Classification of LIM proteins

Classification of LIM proteins

pharmaceutical companies each spend almost $1 billion per year on research and developmentl3. In this context, bacterial genomes are a bargain at appr...

374KB Sizes 2 Downloads 140 Views

pharmaceutical companies each spend almost $1 billion per year on research and developmentl3. In this context, bacterial genomes are a bargain at approximately $lmillion each. It is probable that, in the very near future, complete genome sequencing will become a cost-effective prerequisite to the development of antimicrobial drugs. Therefore, before the decade is out, we can expect that the complete genome sequences of all the usual suspects such as Mycobacterium and Staphylococcus will be known. Let us hope that the sequences make it into the public domain. If they do, then truly we

stand on the threshold of a golden era of prokaryotic genetics.

feremes 1 Fieischmann, K.D. et al. (1995)

8 Wilson, R. et al. (1994) Nature 368, 32-38 9 Devine, K.M. (1995) Trends Biotechnol. 13, 210-216 10 Smith,H.O. et al. (1995) Science

Science269. 496-512

2 Lee,J.J. et al. (1989)J. Bacterial. 171, 3016-3024

3 Dujon, B. et al. (1994) EMBO J. 13, 5795-5809

4 Blackburn, E.H. and Kzrer, KM. (1986) Ann. Rev. Genet. 20, 501-521 Oliver, S.G. et al. (1992) Nature 357, 38-46 6 Feldman, H. (1994) EMBOJ. 13, 5795-5809 7 Vassarotti, A. and Goffeau, A. (1992) Trends Biotechnol. 10, 15-18

5

11

269,538 -540 Parkinson, J.S. (1995) in Two-

Component Signal Trarzsductiol? (Hoch, J.A. and Silhavy, T.J., eds), pp. 9-23, American Society for Microbiology 12 Msadek, T. et al. (1995) in TwoComponent Signal Transduction (Hoch, J.A. and Silhavy, T.J., eds), pp. 447-471, American Society for Microbiology 13 Glaser, V. (1995) Biotechnologll13, 632-635

/

/ /

c

P Classification of LIMproteins A

The LIM domain is defined by a unique cysteine-rich motif consisting of xxCx~Cx,,_~~ HxxCxxCxxCxlGz,Cxx[D/H/C]x (Refs l-3); this domain is found in at least 40 distinct proteins in different eukaryotes (see Refs 4, 5 for additional references). The number of LIM domains within known proteins varies from 1-5, and LIM domains might be associated with various other fUnctiona domains, such as homeodomains, a protein kinase domain, and regions involved in protein-protein interactions; several proteins, however, appear to consist of LIM domains and little else. Although LIM domains bind zinc ions and show structural similarity to zinc fmger&, most evidence favours a role in protein-protein interactions rather than DNA bindiq@7J3. Multiple functions have been ascribed to LIM proteins. Among LIM homeobox (LHX) genes, h-11 and met-3 are involved in cell lineage specificatior172, while the Liml gene functions in muscle and neural induction by the organizer9 and is required for the formation of the fore- and midbrainlo. The LIM-only genes LMO1, -2 (RBIIIvor flTG1, -2) are associated with chromosomal translocations in

T cell leukemiasll-13, and mice in which the Lmo2(Rbtn2) gene has been disrupted die as a result of an early block to erythropoiesisld. While the proteins mentioned above are known or believed to function in transcriptional regulation in the nucleus, other LIM proteins are associated with the cytoskeleton where they mediate cell adhesion and otherproperties4q517. The nature of the LIM domain as a protein interaction motif could account for the broad range of functions in which LIM domain proteins have been implicated. While LIM proteins have been classified primarily on the basis of the presence and nature of associated domains415, such schemes might be refined by considering sequence similarity between LIM domains. At the recent workshop on LIM proteins (Bischenberg, 9-12 May, 19951, we decided to combine our separate sequence analyses (Ref. 5; J-L. Evrard, R. Baltz, V. Bourdon, and A. Steinmetz, unpublished) to provide the classification scheme shown in Fig. 1. Additional sequence analysis of LIM proteins was carried out by K. Kuma, K. Mizuno and T. Miyata (unpublished). TIG NOVEMBER1995 VOL. 11 NO. 11

431

Group 1 proteins always contain paired LIM domains near the N-terminus; many (15 known proteins) contain a homeodomain (LHX proteins). The N-temtial LIM domains fall into one sequence class (type A; Ref. 51, while the

second LIM domains form a distinct class (type B; Ref. 5). A and B domains are always associated. The LHX protein Lmx-1 and LMO (RBTN or TTG) interact functionally or physically with the helix-loop-helix (HLH) proteins Pan1 (E47) and TALl, Group 1

11 It’ t tHDt

1

m A

It 1 a

Ikinasel

[

LHX LMO (RBTN/lTG) LIM k rtase

Group 2 GRIP, DMLPl CRP. CRP2, MLP, (SFB) c Groip rj

I B ril --itI .j YKROSOW 1_ ; 41 i

I

/ . iaP{

1

I

-,

II ij <_ If ij ,-

enigma, Hic5, testin Zyxin, ESP-2, AvLB-1 paxillin PINCH

Ungrouped (yeast GAP proteins) I GAPu Dbmlp II I

n

11

I I GAPuLr!JlP

FIGURE1. Ciassification cf .iJM l,ratcbinc; LIM domains (grey), glycine-rich sequence (dark grey), distinct LIM domain (asterisk) (type E; Ref. 5). Kinase and GAP domains are indicated. Abbreviation: HD, homeodomain.

