- Email: [email protected]
Structural model for the selenocysteine-specific elongation factor SelB
Biochimie (1996) 78, 971-978 © Soci6t6 franqaise de biochimie et bioiogie molEculaire / Elsevier, Paris
Structural model for the selenocysteine-specific elongation factor SelB R Hilgenfeld a, A B6ck b, R Wilting b* alnstitut fiir Molekulare Biotechnologie eV, Beutenbergstra~e 11, D-07708 Jena; bLehrstuhl fiir Mikrobiologie der Universitiit Miinchen, Maria-Ward-Strafe la, D-80638 Miinchen, Germany (Received 10 November 1996; accepted 29 November 1996)
Summary m A structural model was established for the N-terminal part of translation factor SelB which shares sequence similarity with EF-Tu, taking into account the coordinates of the EF-Tu 3D structure and the consensus of SelB sequences from four bacteria. The model showed that SelB is homologous in its N-terminal domains over all three domains of EF-Tu. The guanine nucleotide binding site and the residues involved in GTP hydrolysis are similar to those of EF-Tu, but with some subtle differences possibly responsible for the higher affinity of SelB for GTP compared to GDP. In accordance, the EF-Tu epitopes interacting with EF-Ts are lacking in SelB. Information on the formation of the selenocysteyl-binding pocket is presented. A phylogenetic comparison of the SelE domains homologous to EF-Tu with those from EF-Tu and initiation factor 2 indicated that SelB tbrms a separate class of translation factors.
selenocysteine / elongation factor / protein synthesis / tRNA / prokaryotes Introduction The product of the selB gene from Escherichia coli is required for the cotranslational insertion of selenocysteine into protein tfor a review see [1]). Both genetical and biochemical lines of evidence indicate that SelB has the function of a specialized translation factor homologous to that of elongation factor Tu (EF-Tu). In vitro SelB forms a quaternary complex with selenocysteyl-tRNA sec, GTP and a secondary structure of the selenoprotein mRNA immediately downstream of the UGA codon directing selenocysteine insertion 112-4]. This preformation of the quaternary complex which requires stoichiometric amounts of the individual components is essential for the decoding process in vivo [5]. A recent analysis of the sequences of four bacterial SelB species has shown that the N-terminal part of the molecule is homologous to the three structural domains of EF-Tu [6]. The C-terminal part, on the other hand, displays no sequence similarity to any other protein known. It functions as an RNA binding domain attaching the translation factor to the recognition sequence of the mRNA. SelB in comparison to EF-Tu displays a number of intriguing characteristics. It binds GTP with on approximately 10-fold higher affinity than GDP and, therefore, does not
require any guanine nucleotide release factor to displace GDP [7]. Moreover, SelB only recognizes tRNASec when it is charged with selenocysteine and it does not interact with any significant affinity with the 20 ,~tandard aminoacyltRNAs [8]. To gain somt: insight into the structural basis of these differences in the absence of a three-dimensional structure for SelB we have established a three-dimensional model on the basis of the coordinates of the EF-Tu structure.
Materials and methods Modeling of SelB The high-resolution crystal structure of the Er:-Tu from Thermus thermophilus in complex with a GTP-analogue [91 served as a template for constructing the model for SelB. The available SeIB sequences were aligned with the amino acid sequences of EF-Tu from E coli and T thermophilus. The multiple alignment was revised when suggested by inspection of the three-dimensional structures (fig 2). Visualization and model building was carried out using Frodo [ 10]. The polypeptide geometry around deletions and insertions was initially regularized by the method of Hermans and McQueen [! 1] as built into Frodo. Backbone atoms belonging to topologically equivalent residues as well as identical side chains and common atoms in non-identical side chains were transferred from the EF-Tu structure. The remaining atoms were positioned by calculation of the potential energy as a function of side chain torsion angles [12]. Finally, the entire model was subjected to energy minimization.
Genetic distance calculation *Correspondence and reprints
The Genworks 2.4 program (IntelliGenetics, Inc) was used for genetic distance calculation for the phylogenetic tree (fig 4).
972 Results and discussion
A structural model for Se/B On the basis of the recent elucidation of the three-dimensional structure of EF-Tu in its active GTP-form [9, 13] and its inactive GDP-form [14] and of the significant sequence homology between EF-Tu and SelB [6], a three-dimensional model for the three N-terminal domains of E coli SelB was established. The model was constructed using methods suitable for deriving a structure when that of a homologous protein is known [121. Protease treatment of purified E coli SelB confirms the structural model: trypsin was found to cut between Arg-34 and Gly-35 at the end of helix A I" and between Arg-152 and Gly-153 in the FI helix [151. These positions are situated in solvent exposed regions in the model structure. Secondary structure Domain 1 of E coli SelB has largely the same overall fold as the nucleotide-binding domain (domain 1) of EF-Tu. Both proteins share 35% identical amino acid residues in this domain. However, in SelB major deletions of polypeptide stretches are found (fig 1). The amino-terminal deletion of 11 residues most probably does not influence the structural integrity of ~he domain. On the other hand, in EF-Tu this stretch makes important interactions with domain 2 and ties the two domains together; consequently, this will not occur in SelB. A deletion of 14 residues in SelB comprises not only parts of the A l-helix, but also the following extended chain and the short A I' helix, both of which are believed to be a part of the so-called 'effector region' in EF-Tu (residues {42 ] to {62 ] (residues of SelB are numbered according to the sequence of E co/i SelB and corresponding positions of T thermophilus EF-Tu are given in brackets)) [ 161. In contrast, the A" helix, located at the rim of the tRNA binding cleft, is completely conserved. The following 13-strands b l and c i, and the B I helix which undergoes an important change in orientation upon GTP hydrolysis [91, are also well conserved. The same is true for the completely solvent-inaccessible ~-strand d l, the C I strand and the DI helices and I~-strands e I and fl. However, beyond the fl strand, as many as 20 residues (from { 1751 to { 1941) are absent in SelB. This means that nearly the entire 10-residue El helix, the loop-like structures following it and the first residue of the F I helix are missing. It is worth noting that in EF-Tu, this portion is intimately interacting with the C-terminal half of the A I helix, which is also absent in SelB. All the structural elements present in EF-Tu but absent in SelB are located on one side of the molecule (fig 2A, B). The secondary structure of domain 2 (25% sequence identity between E coli SelB and T thermophilus EF-Tu) is structurally much better conserved than that of the nucleotide-binding domain. His- 1273 } and Gin- 1278 }of EF-Tu,
located in a turn that interacts with the N-terminal residues of the chain, are missing in SelB. As in domain 2, all secondary elements known from EF-Tu are conserved in domain 3 of SelB (22% sequence identity), and deletions and insertiens are only found in loops at the surface of the molecule. A major deletion of eight residues concerns a loop (residues ! 323 } to {330} in EF-Tu) that interacts, via water molecules, with domain 1 in EF-Tu. Another eight-residue deletion (residue {358 } to {3651) is located in a surface loop very ~,Iose to the previous one. Since the deletions in domain 3 are located at the far end of the molecule compared to the c|~,anges in domain I, it is possible that they mark the sites oi' interactions with the C-terminal domain 4 of SelB which cot,dd not be modeled in this study due to the lack of a template. Nuc!eotide binding The binding mode of GTP is similar in EF-Tu and SelB. The phosphate b i n d i n g Io~3p, c o n t a i n i n g the canonical GXXXXGKT (G l-box), i~',well conserved, and so are the residues in the first (Thr- 141251, Thr-37 {621) and second (Asp-26 {511, Asp-57 {81 }) ceordination sphere of the Mg2+ ion (fig 3A, B). The latter is paxrtof the DXXG motif (G3box), the glycine of which (Giy-60{ 84}) plays a pivotal role in the GTPase switch mechanism ,I!71. As in EF-Tu, the guanine base is specifically recognized by the conserved Asp- ! ! 51 ! 39 }, ~!ia hydrogen bonding with the N i and N2 atoms (f;g 3). h~ most GTPases, this residue is part of the NKXD motif (G4-box). in EF-Tu Asn{ 136 } makes important interactions \with other nucleotide-binding segments of the polypeptide c'hain, such as the main-chain carbonyi of residue {22} and ~he main-chain amide of residue I 1751. in SelB this asparag{ne is replaced by Thr-112, the side-chain hydroxyl of which can still donate a hydrogen bond to the main-chain carbonyl of residue {221. The other interaction is missing, because the polypeptide chain takes a different course at the residue corresponding to Ala-{ 1751, Glu-150 in SelB. As a result, the E coil SelB lacks the hydrophobic interaction between the guanine base and Leu-{ 176} of the 'SAL'-motif found in EF-Tu. This is due to the complete deletion of the EI helix in SelB~ which forces the polypeptide chain to enter directly into the FI helix at this location. Therefore, GTP and GDP binding to SelB can be expected to be weaker than that to T thermophilus EF-Tu, and this was also observed experimentally [71. In spite of that, the side-chain of Thr-149 can still donate a hydrogen bond to the 06 atom of the guanine base. Interaction with EF-Ts The crystal structure of E coli EF-Tu in complex with EF-Ts revealed that the contact sites in EF-Tu are located in domains 1 and 3 [181. Only a few of these residues in domain 1 are present in E coli SelB, whereas the contacts in domain 3 (Phe-{335 }, Met-{361 }, and Met-{3631), are absent. The
973
SELB_EC
EFTU_TT
SELB_EC
EFTU_TT
SELB_EC
flREGEFURTE
~
34
PHUHVG
PHVEVEDYGD
LGYRY TAHVE
gPQPDI~PG F ' I ~ K F
LS~LIF~GG
VETeK-vS
94 119
I l eQ
EFTU_TT
~ILLRRQV~ ~/IU~flH~
SELB_EC
. . . . . . EGRG flDA~FIEHLLQ L..-~EREHASQ LEQMHRHPKT RRGEHEWVOE IMEW.DAIDE YI~JTPVRDVD
KP~_MPVEDU
TGTALSGEVE TGRIERGKVE
TRHRGQR iRL EGIAGDHVGV
EFTU_TT
SELB_EC
EFTU_TT
-
SELB_EC
EFTU_TT
SELB_EC
Tu_TT
68
. - - . ~ ~
VGDSLULTGV VGDEVEIVGL
~II~PELL~
H~GDEV
H-EPflRVRRL HRQH-QPTERPETRRTVVT GVEMHRETLQ
PVIRGSRLLA
149 179
239
H iflGDREKEQ LLRG-VSREE
249 298
~ii~--~
c
I HRGDWLLAD VPPE~F~VI VELQT. . . . . . . -~PLTQW UERGQVLaKP GSI T~JH~FE ASVYVLKI(EE GGR~GFFS6
~
L-AELVFDT~ LI,I~DHD~V
L~IS~L
UTFTUELIKW U, EGL . leEG-
.
AGARJ~JilLHP PRRGKRKPEY 352 GAGV~i( ILE
486
SELB_EC
LQWLASLARA QSDADALSVH LERGAVNLAD FAWARQLNGE GMRELLQOPG YIQAGYSLLH
412
SELB_EC
APVAARWQRK
ILDTLATYHE
QHRDEPGPGR
ERLRRMALPM
EDEALVLLLI EKMRESGDIH
472
SELB_EC
SHHGWLHLPD
HKAGFSEEQQ
AIWQKAEPLF
6DEPWWVRDL
AKETGTDEQA
MRLTLRQAAQ
532
SELB_EC
QGIITAIVKD
RYYRNDRIVE
FANMIRDLDQ
ECGSTCAADF
RDRLGVGRKL
AIQILEYFDR
592
SELB_EC
IGFTRRRGHD
HLLRDALLFP
EK
614
Fig 1. Alignment of SelB from E coli (EC) and EF-Tu from Thermus thermophilus (TT) based on a multiple alignment of four SelB sequences and two EF,~2, sequences [6]. Identical residues were colored red for domain l, yellow for domain 2, blue for domain 3 and orange for domain 4. Rouo4ed bars indicate o~-helices, arrows indicate D-sheets of EF-Tu.
974
~i~!~i~¸ !i ¸¸~i¸i¸~i¸i ¸I!' !i~iii i ~ i¸¸!~!~ ~i ~ ......
EF-Tu
E. coli
SELB
Fig 2. Structural comparison of EF-Tu and the homologous part of SelB. A. Three-dimensional structure of T thermophilus EF-Tu [9]. B. The three-dimensional model of the EF-Tu-homologous part of E coil SelB. Only the polypeptide backbone is shown. Colors: red, domain I; yellow, domain 2; blue, domain 3; green, GTP and magnesium ion.
