterborg et al., , 1989) and tobacco' are very different.
This study was initiated to evaluate whether the primary structure of the alfalfa histone H3 variants differ sufficiently in the domain containing the acetylation sites to explain the observed differences in acetylation.
Sequence analysis of substantial parts of purified H3.  Brandt, 1977) as described previously (Waterborg et al., 1987). Fractions containing histone H3 were identified by acid/urea/Triton (AUT)*gel electrophoresis, pooled, dialyzed into 2.5% acetic acid, 0.1 mM 2mercaptoethanol, and lyophilized. Histone H3 variants labeled in uioo were prepared from 20 g of callus cells, dispersed in growth medium, and labeled for 60 min with 1 mCi ["HIacetate as described previously (Waterborg et al., 1989), by the procedure described above, with the omission of Bio-Gel P60 chromatography. Separation of Histone H3 Variants-Histones were solubilized in 7.2 M freshly deionized urea, 50 mM dithiothreitol, 0.75 M ammonium hydroxide, and 0.05% phenolphthalein for 5 min at room temperature, acidified by the addition of l/20 volume of glacial acetic acid, and clarified by centrifugation for 5 min at 13,000 x g. Crude histones prepared from up to 20 g of callus or histone H3-enriched Bio-Gel preparations from up to 300 g of callus were injected on a 4.6-mm X 25-cm Zorbax Protein Plus (Du Pont-New England Nuclear) reversed-phase column, equilibrated in buffer A (0.1% trifluoroacetic acid in water).
The column was developed at 1 ml/min by 5 ml of buffer A, a gradient of 5 ml from 0 to 42% buffer B, a gradient of 60 ml from 42 to 52% buffer B, and a gradient of 5 ml from 52 to 70% buffer B followed by re-equilibration to buffer A. The two histone H3 variant proteins were localized by UV absorbance at 214 nm and by AUT gel analysis, and pooled separately, excluding the small overlap between the two peaks after preparative chromatography. Analytical retention times were 45 min for histone H3.1 and 49 min for H3.2 ( Fig. 1). The pooled histones were judged to be homogeneous by SDS and AUT gel electrophoreses.
A typical recovery of homogeneous H3 variants was 0.6 pg of H3.1 and 0.3 pg of H3.2 protein/g of callus. For the analysis on the total number of detectable acetylation sites by fluorography (Fig. 2), the small overlap between the two histone H3 variants was not discarded to avoid possible bias, but care was taken to avoid inclusion of H3.1 protein into the H3.2 fraction because this would have interfered with the detection of high levels of H3.2 acetylation.
These considerations lead to a minor contamination of histone H3.2 into the H3.1 fraction (Fig. 2) mg/ml and digested with l/40 weight of L-l-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington) twice for 2 h at 37 "C. The digest was lyophilized and fractionated on a 4.6-mm X 25cm Zorbax Protein Plus reversed-phase HPLC column in 0.1% trifluoroacetic acid in water which was developed at 0.5 ml/min with a 120-min gradient to 70% acetonitrile in water with 0.1% trifluoroacetic acid. The tryptic core peptide of both H3 variants (residues 84-115 (Brandt and Von Holt, 1982;Vanfleteren et al., 1987)) eluted at 93 min and was identified by UV absorbance of tyrosine at 275 nm and by amino acid analysis.
Analysis of Histone H3 Variants-SDS and AUT gel electrophoreses, fluorography, and gel quantitation were performed as described previously (Waterborg et al., 1987(Waterborg et al., , 1989 lo-fold more histone H3 per reversed-phase HPLC run (see "Experimental Procedures"). Moreover, a small contamination of minor H2A variant protein and trace amounts of nonhistone proteins were removed from the histone H3 preparations. A single band of identical molecular weight was visible for each preparation in SDS gels (lanes 10 and I1 in Fig. 1). The purity of each histone variant preparation was further proven by amino acid analysis (Table I) and by automated sequence analysis, which indicated a contamination level of less than 1% of non-histone H3 protein.
Analysis of the two purified histone variant proteins by AUT gel electrophoresis showed a series of multiple bands for each protein (lanes 3-4 and 6-7 in Fig. 1) in addition to a variable amount of presumably oxidized histone H3 dimer. To identify acetylation as the cause for this pattern of distinctly charged histone H3 protein forms and to establish the total number of detectable modification sites, alfalfa callus cultures were labeled in uiuo with tritiated acetate, histone H3 variant proteins were fractionated by reversed-phase chromatography of the crude histone preparation, and the fractions were analyzed by AUT gel electrophoresis (Fig. 2). Histone H3.1 clearly showed three modified protein bands with increasing discrete reductions in gel mobility relative to the major protein form (Protein .l in Fig. 2). These modified bands appear to be caused by acetylation, as judged by the acetate label incorporation pattern (Label .l in Fig. 2), but extensive fluorographic analysis revealed the existence of six possible sites ( Fig. 2A). Histone H3.2 had a 2-fold higher level Fm. 2. Acetylation of histones H3.1 and H3.2 in AUT gels.
