University of Groningen

Structure-function relationships in type I signal peptidases of bacilliRoosmalen, Maarten Leonardus van

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2001

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Roosmalen, M. L. V. (2001). Structure-function relationships in type I signal peptidases of bacilli. s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 09-07-2021

https://research.rug.nl/en/publications/structurefunction-relationships-in-type-i-signal-peptidases-of-bacilli(25594878-2063-475d-9acc-a67e14d7d84f).html

61

CChhaapptteerr FFiivvee

DISTINCTION BETWEEN MAJOR ANDMINOR BACILLUS SIGNAL PEPTIDASES

BASED ON PHYLOGENETIC ANDSTRUCTURAL CRITERIA

This chapter was published in:The Journal of Biological Chemistry (2001), 276(27), 25230-25235.

555

Chapter Five

62

DISTINCTION BETWEEN MAJOR AND MINOR BACILLUSSIGNAL PEPTIDASES BASED ON PHYLOGENETIC AND

STRUCTURAL CRITERIA

Maarten L. van Roosmalen, Jan D.H. Jongbloed, Jean-Yves F. Dubois, Gerard Venema, Sierd Bron andJan Maarten van Dijl

Summary

The processing of secretory pre-proteins by signal peptidases (SPases) is essential for cell viability.As previously shown for Bacillus subtilis, only certain SPases of organisms containing multipleparalogous SPases are essential. This allows a distinction between SPases that are of major andminor importance for cell viability. Notably, the functional difference between major and minorSPases is not clearly reflected in sequence alignments. Here, we have successfully used molecularphylogeny to predict major and minor SPases. The results were verified with SPases from variousbacilli. As predicted, the latter enzymes behaved as major or minor SPases when expressed in B.subtilis. Strikingly, molecular modeling indicated that the active site geometry is not a criticalparameter for classification of major and minor Bacillus SPases. Even though the substrate bindingsite of the minor SPase SipV is smaller than that of other known SPases, SipV could be convertedinto a major SPase without changing this site. Instead, replacement of amino-terminal residues ofSipV with corresponding residues of the major SPase SipS was sufficient for conversion of SipVinto a major SPase. This suggests that differences between major and minor SPases are based onactivities other than substrate cleavage site selection.

Introduction

Signal peptidases (SPases) play a key role in thetransport of proteins across membranes in all livingorganisms. The type I SPases are integralmembrane proteins that remove signal peptidesfrom pre-proteins during, or shortly aftertranslocation across the cytoplasmic membrane,thereby releasing the mature proteins from thetrans side of the membrane (for reviews, see Refs.(159) and (89)).

In recent years, type I SPases from many differentorganisms have been identified. Comparison ofthese SPases showed that they can be divided intwo sub-families: P(prokaryotic)-type andER(endoplasmic reticulum)-type SPases (106).The P-type SPases are found in eubacteria andorganelles of eukaryotes. In contrast, the ER-typeSPases are typical for the endoplasmic reticular

membrane. Strikingly, a few ER-type SPases wereshown to be present in sporulating Gram-positiveeubacteria, such as Bacillus subtilis (106). In fact,B. subtilis was the first eubacterium in which thepresence of both P- and ER-type SPases wasdemonstrated. With respect to SPases, B. subtilis isnot only exceptional because it contains both P-and ER-type SPases, but also because it is theorganism with the largest known number of type ISPases. These include the chromosomally encodedP-type SPases SipS, SipT, SipU and SipV, theplasmid-encoded P-type SPases SipP1015 andSipP1040, and the chromosomally-encoded ER-type SPase SipW (103-108). These observationssuggested that at least some of the SPases of B.subtilis have specialized functions. Indeed, it wasrecently shown, that SipS, SipT and SipP1015 areof major importance for the secretion ofdegradative enzymes and cell viability, whereasSipU, SipV and SipW have only a minor role in

Major and minor Bacillus signal peptidases

63

Table 5.I. Plasmids and Bacterial StrainsPlasmids Relevant properties Reference

pGEFdSH-K83A encodes sf-SipS-His K83A of B. subtilis; 3.6 kb; Apr (121)pGDL41 Encodes pre(A13i)-β-lactamase and SipS of B. subtilis; replicates in E. coli and B. subtilis.

