An SH2-SH3 domain hybrid. Srn - We report a striking similarity between the three-dimensional structures of the src homology 2 domain (SH2)r and domain II of ...
SCIENTIFIC CORRESPON DENCE
AnSH2-SH3domainhybrid Srn - We report a striking similarity betweenthe three-dimensional structures of thesrchomology2 domain(SH2)rand domainII of the repressor of the Escherichia coli biotin biosynthetic operon (BirA)2. We used a- protein-structure comparisonalgorithmr to compare the X-ray structure of the chicken src SH2 domainto a non-redundantsubsetof the Brookhaven protein databanka. and found that BirA and SH2share51 equivalent Cry atoms that superimposewith an r.m.s. deviationof2.I3 A. Four strandsof a centralB-sheetand two helicespacking on eithersideofthis sheetlie in equivalent positions in the structure. BirA has an insertion of 49 residuesrelative to SH2 betweenthe third and fourth strands.The insertionincludesan s- helix and threepstrands which extend the SH2 domain sheet (see figure). An analysisof other
r
1
p p p
0
2 0 c c c c c o c s
SH2 sequencess shows that one, avian tensin, has an insertionof 22 residuesat the samelocation.This suggests that the insertion in the tensin SH2 domain may extendthe p-sheetin a similarfashionto BirA. Noble et al.6 have drawn attention to the similarityof domainIII of BirA to the SH3 domain structure. Taken together with our findings, this shows that BirA comprises an SH2-SH3 domain hybrid with an amino-terminal helix-turn-helix domain. There are manv examolesof SH2- and SH3-containingproteinsT,but no structure of an SH2-SH3 complex has yet been determined.BirA thus provides a possiblestructural model for the SH2SH3 complex. Both SH2 and BirA bind phosphate. SH2 binds phosphotyrosine-containing peptides, whereas BirA catalysesthe
3
0
4 p p p p F
0
sH2 E wylF-GTT-Tl R RIET-E-F-nL L NIF=-NT'TG-r-FIT-F_EIS Er r K BirA L D G G s v A V t L p v t p l s T N o t y L L p R t t l c E L K s G p a c t a E l y o o A G [ 1 1 ] p p p
p p F p
c o c c q c
60
F p F p p p p p F
70 8o p p F p p p F p p p p
SH2 GAIT-T5V-S-D-FIDNA KG B n AP F G A N t L YL S I T F W R IELO [ 4 9 1 P A A
B p p p p p p p 9 p 100 p F
c c c c c q c c
90
p p
D S G G F Y I M [ 1 2 l W l T L Q E A G
p p F p F p F p p p
110
sHz r s R r a F s s LIO-d-L-T-IT-YTRT-il B i r AI
NL DRNT
LAlAttLl
formation of biotinyl-5'-adenylate from biotin and ATP. The interaction between phosphotyrosineand SH2 is well understood''o'". but for BirA structural evidenceis aviilable only for the biotinenzyme complex (not for the ATPenzymecomplex).Biotin bindsto BirA on domain II on one face of the central F-sheet,and near the amino terminusof the first helix. Surprisingly,this is nearly identicalto the binding site for phosphotyrosineon SH2. Although BirA doesnot seemto have any of the residuesknown to interact with phosphatewithin the SH2 domain, the ATP-bindingsite is thought to be in a disorderedloop region containing arginine and glycine near the FLVRESET phosphate-bindingregion on SH2. The structural similarity suggests that BirA undergoes a conformational rearrangementsimilar to that of SH2 on binding to phosphate. No prokaryotic counterpart to the SH2 and SH3 domains has oreviouslv been identified. Although the functionsare different,the structuralsimilarity,similar binding site and coincidenceof both SH2 and SH3 domains in a single molecule describedhere, suggeststhat'BirA may share a common ancestorwith SH2 and SH3 domains.It is tempting to suggest such an evolutionary link between the SH2 and SH3 domains. which are involvedin variousstages of thecytoplasmic signallingcascade"', and a transcriptional regulator,whichhasultimatecontrolover gene expression. RobertB.Russell GeoffreyJ.Barton Laboratoryof MoIecuIar Biophysics, Universityof Oxford, SouthParksRoad, OxfordOX73QU,UK
RELRAALI
c c c c c c c c c c c c c c c c
1. Waksman,G. etal. Natule358,64M53 (1992). 2. Wilson, K- P. et a l. Proc.natn.Acad.Sci. U.S.A.89, 9257-926L(7992\. 3. Russell, R. B. & Bafton,G.). Protei ns:Structure, Functionand Genetlcs14, 3O9 323 (1992). 4. Bernstein,F. C. et a L J. molec. Biol. ll2, 535 542 (r977\. 5. Russell,R. B. et a l. FEBSLett. 304, t5 20 (L992). 6. Noble,M. E. M. et a t. EMBOJ. 42. 26L7-2624 (L993\. 7. Heldin,C. Trendsbiochem. Scl.16, 450-452 (1991). 8. Eck,N,f . J. et al. Nature362,87-91(1993). 9 . W a k s m a n , G . , S h o e l s o nE, S . , .P a n t , N . , C o w b u r n , D . & Kutiyan,J. Cell72,779-790 (1993). 10. lvccormick, F. Nature363, 15-16 (1993). 11. Kraulis,P.J.-/.Appl. Cryst.24, 946-950 (1991). 12. Kabsch,W. & Sandet, C. Biopolymers 22, 257 7 -2637 (1983).
a, Molscriptll drawing of the three-dimensional structure of BirA2, excludingthe amino-terminal helix-turn-helixdomain.Theshadedregionsshowthe SH2-likedomainll, and the SH3-likedomainlll. Theunshadedregioncorresponds to the 49 residueinsertion relativeto the srcSH2domain.Dashedlinesin BirA,regionsforwhichno electrondensityis seen.Theexperimentally determinedlocationof boundbiotinylated lysineis shownin stick form. b, The structure of src SH2 domainl in the same orientationas a, showingthe experimentally determinedlocationof bound phosphotyrosine. Shadinghightightsthe regionssimilarto BirA.c, Structure-derived sequencealignmentof SH2 and the BirA equivalentdomain3.Boxed bold-faceregions show structuralequivalences.Seconoary structure assignmentswere made using the program DSSP12:a, u-helix;B, B-strand. Missingresiduesin BirAat alignmentpositions43 and83 aredueto poorelectrondensity in these regionswithin the X-raystructure.The proposedphosphatebinding loop is at position 43 and containsthe sequenceGRGRRG. 49 residueswere removedfrom BirAat position65 for clarity.Only8 of the 51 structurallyequivalentresiduesare identical.
. VOL364 . 26 AUGUST NATURE 1993
765