assumption that SeiB is independent of EF-Ts is further supported by biochemical data: SelB is the only translation factor that has a higher affinity for GTP (Kd = 1.7 mM, similar to E coli EF-Tu) than for GDP (Kd = 10 mM, 1000-fold higher compared to E coli EF-Tu), therefore the nucleotide exchange will occur spontaneously [7]. Binding of aminoacyl-tRNA
EF-Tu binds all aminoacylated tRNAs except formylmethionyl-tRNAMcti which is bound by initiation factor 2, and selenocysteyl-tRNAS~, which is bound exclusively by SeiB, Information on the selenocysteyl-tRNA binding site can be obtained from a comparison with the crystal structure of the ternary complex of Phe-tRNA phc, EF-Tu and GppNHp [ 19]. The binding site for the CCA-Phe end on EF-Tu is formed by the cleft between domains 1 and 2, as has been predicted earlier [9]. The same binding site is also found in the structure of the complex between EFTu,GppNHp and anthranyi adenosine-5'-monophcsphate, a compound regarded as a model for the 3'-aminoacylated adenine of the tRNA (Hilgenfeld et al, manuscript in preparation). In both EF-Tu complexes, the pocket for the amino acid (phenylalanine and anthranylic acid, respectively) is lined by the side chains of Phe- {229 }, Thr- {239 }, Glu1226}, Asp-{2271, His-167} and His-{273} (table I). In SelB, the two former of these residues are conserved (Phe183 and Thr- 193, respectively), while Glu {226 } is replaced by Asp- 180 and His-167 } by Tyr-42. Interestingly, Asp1227 } becomes Arg- 181 in all four SelB sequences, which might protrude into the binding pocket and restrict its size. In addition, its guanidinium group might interact with the negatively charged selenocysteyl side chain (pK ~6.5). Both factors
might contribute to the specificity of SelB for selenocysteine, which can be tested by site-directed mu,tagenesis. In EF-Tu, His-| 2731 is located at the tip of a loop that protrudes into the cleft between domains 1 and 2, thereby closing the amino acid binding site at its bottom (orientation as in fig 2A, B). In the ternary complex between PhetRNA pheand EF-Tu,GppNHp, main chain amide and carbonyl groups from this loop interact with the aminoacyl ester of the tRNA. In SelB, however, the loop is shorter and has a different conformation. As a result, there is no residue in E coil SelB that would correspond to His- {273 } of EF-Tu, so that the amino acid binding pocket is much more open in SelB as compared to EF-Tu. This effect is enhanced by the deletion in SelB of both the eleven N-terminal residues and much of the 'effector region', which in EF-Tu form the entrance of the tRNA-binding cleft. It is possible that these deletions allow portions of domain 4 of SelB to approach the amino acid binding site and are involved in determining the specificity for selenocysteyl-tRNAsec. The only base of the tRNA found to interact with the EF-Tu in the crystal structure of the ternary complex [ 19] is the 3'-terminal adenosine monophosphate. In this complex and in the complex with anthranyl-5'-adenosine monophosphate (Hilgenfeld et al, in preparation) it interacts with the hydrophobic side chains of Val- {237 }, lie- {231 }, and, in part, Leu-{289} on one side, and with Glu-{271 } on the other. In SelB the valine is conserved (Val-191), whereas the isoleucine is exchanged for Val-185, the leucine replaced by Asn-240, and the glutamate is replaced by Gin224. Therefore, with the exception of the exchange of Leu to Asn, SelB provides an environment for the adenine base that is similar to that provided by EF-Tu. Similarly, residues seen to interact with the 5'end of the tRNA in the ternary
975 Table I. Residues involved in aminoacyl-tRNA binding of T thermophilus EF-Tu and corresponding residues of the four SELB sequences. tRNA binding residues a. Adenine 76-interactions EF-Tu
SELB
lie [231 ] Val [2371 Glu [271 ] Leu [289]
Val/Met [ 1851 Val [1911 Gin [2241 Asn [2401
Evolutionary considerations
b. Backbone interactions
Lys [52] Glu [55] His [851 Asp [871 Tyr [881 Lys [90] Gin 1911 Arg [30011 Arg [330] His [331 ] Gin [341] Thr [3501 mcGly [391 ]
Arg/His27 Glu30 His61 Lys/Arg63 Phe64 Ser/Lys/Arg66 Gln/Asn67 Arg252 -/Arg27 ! h His/-272 His285 Th~'394 lle/Glyfl'yr326
Aminoacyl-binding pocket
His 1671 Glu 1226] Asp [227] Phe [229] Thr [2391 mcHis [2731 mcArg [2741 mcAsn 12851
EF-Tu complex [19] are largely conserved in SelB (see lable I). In conclusion, SelB will bind the Sec-tRNAS~ in a way very similar to the way EF-Tu binds aminoacyl-tRNA, b7 vivo analysis supports this conclusion, because with C-terminally truncated SelB-derivatives it was demonstrated that the tRNA binding site is located in the first 378 amino acids of SelB, which corresponds to the EF-Tu homologous part [5].
Tyr/Phe42 Asp180 Arg 18 ! Phe 183 Thr 193 -/His225a GIn/-/Gly226 Arg236
Table II. Amino acid identity between the four domains of SELB from E coli (EC), C thermoaceticum (CT), D microbium (DB), and H influenza (HI). Domain SELB
1--4 (%)
1-3 (%)
4 t%)
CT-EB CT-DB DB-EC CT-HI DB-HI EC-HI EC-HI-DE .CT
31 40 30 32 27 42 15
39 50 38 42 36 49 23
22 26 21 20 16 34 6
The conservation of both primary and three-dimensional structure of the N-terminal 3-domains in comparison to that flom EF-Tu allows the determination of the phylogenetic relationship between SelB and other translation factors. A phylogenetic tree (fig 4) of the amino acid sequences of EF-Tu and the homologous parts of the SelB from Clostridium thermoaceticum, Desulfomicrobium baculatum [6] and Haemophilus influenzae I21] and initiation factor 2 (IF2) was calculated using the Geneworks 2.4 program. The parts of IF-2 homologous to EF-Tu comprise residues 386749 of E coli IF-2 and residues 237-587 of Bacillus stearothermophilus (BS) IF-2. To compare the genetic distances among the SelBs to those of related proteins, two EF-Tu and IF-2 sequences from phylogenetically well separated bacteria were included into the calculations. Taking the branching point error (fig 4: numbers in brackets) into account, the SelB and EF-Tu sequences are slightly closer related to each other than either of them is to IF-2. When the distance values of the SelB sequences are compared it is obvious that SelB is not derived from EF-Tu of the corresponding organism, but that the SelBs form a distinct class of translation factors. The genetic distances between the SeiB sequences are in accord with the phylogeny of the organisms as delineated from the rRNA sequence [20]. This is true for domains 1 to 4, 1 to 3, and 4 alone. Although the C-terminal half of SelB might be younger than domains I to 3, this means that it also has evolved before the phyiogenetic separation into the bacterial subgroups. Whereas domains l to 3 of the SelB sequences share 23% sequence identity, domains 4 only contain 6% identical amino acids (table II). In our model, the domains 1 to 3, as in EF-Tu, have to interact with the ribosome, guanine nucleotides and the tRNA, therefore a higher structural constraint is necessary than for the mRNA binding domain 4. The flexibility in sequence of the mRNA-binding domain is obvious in case of E coli and H influenzae, because they share only 34% identity in domain 4 (table II), although the mRNA stem-loops offdnG of E coli and the homolog of H influenzae (deduced from ORF H influenzae 00006 [21]) differ in only three out of 45 bases.
Conclusions
The model of E coli SelB showed that its N-terminal part is homologous to EF-Tu. The guanine nucleotide binding site
976
A
| HN "'
ILYS241
I
Imc(~ly84~NH. , •
N.-H3 o.'~)'-.
imcLys24[
.HN
i ."" . ..°
. H O ~
'..o-' '-~,I " i I ..
• ""
.o. ~o. r-~
:
.
.,.,
,Thr62,Od'"
,
.,
,,
t
.::lm©Sly231 i
."
"
,~. |
."
i!"
/:. ~o.
~
".tm~Th'=S]
:
I
,
.~
.,
.H3N
:
""" ~
~;.
?
~ N . .
ImcLysl 3
:
' IAsnl=Sl
.....
.'"'
i
n .olmcAspl0l. O •"
• ""
IThr371.OH
"
".
: N,H
,, - -
--
"
;:lm©Sly12i
"
i'
" ,
.
: , i
,,o.
~
':
NH I
"~n~
."
"
o
."
~
~
NH .....
| ~=,~ t~J~~.
;..
'
,k
I ~..,,,~t
:
'
imcThrtSl
. . ~ .... o~
~
~
O .... HO~"S'I
."
. ,
.o,
. ~ |
'
;
""
[
o
O ' m m " P m O ' - " P " " ' O I I '. |
u
....
HN
"'"'.~'"'~'""'.~'"'"': IThrST[NH
;'HN~mcAIa17si
i
"
"."3
. o ~
"'"
......~%,o °
I-°.,.=~1
ILys13i I
_. .....
::
Im~A*P='I
B lmcGlY121 I HN.