8.1 kb; Apr; Kmr(103)

pGDL46.36 pGDL48 derivative carrying the sipC (Bca) gene; Apr; Kmr (118)pGDL48 Lacks the sipS gene and contains a multiple cloning site; otherwise identical to pGDL41; 7.5

kb; Apr; Kmr(107)

pGDL90 pGDL48 derivative carrying the sip (Bli) gene; Apr; Kmr This paperpGDL100 pGDL48 derivative carrying the sipT (Bsu) gene; Apr; Kmr (105)pGDL121 pGDL48 derivative carrying the sipU (Bsu) gene; Apr; Kmr (105)pGDL131 pGDL48 derivative carrying the sipV (Bsu) gene; Apr; Kmr (105)pGDL140 pGDL48 derivative carrying the sipW (Bsu) gene; Apr; Kmr (110)pM0 pGDL41 derivative; the 3' end of the sipS gene is replaced with a multiple cloning site; Apr;

Kmr(113)

pM0V pM0 derivative carrying a SipSV fusion protein; Apr; Kmr This paper

Bacterial strain Genotype ReferenceE. coliBL21(DE3) F- ompT rb-mb- λDE3 (167)

B. subtilis8G5 trpC2; tyr; his; nic; ura; rib; met; ade; lacks the sipP genes (205)8G5 ∆S like 8G5; rib+; sipS (104)8G5 ∆ST pGDL90 like 8G5; sipS sipT; contains the sip (Bli) gene This paper8G5 ∆ST pM0V like 8G5; sipS sipT; contains the hybrid sipSV gene This paper

protein secretion and are probably involved inspecific non-essential processes (106;108). Forexample, SipW is specifically required for theprocessing of two precursors, pre-TasA and pre-YqxM, but not for cell viability (116;117). As thepresence of a single major SPase (i.e. SipS, SipT orSipP1015) is sufficient for growth and cell viabilityof B. subtilis, it seems that the secretory precursorprocessing machinery of this organism isfunctionally redundant (106;108).

In addition to an N-terminal membrane anchordomain (A), all P-type SPases contain four well-conserved domains (B to E) (89). These conserveddomains include residues involved in substraterecognition and catalysis. Specifically, domain Bcontains a strictly conserved Ser residue anddomain D a strictly conserved Lys residue.Together, these residues form a Ser-Lys catalyticdyad (29). The domains B to E of the P-typeSPases are conserved in the ER-type SPases.Nevertheless, instead of a Lys residue, domain Dof the ER-type SPases contains a strictly conservedHis residue, which is required for catalysis. Atpresent, it is not known whether this His residue ispart of a Ser-His catalytic dyad, or a Ser-His-Aspcatalytic triad, as described for the classical serineproteases (110;115).

The distinction between P-type and ER-typeSPases can readily be made on the basis of theconserved Lys or His residues in domain D. Incontrast, it is presently not clear which propertiesdetermine whether a type I SPase is a major or aminor SPase of B. subtilis. A clear definition ofthese properties is important to understand themolecular basis for SPase substrate specificity.Therefore, the present studies were aimed at thecharacterization of differences between major andminor Bacillus SPases, and the identification ofdomains in these enzymes that are critical for theirspecificity. The results show that major and minorBacillus SPases can be distinguished byphylogenetic analyses and that critical informationfor their role in cell viability is provided byresidues that are located amino-terminally of thecatalytic Ser residue. Strikingly, molecularmodeling of the active-site of major and minor P-type SPases of B. subtilis suggests that the activesite cleft of the minor SPase SipV is significantlysmaller than that of the other known Bacillus P-type SPases. Nevertheless, this difference can notexplain why SipV is a minor SPase.

Chapter Five

64

Materials and Methods

Plasmids, Bacterial Strains and MediaTable 5.1 lists the plasmids and bacterial strains used. TYmedium (tryptone/yeast extract) contained Bacto tryptone(1%), Bacto yeast extract (0.5%) and NaCl (1%). If required,media for Escherichia coli were supplemented withampicillin (Ap; 50 µg/ml) or kanamycin (Km; 20 µg/ml);media for Bacillus subtilis were supplemented with Km (20µg/ml) or chloramphenicol (Cm; 5µg/ml).