Histone H3 variants were labeled in uiuo with tritiated acetate and purified from a total histone preparation by reversed-phase HPLC. The H3.1 preparation (.I and A) with a slight inclusion of histone H3.2 (see "Experimental Procedures") and the H3.2 preparation (2 and B) were electrophoresed in parallel AUT gel lanes and stained for Protein by Coomassie and for radioactive Label by quantitative fluorography. The relative positions of non-through multiacetylated H3 bands are numbered separately and plotted to show their relative electrophoretic mobilities. A densitometric scanning analysis is shown after Coomassie protein staining (continuous line) and after fluorography for 4 days (broken line) and 67 days (clotted line). The overexposed portion of the 67-day exposure has been omitted. The scanning patterns of histone H3.1 (A) were aligned with the patterns of histone H3.2 (R) as observed in gel electrophoresis (from left to right). purified to homogeneity by excluding the small overlap between the peaks (Fig. 1).
As a source of histone H3, total cells were used (Waterborg ' ND, not determined. Alfalfa Histone H3 Variants 17159 et al., 1987) rather than isolated nuclei (Waterborg et al., 1989). Artificial heterogeneity arises during the preparation of alfalfa nuclei, probably due to some form of cysteine modification in the H3 proteins (Waterborg, 1990). The extraction of histones from total cells rather than from nuclei increased histone recoveries approximately 2-fold but decreased the histone content of the crude extract to below 50%. The maximal capacity of a Zorbax Protein Plus column (4.6 mm X 25 cm) for separation of the histone H3 variants was reached when 5 mg of total protein was applied. Fractionation of the total crude histone preparation by Bio-Gel P60 chromatography (Waterborg et al., 1987) to obtain a histone H3enriched preparation allowed separation and purification of Alfalfa Histone H3 Variants of steady-state acetylation than did H3.1, as determined by quantitative Gaussian curve-fitting analysis of the Coomassiestained pattern (Protein .2, and the continuous scan in Fig.  2B). The higher level of H3.2 acetylation was also supported by the larger number of detectable acetylation bands (Label .2 in Fig. 2) and their relatively higher abundance in the fluorography pattern with probably more than six sites of modification (Fig. 2B). Automated sequence analysis of both histone H3 variant proteins for 30 residues revealed identical sequences (Fig. 3). At lysine residues 4, 9, 14, 18, 23, and 27, additional, unidentified phenylthiohydantoin-derivative peaks were observed at varying intensities. With the known postsynthetic modification of lysine by acetylation and methylation, the elution pattern of various modified forms of lysine was established using known standards. Extension of the NORMAL-l HPLC gradient elution program enabled all previously unknown peaks to be identified as acetylated, mono-, di-, or trimethylated lysine. With quantitation of these modified residues relative to unmodified lysine, a complex pattern of acetylation and methylation was revealed with distinct quantitative and qualitative differences between histones H3.1 and H3.2 (Table  II). It should be noted that the accuracy of quantitation of lysine, acetyllysine, and monomethyllysine is higher than that Top, protein H3.1 automated protein sequence results (upper caee letters) were aligned with the protein sequence predicted for major alfalfa H3 gene ALH3.1 (lower case letters).

Residue
Histone H3 To establish the extent of primary structure differences for the two histone H3 proteins, additional sequence analyses were performed, guided by the observed differences in the amino acid analysis between the two variant proteins (Table  I) and by the published cDNA sequence data for the major histone H3 gene of alfalfa (Wu et al., 1989). This study identified a minor histone H3 gene from a partial cDNA clone (pH3c.l1), predicting a histone H3 protein from amino acid residues 17-135 which would differ at residues 31,41,87, and 90 from the major gene (Fig. 3). Amino acid analysis of histone H3.1 and H3.2 showed differences only for threonine, serine, alanine, leucine, tyrosine, phenylalanine, and histidine (Table  I), consistent with the major protein variant H3.1 as the product of the major histone H3 genes of alfalfa and the minor protein variant H3.2 as the product of the minor gene(s). This conclusion was supported by automated sequence analysis of histone H3.1 for 44 and of H3.2 for 51 residues, which showed the predicted sequences, including the differences at residues 31 and 41 (Fig. 3), and which supplied the amino-terminal sequence missing from the partial gene clone. Lysines 36 and 37 in both proteins appeared not to be modified by acetylation or methylation. Tryptic core peptides were prepared and sequenced to confirm the predicted sequence differences for residues 87 and 90 (Fig. 3).