.... o~ ~o4,.,,-~
"
"
"m¢i:r26'
j,., ,-
" +r~.P
',1
H3N
'
,-.
....
i
_g O
, ,
O ......
HO
: '
"
." O'" , i
.-"HO ,,
[T'C~ou~ ~N4mcThr14[ Im=A'pl01
?"
o.,o..
ImoHislll Fig 3. Detailed comparison of the GTP-binding amino acids of T thermophilus EF-Tu (A) and E coil SelB (B). Deviations in SelB are shaded.
is similar to that of EF-Tu, as are the amino acids involved in the GTPase mechanism. SelB contains no EF-Ts binding site; a nucleotide exchange factor is not required since SelB has a lower affinity for GDP than for GTP. The mode of tRNA binding can be expected to be similar in EF-Tu and
SelB, although the aminoacyl-pocket of SelB is more open than that of EF-Tu and contains an arginine residue that might be involved in selenocysteyl-recognition. Deacyiation kinetics of selenocysteyl-tRNA in complex with SelB and of seryl-tRNAser bound to EF-Tu indicated the same
977
0.272
N'-SELB_CT
0.272
N'-SELB_DB
o.101 (eo.020) 0.198
(+_0 +22)
0.266
N'-SELB_EC
0.266
N'-SELB_HI
0.161
EF-TU_EC
0.107 0.059
(+0.049) 0.409
"~(+0.015) O.161
m
(+0.061) 0.422
~
'(+_0.017)
EF-TU_TT
0.2fH
N'-IF-2_EC
0.207
N'-IF-2_BS
Fig 4. Phylogenetic tree of the EF-Tu-hemologous parts of SelB, EF-Tu, and IF-2 sequences. CT, Clostridium thermoaceticum, DB, Desulfomicrobium baculatum; EC, Escherichia coil; HI, Haemophilus influenzae; "IT, Thermus thermophilus; BS, Bacillus stearothermophilus.
degree of ester bond protection [22]. Therefore, it is unlikely that the aminoacyl-binding pocket in SelB is more accessible to the solvent. Instead, it can be assumed that another part of SelB, probably domain 4, 'closes' the pocket. Domain 4, which has no counterpart in EF-Tu, is much less conserved among the four SelB molecules than the other three domains. This divergence may be related to the fact that this domain interacts with the selenoprotein mRNA [6] and thereby reflects the co-evolution of the different mRNA recognition sequences with the cognate translation factor. This is supported by our recent finding that SelB from D baculatum is unable to functionally substitute for the E coil SelB in selenoprotein synthesis in E coil [23]. On the basis of the structural model presented, targeted mutational analysis of residues involved in recognition of selenocysteyl-tRNA sec can be performed for its proof.
Acknowledgments We thank H Schtitz for help with computer calculations. This work was supported by grants from the Deutsche Forschungsgemein-
schaft (SFB 369), from the DESY-PS, 7rom European Commission via its Human Capital and Mobility Programme, and from the Fonds der Chemischen lndustrie.
References ,~ Baron C, B6ck A (1995) The selenocysteine-inserting tRNA species: Structure and function. In: tRNA: Stnwture, Biosynthesis and Ftowtion (S011 D, RhajBhandary U, eds) American Society for Microbiology, ASM Press, Washington DC, 529-544 2 Heider J, Baron C, B~ck A (1992) Coding from a distance: dissection of the mRNA determinants required for the incorporation of selenocysteine into protein. EMBO J i 1, 3759-3766 3 Baron C, Heider J, B0ck A (1993) Interaction of translation factor SelB with the formate dehydrogenase selenopolypeptide mRNA. Proc Natl Acad Sci USA 90, 4181- 4185 4 Ringquist S, Schneider P, Gibson T, Baron C, BOck A, Gold L 11994) Recognition of the mRNA selenocysteine insertion sequences !~y the specialized translation factor SelB. Genes Dev 8, 376-385 5 Tormay P, Sawers A, B6ck A (1996) Role of the stoichiometry between mRNA, translation factor SelB and selenocysteyl-tRNA in seleaoprotein synthesis. Mol Microbiol 2 i, 1253- i 259 6 Kromayer M, Wilting R, Tormay P, B6ck A (1996) Domain structure of the prokaryotic selenocysteine-specific elongation factor SelB. J Mol Biol 262, 413-420
978 7 Forchhammer K, Leinfelder W, B0ck A (1989) Identificationof a novel translation factor necessary for incorporation of selenocysteine into protein. Nature (Lond) 342, 453-456 8 F0rster C, Ott G, Forchhammer K, Sprinzl M (1990) Interaction of selenocysteine-incorporating tRNA with elongation factor Tu from E coli. Nucleic Acids Res 18, 487-49 l 9 Berchtold Hi Reshetnikova L, Reiser CO, Schirmer NK, Sprinzl M, Hilgenfeld R (1~3)Crystal structure of active elongation factor Tu reveals major domain rearrangements. Nature 365, 126-132 l0 Jones TA (1985) Interactive computer graphics: FRODO. Methods Enzymol 115, 157-171 I l Hermans J, McQueen JE (1974) Computer manipulation of (macro) molecules with the method of local exchange. Acta Crystallogr Sect A 30, 730 12 Summers NL, Karplus M (1989) Construction of side-chains in homology modelling. Application to the c-terminal lobe of rhizopuspepsin. J Mol Bio1210, 785--811. 13 Kjeldgaard M, Nissen P, Thirup S, Nyborg J (1993) The crystal structure of elongation factor Tu from Thermus aquaticus in the GTP conformation. Structure l, 35-50 14 Kjeldgaard M, Nyborg J (1992) Refined structure of elongation factor EF-Tu from Escherichia coli. J Mol Bioi 223, 721-742
15 Forchhammer K, Riicknagel K-P Brick A (1990) Purification and biochemical charactedsation of SelB, a translation factor involved in selenoprotein synthesis. J Biol Chem 265, 9346-9350 16 Peter ME, Schirmer NU, Reiser COA Sprinzl M (1990) Mapping of the effectorregion in Themms thermophilus EF-Tu. Biochemist~. 29, 2876--2884 17 Hilgenfeid R (1995) How do GTPases really work? Nature Struct Biol 2, 3--6 18 Kawashima T, Berthet-Colominas C, Wulff M, Cusack S, Leberman R 996) The structure of Escherichia coli EF-Tu. EF-Ts complex at 2.5 resolution. Nature 379, 511-518 19 Nissen P, Kjeldgaard M, Thirup S, Polekhina G, Reshetnikova L, Clark BFC, Nvborg J (1995) Crystal structure of the ternary complex of PhetRNAPhr:.EF-Tu and a GTP analog. Science 270, 1464-1472 20 Olsen GJ, Woese CR, Overbeck R (!994) The winds of (evolutionary) changes: Breathing ne,:+llife into microbiology. J Bacteriol 176, 1--6 , ! ~leischmann RD, Adams MD, White O, et al (1995) The genome of Haemophilus b~uenzae Rd. Science 269, 496-512 22 Baron C, B6ck A (l~+! ~The lengh of the aminoacyl-acceptor stem of selenocysteine-specific~NA ofEsd+erichia coli is the determinant for binding to elongation factors St:'~Bor EF-'i'u. J Biol Chem 266, 20375-20379 23 Tormay P, B0ck A (!'~97) Barriers for the heterologous expression of a selenoprotein gene in bacteria. J Bacterioi 179, 576-582
~
Structural model for the selenocysteine-specific elongation factor SelB R Hilgenfeld a, A B6ck b, R Wilting b* alnstitut fiir Molekulare Biotechnologie eV, Beutenbergstra~e 11, D-07708 Jena; bLehrstuhl fiir Mikrobiologie der Universitiit Miinchen, Maria-Ward-Strafe la, D-80638 Miinchen, Germany (Received 10 November 1996; accepted 29 November 1996)
Summary m A structural model was established for the N-terminal part of translation factor SelB which shares sequence similarity with EF-Tu, taking into account the coordinates of the EF-Tu 3D structure and the consensus of SelB sequences from four bacteria. The model showed that SelB is homologous in its N-terminal domains over all three domains of EF-Tu. The guanine nucleotide binding site and the residues involved in GTP hydrolysis are similar to those of EF-Tu, but with some subtle differences possibly responsible for the higher affinity of SelB for GTP compared to GDP. In accordance, the EF-Tu epitopes interacting with EF-Ts are lacking in SelB. Information on the formation of the selenocysteyl-binding pocket is presented. A phylogenetic comparison of the SelE domains homologous to EF-Tu with those from EF-Tu and initiation factor 2 indicated that SelB tbrms a separate class of translation factors.