respectivelyl5916.In addition, group 1 LIMdomains might have a negative regulatory function within LHXproteins, repressing their DNA binding and transcriptional activating abilities (see Refs 4, 5). I.&Eand LMO(RBTNor ‘ITG) proteins are more closely related among and to each other than to LIMK. Group 2 proteins contain one or two copies of a single sequence type of LIMdomain (type C; Ref. 5); all, except the plant protein SF3, contain a short glycine-rich domain, and the distance between domains is larger than in other groups. Group 2 proteins are involved in interaction with identical (e.g. CRP homodimers) or different LIM domains (e.g. binding of CRP to zyxin)7ls;their function might be to regulate the function of other proteins. Group 3 proteins contain LIM domains that are more heterogeneous in sequence than group 1 and 2 domains, and that are localized in the C-terminal region. Non-L&f domains in group 3 are implicated in binding other proteins, such as actinin, vinculin, focal adhesion kinase, insulin receptor, and SH2- and SH3-

containing proteins (see Refs 4, 5). Known group 3 LIMproteins interact with the cytoskeleton or membranous organelles and might function in adhesiveness, defining cell shape, and in intracellular trafficking. Many, but not all, of the LIMdomains in this group can be assigned to a sequence class (type D; Ref. 5), but several subclasses might be recognized (J-L.Evrard et al, unpublished). Two yeast proteins containing unrelated N-terminal LIMdomains and a GAP domain remain unclassified at this time. While some issues remain unresolved, this classification for LIMproteins combines sequence C similarity with structural considerations to provide a scheme that organizes this complex class of proteins in a functionally meaningful manner. MASANORI TAM*, JEAN-LucEVMRD$ ANDRB sTElNMETz*AND

IGOR B. DAW*

Qboratoty of MolecularGenetics, National Institute of ChildHealtband Human Development, NIH,Bethesda, m 20892, USA; htitut de Biologic Moltkulaire desPlantes, Centre National de laRechercbeScientijique, 67084 StrasbourgFrance.

I Returnto sobrietvafterthecatalytic partyn4

I

The apparently non-uniform mode of strand cleavage, cfs or trans, among members of the so-called Int family of recombinases has been irksome. Stark and Boocock, after gate crashing at the catalytic partyl, offer three potential resolutions to the cis-tram paradox: (1) the abnormal mode of cleavage (be it c&or trun$ results from the nature of the specially configured substrates used in several of the assays; (2) the ‘wrong’cleavage is executed by an exposed surrogate nucleophile from the recombinase; and (3) a dual cleavage mode (cis and tram), if this does occur, reflects the different functions of the

recombinase monomers during different stages of the reaction. The notion that the DNA substrate might influence the assembly of the recombinase active site is a valid one. As the ‘Int family recombinase’ binds to its DNA target as a monomer, building an active site at the interface of a recombinase dimer, for example, will depend on whether the DNA arms have the requisite stacking freedom. For the Flp system, however, half-sites (one Flpbinding armj2, full-sites (two Flpbinding arm@, Y junctions (three Flp-binding arms) and Holliday junctions (four Flp-binding arms) TIG NOVEMBER1995 VOL. 11 No. 11

432

References 1 Way, J.C. and Chafie, M. (1988) Cell

54,~16

2 Freyd, G., Kim, S.K. and

Horvitz, H.R. (1990)Nature 344,

876-S79

Karlsson, 0. et al. (1990)Nature

344,879-882 Sgnchez-Garcia, I. and Rabbitts, T.H. (1994)Trends Genet. 10,315-320 Dawid, I.B., Toyama, R. and Taira, M. (1995)CR. Acad. Sci. Paris 318,295-306 Perez-Aivarado,G.C. et al. (1994) Struct. Biol. 1, 388-398 Schrneichel,K.M. and Beckerle, M.C. (1994)Cell79,211-219 Feuerstein,R. et al. (1994)Proc. Nat1Acad. Sci. USA91, 10655-10659 9 Taira,M. et al. (1994)Nature372, 677-679 10 Shawlot,W. and Behringer,R.R. (1995)Nature 274,425-430 11 McGuire,E.A.et al. (1989)Mol.Cell. Biol. 9, 2124-2132 12 Boehm,T. et al. (1990)EMBOJ. 9, 857-S68 13 Sinchez-Garcia,I. and Rabbit&T.H. (1993)Cancer Biology 4,349~358 14 Warren, A.J.et al. (1994)Cell 78, 45-57 V.E.et al. (1994) 15 Valge-Archer, Proc. Nat1Acad. Sci. USA91, 8617-8621

16 German, M.S. et al. (1992) Genes Dev. 6,2165-2176

0. Lee and M. Jayaram, unpublished) have so far revealed only trans DNA cleavage. The reasoning by Stark and Boocockl that the ‘wrong’cleavage could be a wanton trick played by a surrogate nucleophile does not persuade us. The active site tyrosine has been biochemically mapped for Int (Tyr342)4 and for Flp (Tyr343)5. They correspond to the only invariant tyrosine of the Int family. There is strong evidence that Flp mediates trans-cleavage via Tyr343 (Ref. 6). The nearest tyrosine neighbor of Tyr343 in Flp is Tyr354, followed by Tyr361, Tyr362 and Tyr364. Changing each of these four residues to phenylalanine has no effect on the recombination reaction as long as