DISCUSSION
This study identifies the three major sites of acetylation in alfalfa histone H3 as lysines 14, 18, and 23. Minor sites of acetylation were only quantitated in histone variant H3.2 as lysines 27 and 9, with a trace of modification detected at lysine 4 (Table II). This shows that the sensitivity for quantitation of acetylation by sequencing is similar to that by quantitative analysis of Coomassie-stained AUT gels (Fig. 2). Calculation from the sequence data of acetylated lysine residues per protein molecule for H3.1 and H3.2 yields values of 0.38 and 0.73 acetyllysine, respectively. These are similar to values determined from AUT gel analysis of 0.44-0.56 for H3.1 and 0.97-1.01 for H3.2. In both analyses, the relative level of acetylation is estimated to be P-fold higher for histone variant H3.2 than for variant H3.1.
Fluorographic detection of the total number of acetylation sites of alfalfa histone H3 is more sensitive than either method, but it is limited to a qualitative assessment. At least seven sites of acetylation were detectable in histone H3.2 (Fig.  2), six of which are presumed to be the 6 lysines observed to be acetylated in histone H3.2 (Fig. 3). This result may indicate that, in trace amounts, lysines other than those in the aminoterminal region of histone H3 may become acetylated. An alternative explanation would be the generation of modified forms of histone H3 by modifications other than acetylation. The involvement of phosphorylation appears excluded by the failure to label histone H3 in uiuo with radioactive inorganic phosphate (Waterborg et al., 1989).
Histone H3 can be divided into two distinct structural domains. The amino-terminal domain of residues l-27 contains all known lysine acetylation and methylation sites and appears highly accessible in chromatin as judged by its sensitivity towards proteolytic fragmentation, especially near residue 27 (Rill and Oosterhof, 1982;Marion et al., 1983;Crane-Robinson and Boehm, 1985;Dimitrov et al., 1986;Dumunis-Kervabon et al., 1986;Encontre and Parello, 1988;Ausio et al., 1989). The carboxyl-terminal part of histone H3 appears to be a structured, globular domain with major hydrophobic characteristics. The sequence differences between histone H3.1 and H3.2 at residues 31, 41,87, and 90 all reside in the globular domain. This makes it highly unlikely that these primary structure differences could be the determining factor in the distinct differences in lysine acetylation and methylation within the amino-terminal domain of the histone H3 variants (Table II). The pattern of acetylation of histone H3.1 covers three contiguous lysines at positions 14, 18, and 23. Acetylation of histone H3.2 is similar, with higher levels of modification of these lysines and low levels in additional, contiguous lysines 27,9, and 4 (Table II). This pattern suggests that the histoneacetylating enzymes lack a strict sequence specificity and that their action may be directed more by steric determinants. Lysines within the globular domain beyond residue 27, such as lysines 36 and 37, remain completely unmodified. The decrease in acetylation levels from lysine 14 to lysine 27 may be caused by increasing steric interference for acetylation by the neighboring globular histone domain. However, a severe restriction towards acetylation of lysines 9 and 4 is apparent. It is tempting to speculate that this area of histone H3 is closely associated with some chromatin structural element when histone H3 is part of chromatin. Highly methylated lysine 9 may be important for this interaction, because in histone H3.2 this residue is less methylated and, as predicted, acetylated to a higher extent than in histone H3.1 (Table II).
The higher levels of acetylation of histone H3.2 with an apparently wider pattern of accessibility are consistent with the concept that histone H3.2 with its higher acetylation levels is part of transcriptionally active chromatin. Such chromatin is expected to be partially unfolded with increased accessibility for proteins. This speculation suggests that the specific or preferential localization of histone H3.2 in activated chromatin causes its higher level of acetylation. It leaves open the question on the mechanism by which histone variants may become localized in different chromatin environments.
Histone methylation is generally considered to be an irreversible modification of lysines in soluble histones and in chromatin for which no function has been established (Thomas et al., 1975;Honda et al., 1975b;Duerre and Buttz, 1990). In histone H3, partial methylation and acetylation at the same lysines (residues 4 and 9) has been observed during sequencing of bovine and trout histone H3 (DeLange et al., 1972;Honda et al., 1975a). Some histone methylation has recently been shown to change in Drosophila when heat shock genes were induced or transcription was inhibited (Tanguay and Desrosiers, 1988;Desrosiers and Tanguay, 1989). Histone methylation may thus be important in the regulation of chromatin activity. However, it is unknown whether mono-, diand trimethylation are functionally identical. The distinct differences observed in the patterns for all three forms of lysine methylation (Table II) might suggest functional differences, but not enough is known for fruitful speculation. For this paper, it appears most important to realize that prior methylation of a lysine, especially of lysines 9 and 27, will limit the possibility for histone acetylation.