Minireview 0 1993 by the American Society for Biochemistry and Molecular Biology, Inc
Total Page:16
File Type:pdf, Size:1020Kb
THEJOURNAL OF BIOL~~ICAL CHEMISTRY Vol. 268, No. 36, Issue of December 25, pp. 26813-26816, 1993 Minireview 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. X-ray crystallographicanalysis of the DNA-bound engrailed (6) Understanding the and a2 homeodomains (7) and an NMR spectroscopy-derived Homeodomain* structure of the Antennapedia-DNA complex revealed nearly identical structures. Eachprotein contains threea-helices that Thomas B. Kornberg are spaced and arrangedsimilarly, and each contactsDNAwith From the Department of Biochemistry, University of a pincer-like grip. In each, the N-terminal arm of 9 residues California, San Francisco, California 94143 extends across the DNA binding site making contacts in the minor groove, and residues in the third a-helixcontact DNA in The homeotic genes of Drosophila have long captured the theadjacent major groove. Moreover, despite only 27% se- fascination of biologists, in partfor the remarkablephenotypes quence identity and a 3-residue insertion in a2 between resi- that mutant animalsdisplay. Homeotic mutants arecapable of dues 23 and 24 (Fig. IA), the a-carbon backbones between placing whole body parts in inappropriate locations, in effect residues 9 and 58 of engrailed and a2 homeodomains align with replacing halteres withwings, or antennae withlegs, or mouth a root mean square difference of 1.0 A. These models under- parts withlegs. These phenotypes suggested that thegroups of score the similarities of even disparate homeodomains, as well cells whose descendents will make these body parts choose as the modular natureof the proteinsof which these segments between alternative developmental pathwaysduring the are a part. Such remarkable conformational identity between course of normal development and implied that thesedecisions distantly related membersof the homeodomain family suggests are directed by the products of the homeotic genes. It follows that their salient structural traits functional and properties are that the cells whose developmental program can be switched shared among all homeodomains. between alternative fates share an inherenthomology and that the differenthomeotic genes that direct their ultimatediversi- The N-terminal Arm fication function in similar capacities. The suggestion that the Although the NMR and crystallographic data did not resolve homeotic genes might sharea common ancestry was madesev- the N-terminal residuesof either the Antp (residues0-61, en- eral years before any were isolated (1). grailed (residues 0-21, or a2 (residues 0-3) homeodomains, or The notion that the presentday homeotic genes might have the side chains of several of the succeeding residues, contacts arisen by duplication and divergence gained immediate accept- with several bases in the minor groove were detected. In en- ance when their isolation and characterization revealed, in grailed, Arg-3 and Arg-5 contacted three or four AT base pairs; each, a conserved homeobox region (2, 3). The homeobox has in a2, Arg-7 contacted two AT base pairs, and the N-terminal since been found to be present in a large family of eukaryotic arm of Antennapedia (Antp)was in a position to make similar regulatory proteins (4). The homeobox family now numbers contacts. The possibility that these are artifactualcontacts that more than 300, continues to grow, and includes genes from form only in the environmentof the crystalsseems unlikely in organisms broadly representative of both the plant and animal view of several relatedobservations. For instance, conservation kingdoms. Although it is now clear that most members of this of Arg at residue 5 (97% among the >300 members of the ho- gene family have developmental roles that are unrelated to meodomain family) implies that Arg-5 has a particularly im- homeosis, the homeobox remains a hallmark of homeotic genes, portant role. In addition, minor groove base and phosphate and a number of these genes appearto have retainedboth their backbone contacts are also indicatedby patterns of methylation precisely ordered tandem arrangement in thegenome, as well and ethylation thatblocked binding of Antp and fushi tarazu as their developmental roles in axial patterning across vast (ftz) homeodomains (8, 9). The contacts inferred from these evolutionary time (5). interference studies are consistent with every minor groove The impressive conservation of nucleotide sequences of the base and phosphatecontact observed in theengrailed cocrystal. homeoboxes has fostered the efficient and rapid isolation of Finally, the importance of the N-terminalarm-DNA contactsis homeobox-containing genes from diverse sources. Even more underscored by the behavior of mutant homeodomains. A ftz impressive is the conservation of the protein segmentencoded homeodomain thatlacks the N-terminal 6 residuesmakes by the homeobox, the homeodomain. For instance, the Dro- fewer base and phosphate backbone contacts and binds DNA sophila Antennapedia(Antp) and human HoxB7 homeodo- with 130-fold lower affinity than does the complete homeodo- mains