selenocysteine / elongation factor / protein synthesis / tRNA / prokaryotes Introduction The product of the selB gene from Escherichia coli is required for the cotranslational insertion of selenocysteine into protein tfor a review see [1]). Both genetical and biochemical lines of evidence indicate that SelB has the function of a specialized translation factor homologous to that of elongation factor Tu (EF-Tu). In vitro SelB forms a quaternary complex with selenocysteyl-tRNA sec, GTP and a secondary structure of the selenoprotein mRNA immediately downstream of the UGA codon directing selenocysteine insertion 112-4]. This preformation of the quaternary complex which requires stoichiometric amounts of the individual components is essential for the decoding process in vivo [5]. A recent analysis of the sequences of four bacterial SelB species has shown that the N-terminal part of the molecule is homologous to the three structural domains of EF-Tu [6]. The C-terminal part, on the other hand, displays no sequence similarity to any other protein known. It functions as an RNA binding domain attaching the translation factor to the recognition sequence of the mRNA. SelB in comparison to EF-Tu displays a number of intriguing characteristics. It binds GTP with on approximately 10-fold higher affinity than GDP and, therefore, does not
require any guanine nucleotide release factor to displace GDP [7]. Moreover, SelB only recognizes tRNASec when it is charged with selenocysteine and it does not interact with any significant affinity with the 20 ,~tandard aminoacyltRNAs [8]. To gain somt: insight into the structural basis of these differences in the absence of a three-dimensional structure for SelB we have established a three-dimensional model on the basis of the coordinates of the EF-Tu structure.
Materials and methods Modeling of SelB The high-resolution crystal structure of the Er:-Tu from Thermus thermophilus in complex with a GTP-analogue [91 served as a template for constructing the model for SelB. The available SeIB sequences were aligned with the amino acid sequences of EF-Tu from E coli and T thermophilus. The multiple alignment was revised when suggested by inspection of the three-dimensional structures (fig 2). Visualization and model building was carried out using Frodo [ 10]. The polypeptide geometry around deletions and insertions was initially regularized by the method of Hermans and McQueen [! 1] as built into Frodo. Backbone atoms belonging to topologically equivalent residues as well as identical side chains and common atoms in non-identical side chains were transferred from the EF-Tu structure. The remaining atoms were positioned by calculation of the potential energy as a function of side chain torsion angles [12]. Finally, the entire model was subjected to energy minimization.
Genetic distance calculation *Correspondence and reprints
The Genworks 2.4 program (IntelliGenetics, Inc) was used for genetic distance calculation for the phylogenetic tree (fig 4).
972 Results and discussion
A structural model for Se/B On the basis of the recent elucidation of the three-dimensional structure of EF-Tu in its active GTP-form [9, 13] and its inactive GDP-form [14] and of the significant sequence homology between EF-Tu and SelB [6], a three-dimensional model for the three N-terminal domains of E coli SelB was established. The model was constructed using methods suitable for deriving a structure when that of a homologous protein is known [121. Protease treatment of purified E coli SelB confirms the structural model: trypsin was found to cut between Arg-34 and Gly-35 at the end of helix A I" and between Arg-152 and Gly-153 in the FI helix [151. These positions are situated in solvent exposed regions in the model structure. Secondary structure Domain 1 of E coli SelB has largely the same overall fold as the nucleotide-binding domain (domain 1) of EF-Tu. Both proteins share 35% identical amino acid residues in this domain. However, in SelB major deletions of polypeptide stretches are found (fig 1). The amino-terminal deletion of 11 residues most probably does not influence the structural integrity of ~he domain. On the other hand, in EF-Tu this stretch makes important interactions with domain 2 and ties the two domains together; consequently, this will not occur in SelB. A deletion of 14 residues in SelB comprises not only parts of the A l-helix, but also the following extended chain and the short A I' helix, both of which are believed to be a part of the so-called 'effector region' in EF-Tu (residues {42 ] to {62 ] (residues of SelB are numbered according to the sequence of E co/i SelB and corresponding positions of T thermophilus EF-Tu are given in brackets)) [ 161. In contrast, the A" helix, located at the rim of the tRNA binding cleft, is completely conserved. The following 13-strands b l and c i, and the B I helix which undergoes an important change in orientation upon GTP hydrolysis [91, are also well conserved. The same is true for the completely solvent-inaccessible ~-strand d l, the C I strand and the DI helices and I~-strands e I and fl. However, beyond the fl strand, as many as 20 residues (from { 1751 to { 1941) are absent in SelB. This means that nearly the entire 10-residue El helix, the loop-like structures following it and the first residue of the F I helix are missing. It is worth noting that in EF-Tu, this portion is intimately interacting with the C-terminal half of the A I helix, which is also absent in SelB. All the structural elements present in EF-Tu but absent in SelB are located on one side of the molecule (fig 2A, B). The secondary structure of domain 2 (25% sequence identity between E coli SelB and T thermophilus EF-Tu) is structurally much better conserved than that of the nucleotide-binding domain. His- 1273 } and Gin- 1278 }of EF-Tu,
