Technical Comments

Second Family of Histone Deacetylases

Science  22 May 1998:
Vol. 280, Issue 5367, pp. 1167
DOI: 10.1126/science.280.5367.1167a

Histone deacetylases, enzymes that remove acetyl groups from ɛ-amino groups of lysines that are clustered near the NH2-termini of core histones, play a key role in the repression of transcription (1). Several histone deacetylases from many eukaryotic species form a highly conserved protein family, which also includes a variety of archaeal and bacterial enzymes that catalyze the removal of acetyl groups from small molecules; for example, from acetylpolyamines (2). However, the histone deacetylase 2 (HD2), recently identified by A. Lusseret al. (3), is unrelated to this family. HD2 has been found to contain stretches of acidic amino acid residues that are typical of nucleolar proteins, and it has been shown to localize to the nucleoli (3). A standard search of the nonredundant protein database at the National Center for Biological Information (NCBI) with the use of the gapped BLASTP program (4) did not detect any homologs of HD2 except for the two recently cloned homologs from Arabidopsis. However, a subsequent search with the use of the PSI-BLAST program (4), which incorporated a conservation profile of the two plant histone deacetylases, revealed statistically significant (the probability of a random match was below 10−3 in each case and was as low as 5 × 10−7 for FKB1_SPOFR) sequence similarity to insect proteins identified as FKBP family peptidyl-prolyl cis-trans isomerases (PPIases) (5) and to a trypanosomal RNA-binding protein (6). The conserved region included an NH2-terminal domain found in each of these proteins and was distinct from the PPIase domain and the RNA-binding domain, respectively. Further searches of the nonredundant database and the database of Expressed Sequence Tags (ESTs) with these sequences resulted in the characterization of a novel family, which includes proteins from plants, yeast, and two parasitic apicomplexans,Toxoplasma gondii and Cryptosporidium parvum(Fig. 1). Thus, this new protein family (hereafter “HD2 family”), for which we predict histone deacetylase activity, appears to be widespread among eukaryotes, although the absence (so far) of members from vertebrates is conspicuous.

Figure 1

Multiple alignment of the HD2 family. Alignment was constructed using the CLUSTALW program (14) and modified taking into account the PSI-BLAST search results. Consensus and the corresponding highlighting show amino acid residues that are conserved in the aligned sequences, with two possible exceptions. U indicates a hydrophobic residue, and X indicates an aromatic residue. indicates the insertion of acidic residue stretches in the yeast FKBPs. Secondary structure prediction (7) is shown above the alignment; E stands for Extended conformation (β-sheet); upper case shows the most confident prediction (estimated accuracy >85%). Predicted catalytic residues are highlighted blue. A Gene Identification number is indicated for each protein. Species name abbreviations: Zm,Zea mays, Os, Oryza sativa (rice), Mtc,Medicago truncatula (barrel medic), At, Arabidopsis thaliana, Tg, Toxoplasma gondii, Cp,Cryptosporidium parvum, Sf, Spodoptera frugiperda, Dm, Drosophila melanogaster, Tb,Trypanosoma brucei, Sp, Schizosaccharomyces pombe, Sc, Saccharomyces cerevisiae.

Inspection of the HD2 family alignment shows a number of conserved hydrophobic positions as well as two conserved polar residues, namely, an invariant aspartic acid and a histidine, which is replaced by an arginine in the trypanosomal RNA-binding protein Nopp44/46 and in the yeast FKBP (Fig. 1). It appears likely that the invariant aspartic acid is the nucleophile involved directly in lysine deacetylation, which may be facilitated through a charge relay system with the conserved histidine (arginine). Multiple alignment-based secondary structure prediction (7) indicated an all-beta structure for the histone deacetylase domain (Fig. 1), without detectable similarity to any known protein fold (8).

The domain organization of the (predicted) histone deacetylases of the HD2 family is of particular interest (Fig.2). In addition to the deacetylase domain, they all contain acidic stretches of various length, which may be diagnostic of nucleolar localization, or of association with basic tails of histones (9). Besides HD2, nucleolar localization has been shown for the trypanosomal RNA-binding protein Nopp44/46 (6) and for one of the yeast FKBPs (10), whereas the Spodoptera FKBP46 is a nuclear protein that binds DNA in vitro (11). The presence of a histone deacetylase and a PPIase in one protein as distinct domains makes functional sense, because in order to be targeted to the specific sites of their action on chromatin, histone deacetylases form complexes with a variety of chromatin-associated proteins (1). The chaperone-like activity of FKBPs (11) may be required for the proper assembly of such complexes. Alternatively or additionally, the FKBP domain of HD2 proteins may be involved in changing the conformation of proline-rich segments in histone COOH-terminal tails, perhaps concomitantly with deacetylation. Recent observations suggest that PPIases may have diverse and not yet fully explored nuclear functions (12).