differ at only 1 of 61 residues, and 16 other homeodo- main peptide (8).An a2 mutant inwhich Ala replaces Arg-7 has mains from fly, human, sea urchin, mouse, rat, and frog that significantly lower repressor activity in vivo (cited in Ref. 7). A are related by sequence to the Antp homeodomain differ at POU domain protein that lacks residues4 and 5 failed to bind fewer than 7 residues. Such extraordinary conservation must DNA in vitro (10). be indicative of conserved functions. Despite the apparent importance of the minor groove con- Structural models have been obtained for three homeodo- tacts by the N-terminal armto DNA binding, the contribution mains: 1) a polypeptide representing the Drosophila Antp ho- that thesecontacts make tosequence selectivityor specificity of meodomain and seven additionalC-terminal residues; 2) a binding is unclear. It is thought that minor groove contacts polypeptide representing the Drosophila engrailed homeodo- cannot discriminate between AT and TA base pairs (11)but that main; and 3) a C-terminal 82-residue fragment of the yeast the extracyclic NH2 of the guanineshould allow for CG and GC MATa2 protein that contains the a2 homeodomain and a 21- base pairs tobe readily distinguished from AT base pairs. The amino acid C-terminal tail. Each of these protein fragments Arg-5 and Arg-7 side chains of the engrailed, a2, and Antp adopts a stable structure that binds DNA with high affinity. homeodomains that contact AT base pairs in theminor groove presumably would have unfavorable interactions with a gua- * This minireview will be reprinted in the Minireview Compendium, nine base. These considerations may explain the preference of which will be available in December, 1993. many homeodomain proteins for AT-rich sites, but selecting 26813 26814 A. Consensus en a2 TKPYRGHRFTKENVRILESWFAKNIENPYLDTKGLENLMKNTSLSRIQIKNWVSNRRRKEKTIT mB BMMBMBBB BB ERKRGRQTYTRYQTLELEKEFHFN RYLTRRRRIEIAHALCLTERQIKIWQNRRMKWKKEN BB B MB B E. N-mminus Helix I Helix I1 Helix III c-taminus 0 10 20 30 40 50 60 scr Ubx Dfd oct-1 Oct-2 FIG.1. Sequence comparisons of homeodomains.A, indicated at the top are the a-helices common to engrailed, a2, andAntp homeodomains. They have been aligned with the sequences of the engrailed, a2, Antp, and Oct-1 homeodomains. The "consensus" sequence has been determined by comparing 87 homeodomains and the conserved residues are shown interspersed among positions (indicated by dashes) with more sequence variation (4). The numbering is that of Kissinger et al. (6). Below the engrailed, a2, and Antp sequences are indicated residues that contact the phosphate backbone (B)or that contact bases in theminor (m)or major (MIgroove of DNA.B, comparisons of the Scr, Ubx, and Dfd homeodomains to Antp and of Oct-2 to Oct-1 represent common residues as dashes and specify different residues with the single letter code. Underlined residues indicate those amino acids whose identity alters specificity in chimeric Drosophila proteins or are critical to the interaction of Oct-1 with VP16. among AT-rich sites might not be possible on the basis of such homeodomain with Scr-specific residues was found to be suffi- contacts. It is therefore surprising that much of the target cient to change the specificity of the Antp protein to that of Scr specificity of the homeodomain has been found to be in the (16).Criteria used in these assays were Antp-specificinduction N-terminal arm. of ectopic anterior teashirt expression in theembryonic midgut Drosophila homeotic genes such as Ultrabithorax (Ubx), and head epidermis, Scr-specific induction of ectopic larval sal- Antp, Deformed (Dfd), and Sex combs reduced (Scr) encode ivary glands, and larval cuticular and adult head transforma- proteins whose homeodomains have very similar sequences tions which are different in HS-Scr and HS-Antp animals. (Fig. 1B).The number of amino acids that distinguish the Ubx, Similarly, replacing the 6 distinguishing residues in the N- Scr, and Dfd homeodomains fromthe Antp homeodomain are 4, terminal arm of the Dfd protein with Ubx-specific residues was 6,and 10,respectively. In contrast, the Antp, Ubx, Scr,and Dfd sufficient to change the specificity of the Dfd protein to that of proteins have few other sequence similarities. These homeodo- Ubx (14). Replacing two N-terminal residues was found to be main proteins are expressed in different regions of the devel- necessary but not sufficient to change the specificity of Ubx oping fly. and are thought to each direct developmental path- protein to that of Antp (15). ways by regulating different sets of downstream target genes. If the engrailed-DNA cocrystal structure is an appropriate The basis of their target specificity has been investigated by model for these other homeodomains, then we might conclude creating chimeric genes in which the homeodomains and/or that theresidues in the N-terminal arm thatdetermine speci- flanking regions were exchanged, and the capacity of the chi- ficity do not interact with DNA directly. Whereas Arg-3 and meric proteins to direct different developmental pathways has Arg-5 of engrailed contact DNA and these two arginines are been assessed in vivo.