located in a turn that interacts with the N-terminal residues of the chain, are missing in SelB. As in domain 2, all secondary elements known from EF-Tu are conserved in domain 3 of SelB (22% sequence identity), and deletions and insertiens are only found in loops at the surface of the molecule. A major deletion of eight residues concerns a loop (residues ! 323 } to {330} in EF-Tu) that interacts, via water molecules, with domain 1 in EF-Tu. Another eight-residue deletion (residue {358 } to {3651) is located in a surface loop very ~,Iose to the previous one. Since the deletions in domain 3 are located at the far end of the molecule compared to the c|~,anges in domain I, it is possible that they mark the sites oi' interactions with the C-terminal domain 4 of SelB which cot,dd not be modeled in this study due to the lack of a template. Nuc!eotide binding The binding mode of GTP is similar in EF-Tu and SelB. The phosphate b i n d i n g Io~3p, c o n t a i n i n g the canonical GXXXXGKT (G l-box), i~',well conserved, and so are the residues in the first (Thr- 141251, Thr-37 {621) and second (Asp-26 {511, Asp-57 {81 }) ceordination sphere of the Mg2+ ion (fig 3A, B). The latter is paxrtof the DXXG motif (G3box), the glycine of which (Giy-60{ 84}) plays a pivotal role in the GTPase switch mechanism ,I!71. As in EF-Tu, the guanine base is specifically recognized by the conserved Asp- ! ! 51 ! 39 }, ~!ia hydrogen bonding with the N i and N2 atoms (f;g 3). h~ most GTPases, this residue is part of the NKXD motif (G4-box). in EF-Tu Asn{ 136 } makes important interactions \with other nucleotide-binding segments of the polypeptide c'hain, such as the main-chain carbonyi of residue {22} and ~he main-chain amide of residue I 1751. in SelB this asparag{ne is replaced by Thr-112, the side-chain hydroxyl of which can still donate a hydrogen bond to the main-chain carbonyl of residue {221. The other interaction is missing, because the polypeptide chain takes a different course at the residue corresponding to Ala-{ 1751, Glu-150 in SelB. As a result, the E coil SelB lacks the hydrophobic interaction between the guanine base and Leu-{ 176} of the 'SAL'-motif found in EF-Tu. This is due to the complete deletion of the EI helix in SelB~ which forces the polypeptide chain to enter directly into the FI helix at this location. Therefore, GTP and GDP binding to SelB can be expected to be weaker than that to T thermophilus EF-Tu, and this was also observed experimentally [71. In spite of that, the side-chain of Thr-149 can still donate a hydrogen bond to the 06 atom of the guanine base. Interaction with EF-Ts The crystal structure of E coli EF-Tu in complex with EF-Ts revealed that the contact sites in EF-Tu are located in domains 1 and 3 [181. Only a few of these residues in domain 1 are present in E coli SelB, whereas the contacts in domain 3 (Phe-{335 }, Met-{361 }, and Met-{3631), are absent. The
973
SELB_EC
EFTU_TT
SELB_EC
EFTU_TT
SELB_EC
flREGEFURTE
~
34
PHUHVG
PHVEVEDYGD
LGYRY TAHVE
gPQPDI~PG F ' I ~ K F
LS~LIF~GG
VETeK-vS
94 119
I l eQ
EFTU_TT
~ILLRRQV~ ~/IU~flH~
SELB_EC
. . . . . . EGRG flDA~FIEHLLQ L..-~EREHASQ LEQMHRHPKT RRGEHEWVOE IMEW.DAIDE YI~JTPVRDVD
KP~_MPVEDU
TGTALSGEVE TGRIERGKVE
TRHRGQR iRL EGIAGDHVGV
EFTU_TT
SELB_EC
EFTU_TT
-
SELB_EC
EFTU_TT
SELB_EC
Tu_TT
68
. - - . ~ ~
VGDSLULTGV VGDEVEIVGL
~II~PELL~
H~GDEV
H-EPflRVRRL HRQH-QPTERPETRRTVVT GVEMHRETLQ
PVIRGSRLLA
149 179
239
H iflGDREKEQ LLRG-VSREE
249 298
~ii~--~
c
I HRGDWLLAD VPPE~F~VI VELQT. . . . . . . -~PLTQW UERGQVLaKP GSI T~JH~FE ASVYVLKI(EE GGR~GFFS6
~
L-AELVFDT~ LI,I~DHD~V
L~IS~L
UTFTUELIKW U, EGL . leEG-
.
AGARJ~JilLHP PRRGKRKPEY 352 GAGV~i( ILE
486
SELB_EC
LQWLASLARA QSDADALSVH LERGAVNLAD FAWARQLNGE GMRELLQOPG YIQAGYSLLH
412
SELB_EC
APVAARWQRK
ILDTLATYHE
QHRDEPGPGR
ERLRRMALPM
EDEALVLLLI EKMRESGDIH
472
SELB_EC
SHHGWLHLPD
HKAGFSEEQQ
AIWQKAEPLF
6DEPWWVRDL
AKETGTDEQA
MRLTLRQAAQ
532
SELB_EC
QGIITAIVKD
RYYRNDRIVE
FANMIRDLDQ
ECGSTCAADF
RDRLGVGRKL
AIQILEYFDR
592
SELB_EC
IGFTRRRGHD
HLLRDALLFP
EK
614
Fig 1. Alignment of SelB from E coli (EC) and EF-Tu from Thermus thermophilus (TT) based on a multiple alignment of four SelB sequences and two EF,~2, sequences [6]. Identical residues were colored red for domain l, yellow for domain 2, blue for domain 3 and orange for domain 4. Rouo4ed bars indicate o~-helices, arrows indicate D-sheets of EF-Tu.
974
~i~!~i~¸ !i ¸¸~i¸i¸~i¸i ¸I!' !i~iii i ~ i¸¸!~!~ ~i ~ ......
EF-Tu
E. coli
SELB
Fig 2. Structural comparison of EF-Tu and the homologous part of SelB. A. Three-dimensional structure of T thermophilus EF-Tu [9]. B. The three-dimensional model of the EF-Tu-homologous part of E coil SelB. Only the polypeptide backbone is shown. Colors: red, domain I; yellow, domain 2; blue, domain 3; green, GTP and magnesium ion.