Evolutionary tree computationsAmino acid sequences of Bacillus SPases were collectedfrom the SubtiList and GenBank databases. Alignments wereperformed with the ClustalX software (218), using the"Gonnet 250" and the "Gonnet series" matrices as thepairwise alignment parameters and multiple alignmentparameters, respectively. Default gap opening and gapextension parameters were applied. When the SPase of E. coli(Lep) was included in the alignments, using the sameparameters, the aligned sequences showed highly congruentareas that correspond to α-helices, β-strands, and previouslydefined conserved domains (D) of Lep and other type ISPases (89) (29). Therefore, the complete data set (CS),predicted α-helices, β-strands and conserved domains wereused in the phylogenetic analyses. Autapomorphic insertionsor deletions (Indels) were removed from all data sets. Treereconstructions were performed according to two differentmethods. Firstly, the Maximum Likelihood (ML) method wasused as implemented in the PUZZLE 4.02 software (219).The VT (variable time) matrix (http://www.dkfz-heidelberg.de/tbi/people/tmueller/paper/VT-matrix/) wasapplied, with 4 gamma rates. One thousand replications wereused to calculate the "Quartet Puzzling" values (QP).Secondly, the Maximum Parsimony (MP) method was usedas implemented in the program PAUP 4.03b(http://www.lms.si.edu/PAUP/). The MP tree reconstructionwas done with the “Branch Swapping: Tree-Bisection-Reconnection” (TBR) algorithm, applying 10 randomadditions of sequences. One thousand replications were usedto calculate the bootstrap values (BP).

DNA techniquesProcedures for DNA purification, restriction, ligation,agarose gel electrophoresis, and transformation of E. coliwere carried out as described in Ref. (161). Enzymes werefrom Roche Molecular Biochemicals. The polymerase chainreaction (PCR) was carried out with Vent DNA polymerase(New England Biolabs) as described in Ref. (113). DNA andprotein sequences were analyzed with the PCGene AnalysisProgram (version 6.7; Intelligenetics Inc.) and ClustalWversion 1.74 (162). B. subtilis was transformed as described inRef. (106). Correct integration of plasmids or resistancemarkers into the chromosome of B. subtilis was verified bySouthern blotting, or PCR.Plasmid pGDL90, specifying Sip (Bli) from Bacilluslicheniformis, was constructed by ligating an EcoRI- andSalI-cleaved PCR-amplified fragment of sip (Bli), into the

corresponding sites of pGDL48. The sip (Bli)-specificfragment was amplified with primers Lbl1 (5'-ACGCGTCGAC TATGC TGTGA CAGAC TG-3') and Lbl2 (5'-CGGAA TTCGC AGTGC TGGCA TCA-3'), using B.licheniformis chromosomal DNA as a template. PlasmidpM0V, specifying a hybrid of SipS and SipV from B. subtilis,was constructed by ligating an EcoRI- and SalI-cleaved PCR-amplified fragment of sipV, into the corresponding sites ofpM0. The sipV-specific fragment was amplified with primersJBV1 (5'-TGTCG TCGAC GGTGA CAGTA TGAACCCGAC CTTCC-3') and JBV2 (5'-CGGAA TTCGC TAGCGACGCC TCTTC AATTA GCA-3'), using B. subtilischromosomal DNA as a template. Note that primer JBV1 isdesigned such that the resulting SipSV fusion proteincontains the amino-terminal fragment of SipS (residues 1-43)that includes the active site Ser43 residue, fused to thecarboxy-terminal fragment starting at the correspondingposition in SipV (residues 34-168).

Western Blot AnalysisPolyclonal antibodies against SipS of B. subtilis wereprepared by immunization of rabbits (Eurogentec) with apurified soluble form of this protein (sf-SipS-His K83A).This protein consists of the catalytic domain of SipS, lackingresidues 2-29. Furthermore, sf-SipS-His K83A (Bsu) containsa carboxyl-terminal hexa-histidine tag, facilitating thepurification by metal-affinity chromatography (121). Westernblotting was performed as described in Ref. (184). Afterseparation by SDS-PAGE, proteins were transferred toImmobilon-PVDF membranes (Millipore Corporation). Todetect SPases, B. subtilis or E. coli cells were separated fromthe growth medium by centrifugation (5 min, 10.000 x g,room temperature), and samples for SDS-PAGE wereprepared as previously described (121;164). SPases werevisualized with specific antibodies and horseradishperoxidase-anti-rabbit-IgG conjugates (AmershamInternational).

Molecular modeling and molecular dynamics simulationsThree-dimensional models of Bacillus SPases were built onthe basis of homology with the E. coli SPase (PDB ProteinData Bank: 1b12) using the molecular modeling programWHAT-IF (http://www.cmbi.kun.nl/whatif/) (220). Themolecular dynamics program GroMacs(http://md.chem.rug.nl/software.html) was used to perform astandard energy minimization in vacuo of a pentapeptidesubstrate in the three-dimensional model of SipS.