Figure 2

Domain organization of the HD2 family proteins. NGD stands for Non-Globular Domain predicted with the use of the SEG program (15). In the yeast (S. pombe andS. cerevisiae) FKBPs, the acidic NGD is inserted within the deacetylase domain. In Nop44/46, the RGG repeat is the RNA-binding domain (6). The two conserved motifs in the deacetylase domain, centering around the putative catalytic residues, are shown by differential coloring.

The characterization of the novel family of histone deacetylases, particularly the identification of an HD2 domain in two yeast FKBPs, which should be readily amenable to genetic and biochemical manipulation, might open new avenues of research in this critical aspect of the regulation of eukaryotic gene expression. Furthermore, the prediction of HD2-like histone deacetylases in Toxoplasma and Cryptosporidia may be of particular interest, given the recent description of an antimalarial drug candidate that acts by inhibiting histone deacetylase activity (13).


Response: With the use of database sequence comparison, Aravind and Koonin present a relation between the maize nucleolar histone deacetylase HD2, recently cloned in our laboratory (1), and immunophilins of the FKBP type peptidyl-prolyl cis-trans isomerases (2). They state that the presence of a histone deacetylase domain and a distinct PPlase domain in one protein makes functional sense, because complex formation of histone deacetylases with other regulatory proteins would probably be facilitated by the rotamase activity. The references in their comment, however, refer to complex formation of Rpd3-type histone deacetylases, which are not related to the nucleolar histone deacetylase HD2 (3). Complex formation of nucleolar HD2 with other regulatory proteins, although likely to occur, has not yet been demonstrated. A rotamase activity and a putative deacetylase domain in the FKBP-type peptidyl-prolyl cis-trans isomerases could also be involved in structural rearrangements occuring in chromatin, especially during nucleosome assembly; apart from proline residues in the folded domains of core histones, H3, H2A, and especially H2B have proline residues within their flexible NH2-terminal extensions. With respect to nuclear functions of PPlases, it is interesting that the transcriptional regulator YY1 which is associated with Rpd3-type histone deacetylase interacts with members of the cyclophilin-type and FKBP-type PPlases (4). Recently, it has been shown that the DNA-binding activity of c-Myb is regulated by the interaction with Cyp40, a member of the cyclophilin type PPlases (5). However, the lack of the PPlase-domain in maize nucleolar HD2 and its close homologs in rice and Arabidopsisand in RPD3-type deacetylases would appear to exclude the possibility of a rotamase activity of histone deacetylases. Given the sequence homology between HD2 and FKBP-type PPlases, it was unexpected that the FKBP family peptidyl-prolyl cis-trans isomerases that we tested did not have histone deacetylase activity (6). We would therefore favor the idea that HD2-type histone deacetylases and certain PPlases have developed from a common ancestor enzyme.

In figure 2 of their comment, Aravind and Koonin included a putative zinc finger (C2H2) at the COOH-terminal end of HD2. We have sequenced all HD2-related cDNAs of maize and EST-clones that encode putative HD2 homologs of Oryza sativa and Arabidopsis thaliana. COOH-terminal sequence alignment (Fig.11) shows that one of the Arabidopsissequences (AT 2351064) not only lacks the zinc finger motif, but also differs significantly from the other homologs in the sequence surrounding the zinc finger. It is possible that this zinc finger has a function as a protein interaction domain rather than in specific DNA recognition and confers specificity to the HD2-subtypes; HD2 is a 400-kDa complex composed of highly homologous subunits that differ with respect to their phosphorylation state (1, 7); maize embryos contain at least three members of the HD2 family (6).

Figure 1

Multiple alignment of the C-terminal sequences of close homologs of maize HD2. The alignment was constructed using CLUSTALW program. Identical amino acids in at least 3 sequences are shown in yellow; the putative zinc finger is boxed in blue. The gene identification number is indicated for each protein. Species name abbreviations: Zm, Zea mays; Os, Oryza sativa; At, Arabidopsis thaliana.

In maize embryos, three different and unrelated histone deacetylases are present: the nucleolar phosphoprotein HD2, Rpd3-homologs, and a third enzyme that is related to neither Rpd3 nor HD2 (8). The fact that histone deacetylases belong to structurally distinct and rather divergent protein families underlines the impact of histone deacetylation for different cellular functions.


Navigate This Article