assumption that SeiB is independent of EF-Ts is further supported by biochemical data: SelB is the only translation factor that has a higher affinity for GTP (Kd = 1.7 mM, similar to E coli EF-Tu) than for GDP (Kd = 10 mM, 1000-fold higher compared to E coli EF-Tu), therefore the nucleotide exchange will occur spontaneously [7]. Binding of aminoacyl-tRNA
EF-Tu binds all aminoacylated tRNAs except formylmethionyl-tRNAMcti which is bound by initiation factor 2, and selenocysteyl-tRNAS~, which is bound exclusively by SeiB, Information on the selenocysteyl-tRNA binding site can be obtained from a comparison with the crystal structure of the ternary complex of Phe-tRNA phc, EF-Tu and GppNHp [ 19]. The binding site for the CCA-Phe end on EF-Tu is formed by the cleft between domains 1 and 2, as has been predicted earlier [9]. The same binding site is also found in the structure of the complex between EFTu,GppNHp and anthranyi adenosine-5'-monophcsphate, a compound regarded as a model for the 3'-aminoacylated adenine of the tRNA (Hilgenfeld et al, manuscript in preparation). In both EF-Tu complexes, the pocket for the amino acid (phenylalanine and anthranylic acid, respectively) is lined by the side chains of Phe- {229 }, Thr- {239 }, Glu1226}, Asp-{2271, His-167} and His-{273} (table I). In SelB, the two former of these residues are conserved (Phe183 and Thr- 193, respectively), while Glu {226 } is replaced by Asp- 180 and His-167 } by Tyr-42. Interestingly, Asp1227 } becomes Arg- 181 in all four SelB sequences, which might protrude into the binding pocket and restrict its size. In addition, its guanidinium group might interact with the negatively charged selenocysteyl side chain (pK ~6.5). Both factors
might contribute to the specificity of SelB for selenocysteine, which can be tested by site-directed mu,tagenesis. In EF-Tu, His-| 2731 is located at the tip of a loop that protrudes into the cleft between domains 1 and 2, thereby closing the amino acid binding site at its bottom (orientation as in fig 2A, B). In the ternary complex between PhetRNA pheand EF-Tu,GppNHp, main chain amide and carbonyl groups from this loop interact with the aminoacyl ester of the tRNA. In SelB, however, the loop is shorter and has a different conformation. As a result, there is no residue in E coil SelB that would correspond to His- {273 } of EF-Tu, so that the amino acid binding pocket is much more open in SelB as compared to EF-Tu. This effect is enhanced by the deletion in SelB of both the eleven N-terminal residues and much of the 'effector region', which in EF-Tu form the entrance of the tRNA-binding cleft. It is possible that these deletions allow portions of domain 4 of SelB to approach the amino acid binding site and are involved in determining the specificity for selenocysteyl-tRNAsec. The only base of the tRNA found to interact with the EF-Tu in the crystal structure of the ternary complex [ 19] is the 3'-terminal adenosine monophosphate. In this complex and in the complex with anthranyl-5'-adenosine monophosphate (Hilgenfeld et al, in preparation) it interacts with the hydrophobic side chains of Val- {237 }, lie- {231 }, and, in part, Leu-{289} on one side, and with Glu-{271 } on the other. In SelB the valine is conserved (Val-191), whereas the isoleucine is exchanged for Val-185, the leucine replaced by Asn-240, and the glutamate is replaced by Gin224. Therefore, with the exception of the exchange of Leu to Asn, SelB provides an environment for the adenine base that is similar to that provided by EF-Tu. Similarly, residues seen to interact with the 5'end of the tRNA in the ternary
975 Table I. Residues involved in aminoacyl-tRNA binding of T thermophilus EF-Tu and corresponding residues of the four SELB sequences. tRNA binding residues a. Adenine 76-interactions EF-Tu
SELB
lie [231 ] Val [2371 Glu [271 ] Leu [289]
Val/Met [ 1851 Val [1911 Gin [2241 Asn [2401
Evolutionary considerations
b. Backbone interactions
Lys [52] Glu [55] His [851 Asp [871 Tyr [881 Lys [90] Gin 1911 Arg [30011 Arg [330] His [331 ] Gin [341] Thr [3501 mcGly [391 ]
Arg/His27 Glu30 His61 Lys/Arg63 Phe64 Ser/Lys/Arg66 Gln/Asn67 Arg252 -/Arg27 ! h His/-272 His285 Th~'394 lle/Glyfl'yr326
Aminoacyl-binding pocket
His 1671 Glu 1226] Asp [227] Phe [229] Thr [2391 mcHis [2731 mcArg [2741 mcAsn 12851
EF-Tu complex [19] are largely conserved in SelB (see lable I). In conclusion, SelB will bind the Sec-tRNAS~ in a way very similar to the way EF-Tu binds aminoacyl-tRNA, b7 vivo analysis supports this conclusion, because with C-terminally truncated SelB-derivatives it was demonstrated that the tRNA binding site is located in the first 378 amino acids of SelB, which corresponds to the EF-Tu homologous part [5].
Tyr/Phe42 Asp180 Arg 18 ! Phe 183 Thr 193 -/His225a GIn/-/Gly226 Arg236
Table II. Amino acid identity between the four domains of SELB from E coli (EC), C thermoaceticum (CT), D microbium (DB), and H influenza (HI). Domain SELB
1--4 (%)
1-3 (%)
4 t%)
CT-EB CT-DB DB-EC CT-HI DB-HI EC-HI EC-HI-DE .CT
31 40 30 32 27 42 15
39 50 38 42 36 49 23
22 26 21 20 16 34 6
The conservation of both primary and three-dimensional structure of the N-terminal 3-domains in comparison to that flom EF-Tu allows the determination of the phylogenetic relationship between SelB and other translation factors. A phylogenetic tree (fig 4) of the amino acid sequences of EF-Tu and the homologous parts of the SelB from Clostridium thermoaceticum, Desulfomicrobium baculatum [6] and Haemophilus influenzae I21] and initiation factor 2 (IF2) was calculated using the Geneworks 2.4 program. The parts of IF-2 homologous to EF-Tu comprise residues 386749 of E coli IF-2 and residues 237-587 of Bacillus stearothermophilus (BS) IF-2. To compare the genetic distances among the SelBs to those of related proteins, two EF-Tu and IF-2 sequences from phylogenetically well separated bacteria were included into the calculations. Taking the branching point error (fig 4: numbers in brackets) into account, the SelB and EF-Tu sequences are slightly closer related to each other than either of them is to IF-2. When the distance values of the SelB sequences are compared it is obvious that SelB is not derived from EF-Tu of the corresponding organism, but that the SelBs form a distinct class of translation factors. The genetic distances between the SeiB sequences are in accord with the phylogeny of the organisms as delineated from the rRNA sequence [20]. This is true for domains 1 to 4, 1 to 3, and 4 alone. Although the C-terminal half of SelB might be younger than domains I to 3, this means that it also has evolved before the phyiogenetic separation into the bacterial subgroups. Whereas domains l to 3 of the SelB sequences share 23% sequence identity, domains 4 only contain 6% identical amino acids (table II). In our model, the domains 1 to 3, as in EF-Tu, have to interact with the ribosome, guanine nucleotides and the tRNA, therefore a higher structural constraint is necessary than for the mRNA binding domain 4. The flexibility in sequence of the mRNA-binding domain is obvious in case of E coli and H influenzae, because they share only 34% identity in domain 4 (table II), although the mRNA stem-loops offdnG of E coli and the homolog of H influenzae (deduced from ORF H influenzae 00006 [21]) differ in only three out of 45 bases.
Conclusions
The model of E coli SelB showed that its N-terminal part is homologous to EF-Tu. The guanine nucleotide binding site
976
A
| HN "'
ILYS241
I
Imc(~ly84~NH. , •
N.-H3 o.'~)'-.
imcLys24[
.HN
i ."" . ..°
. H O ~
'..o-' '-~,I " i I ..
• ""
.o. ~o. r-~
:
.
.,.,
,Thr62,Od'"
,
.,
,,
t
.::lm©Sly231 i
."
"
,~. |
."
i!"
/:. ~o.
~
".tm~Th'=S]
:
I
,
.~
.,
.H3N
:
""" ~
~;.
?
~ N . .
ImcLysl 3
:
' IAsnl=Sl
.....
.'"'
i
n .olmcAspl0l. O •"
• ""
IThr371.OH
"
".
: N,H
,, - -
--
"
;:lm©Sly12i
"
i'
" ,
.
: , i
,,o.
~
':
NH I
"~n~
."
"
o
."
~
~
NH .....
| ~=,~ t~J~~.
;..
'
,k
I ~..,,,~t
:
'
imcThrtSl
. . ~ .... o~
~
~
O .... HO~"S'I
."
. ,
.o,
. ~ |
'
;
""
[
o
O ' m m " P m O ' - " P " " ' O I I '. |
u
....
HN
"'"'.~'"'~'""'.~'"'"': IThrST[NH
;'HN~mcAIa17si
i
"
"."3
. o ~
"'"
......~%,o °
I-°.,.=~1
ILys13i I
_. .....
::
Im~A*P='I
B lmcGlY121 I HN.
.... o~ ~o4,.,,-~
"
"
"m¢i:r26'
j,., ,-
" +r~.P
',1
H3N
'
,-.
....
i
_g O
, ,
O ......
HO
: '
"
." O'" , i
.-"HO ,,
[T'C~ou~ ~N4mcThr14[ Im=A'pl01
?"
o.,o..
ImoHislll Fig 3. Detailed comparison of the GTP-binding amino acids of T thermophilus EF-Tu (A) and E coil SelB (B). Deviations in SelB are shaded.
is similar to that of EF-Tu, as are the amino acids involved in the GTPase mechanism. SelB contains no EF-Ts binding site; a nucleotide exchange factor is not required since SelB has a lower affinity for GDP than for GTP. The mode of tRNA binding can be expected to be similar in EF-Tu and
SelB, although the aminoacyl-pocket of SelB is more open than that of EF-Tu and contains an arginine residue that might be involved in selenocysteyl-recognition. Deacyiation kinetics of selenocysteyl-tRNA in complex with SelB and of seryl-tRNAser bound to EF-Tu indicated the same
977
0.272
N'-SELB_CT
0.272
N'-SELB_DB
o.101 (eo.020) 0.198
(+_0 +22)
0.266
N'-SELB_EC
0.266
N'-SELB_HI
0.161
EF-TU_EC
0.107 0.059
(+0.049) 0.409
"~(+0.015) O.161
m
(+0.061) 0.422
~
'(+_0.017)
EF-TU_TT
0.2fH
N'-IF-2_EC
0.207
N'-IF-2_BS
Fig 4. Phylogenetic tree of the EF-Tu-hemologous parts of SelB, EF-Tu, and IF-2 sequences. CT, Clostridium thermoaceticum, DB, Desulfomicrobium baculatum; EC, Escherichia coil; HI, Haemophilus influenzae; "IT, Thermus thermophilus; BS, Bacillus stearothermophilus.
degree of ester bond protection [22]. Therefore, it is unlikely that the aminoacyl-binding pocket in SelB is more accessible to the solvent. Instead, it can be assumed that another part of SelB, probably domain 4, 'closes' the pocket. Domain 4, which has no counterpart in EF-Tu, is much less conserved among the four SelB molecules than the other three domains. This divergence may be related to the fact that this domain interacts with the selenoprotein mRNA [6] and thereby reflects the co-evolution of the different mRNA recognition sequences with the cognate translation factor. This is supported by our recent finding that SelB from D baculatum is unable to functionally substitute for the E coil SelB in selenoprotein synthesis in E coil [23]. On the basis of the structural model presented, targeted mutational analysis of residues involved in recognition of selenocysteyl-tRNA sec can be performed for its proof.
Acknowledgments We thank H Schtitz for help with computer calculations. This work was supported by grants from the Deutsche Forschungsgemein-
schaft (SFB 369), from the DESY-PS, 7rom European Commission via its Human Capital and Mobility Programme, and from the Fonds der Chemischen lndustrie.
References ,~ Baron C, B6ck A (1995) The selenocysteine-inserting tRNA species: Structure and function. In: tRNA: Stnwture, Biosynthesis and Ftowtion (S011 D, RhajBhandary U, eds) American Society for Microbiology, ASM Press, Washington DC, 529-544 2 Heider J, Baron C, B~ck A (1992) Coding from a distance: dissection of the mRNA determinants required for the incorporation of selenocysteine into protein. EMBO J i 1, 3759-3766 3 Baron C, Heider J, B0ck A (1993) Interaction of translation factor SelB with the formate dehydrogenase selenopolypeptide mRNA. Proc Natl Acad Sci USA 90, 4181- 4185 4 Ringquist S, Schneider P, Gibson T, Baron C, BOck A, Gold L 11994) Recognition of the mRNA selenocysteine insertion sequences !~y the specialized translation factor SelB. Genes Dev 8, 376-385 5 Tormay P, Sawers A, B6ck A (1996) Role of the stoichiometry between mRNA, translation factor SelB and selenocysteyl-tRNA in seleaoprotein synthesis. Mol Microbiol 2 i, 1253- i 259 6 Kromayer M, Wilting R, Tormay P, B6ck A (1996) Domain structure of the prokaryotic selenocysteine-specific elongation factor SelB. J Mol Biol 262, 413-420
978 7 Forchhammer K, Leinfelder W, B0ck A (1989) Identificationof a novel translation factor necessary for incorporation of selenocysteine into protein. Nature (Lond) 342, 453-456 8 F0rster C, Ott G, Forchhammer K, Sprinzl M (1990) Interaction of selenocysteine-incorporating tRNA with elongation factor Tu from E coli. Nucleic Acids Res 18, 487-49 l 9 Berchtold Hi Reshetnikova L, Reiser CO, Schirmer NK, Sprinzl M, Hilgenfeld R (1~3)Crystal structure of active elongation factor Tu reveals major domain rearrangements. Nature 365, 126-132 l0 Jones TA (1985) Interactive computer graphics: FRODO. Methods Enzymol 115, 157-171 I l Hermans J, McQueen JE (1974) Computer manipulation of (macro) molecules with the method of local exchange. Acta Crystallogr Sect A 30, 730 12 Summers NL, Karplus M (1989) Construction of side-chains in homology modelling. Application to the c-terminal lobe of rhizopuspepsin. J Mol Bio1210, 785--811. 13 Kjeldgaard M, Nissen P, Thirup S, Nyborg J (1993) The crystal structure of elongation factor Tu from Thermus aquaticus in the GTP conformation. Structure l, 35-50 14 Kjeldgaard M, Nyborg J (1992) Refined structure of elongation factor EF-Tu from Escherichia coli. J Mol Bioi 223, 721-742
15 Forchhammer K, Riicknagel K-P Brick A (1990) Purification and biochemical charactedsation of SelB, a translation factor involved in selenoprotein synthesis. J Biol Chem 265, 9346-9350 16 Peter ME, Schirmer NU, Reiser COA Sprinzl M (1990) Mapping of the effectorregion in Themms thermophilus EF-Tu. Biochemist~. 29, 2876--2884 17 Hilgenfeid R (1995) How do GTPases really work? Nature Struct Biol 2, 3--6 18 Kawashima T, Berthet-Colominas C, Wulff M, Cusack S, Leberman R 996) The structure of Escherichia coli EF-Tu. EF-Ts complex at 2.5 resolution. Nature 379, 511-518 19 Nissen P, Kjeldgaard M, Thirup S, Polekhina G, Reshetnikova L, Clark BFC, Nvborg J (1995) Crystal structure of the ternary complex of PhetRNAPhr:.EF-Tu and a GTP analog. Science 270, 1464-1472 20 Olsen GJ, Woese CR, Overbeck R (!994) The winds of (evolutionary) changes: Breathing ne,:+llife into microbiology. J Bacteriol 176, 1--6 , ! ~leischmann RD, Adams MD, White O, et al (1995) The genome of Haemophilus b~uenzae Rd. Science 269, 496-512 22 Baron C, B6ck A (l~+! ~The lengh of the aminoacyl-acceptor stem of selenocysteine-specific~NA ofEsd+erichia coli is the determinant for binding to elongation factors St:'~Bor EF-'i'u. J Biol Chem 266, 20375-20379 23 Tormay P, B0ck A (!'~97) Barriers for the heterologous expression of a selenoprotein gene in bacteria. J Bacterioi 179, 576-582
~