Results

Phylogenetic clustering of major Bacillus signalpeptidasesTo investigate the relationships between major andminor SPases of Bacillus species, phylogeneticanalyses were performed, applying the MaximumLikelihood (ML) and Maximum Parsimony (MP)

Major and minor Bacillus signal peptidases

65

Table 5.II. Summary of Quartet Puzzling and Bootstrap values.Bootstrap values were obtained with the maximum likelihood method (1000 replications each), and quartet puzzling valueswith the maximum parsimony method. Values lower than 50 are not considered significant. Values higher than 80 are shown inbold. CS, complete sequences; α, α-helices; β, β-strands; D, conserved domains.

Maximum Likelihood Maximum ParsimonyCS αααα ββββ D CS αααα ββββ D

Number of variable (ML) /informative (MP) sites 236 34 82 42 159 23 59 32

SipW (1) 91 71 75 93 100 75 99 99SipV (2) 84 84 98 72 99 56 98 98SipT Bsu + SipT Bam (3) 78 80 75 80 86SipS Bsu + SipS Bam (4) 90 89 75 98 87 72SipP (5) 91 60 53 61(3) + (4) + (5) + Sip Bli 93 79 60 79 77(3) + (4) + (5) + Sip Bli + SipU 90 96 85 80(1) + SipU 62(1) + SipX Ban 70 100 99(2) + SipC Bca 66 96 54 67(3) + SipX Ban 81

methods. For this purpose, either the completesequences (CS), conserved α-helices (α), β-strands(β) or domains (D) were used (Table 5.II).Consistent with the fact that only few α-helices arepresent in type I SPases (29), the ML analysis thatwas based on α-helices resulted in a tree with apoorly resolved topology, and the equivalent MPanalysis in 108 “most parsimonious trees” (112steps long; CI1 excluding uninformative characters= 0.8077, RI1 = 0.7101, RC1 = 0.5833). Far betterresults were obtained when complete sequences, β-strands, or conserved domains were used in the MLand MP analyses (Table 5.II). In fact, the MPanalysis with complete sequences resulted in one"most parsimonious tree" (805 steps, CI excludinguninformative characters = 0.8188, RI = 0.7152,RC = 0.5988; Fig. 5.1). One "most parsimonioustree" was obtained when conserved β-strands wereused (283 steps, CI excluding uninformativecharacters = 0.8618, RI = 0.8046, RC = 0.7079);and three “most parsimonious trees” were obtainedwith the conserved domains (134 steps, CIexcluding uninformative characters = 0.8319, RI =0.7938, RC = 0.6753). Notably, all data sets arecongruent with respect to the clustering of the fourbest supported groups of Bacillus SPases: I), theSipW group; II), the SipV group; III), the SipT ofB. subtilis (Bsu) + SipT of Bacillusamyloliquefaciens (Bam) group; and IV), the SipSgroup (Table 5.II). As shown in Fig. 5.1, SipT ofBacillus anthracis (Ban) seems to be more closelyrelated to the SipW group (bootstrap percentages[BP]/quartet puzzling values [QP] are 100/nc for

Fig. 5.1. Most parsimonious tree of known Bacillus SPases.A most parsimonious tree of different Bacillus SPases is shown.The calculated Maximum Parsimony (top) and MaximumLikelihood (bottom) values are shown at junctions. The lengthof the branches reflects the number of amino acid changesbetween different SPases, as indicated by the bar. See Materialsand Methods for details concerning the calculations. The clusterof major SPases is encircled. The following SPases are shown:SipS, SipT, SipU, SipV, SipW, SipP1015 and SipP1040 fromB. subtilis (Bsu); SipS, SipT, SipV and SipW from B.amyloliquefaciens (Bam); SipC from B. caldolyticus (Bca); Sipfrom B. licheniformis (Bli); and SipX and SipW from B.anthracis (Ban). Note that SipX was originally annotated asSipT (NCBI accession #AAF13664) but, to avoid possiblemisinterpretations, this SPase was renamed.

CS and 100/70 for β-strands) than to the SipT(Bsu) + SipT (Bam) group. To prevent the possiblemisinterpretation that SipT (Ban) is related to themajor SPases SipT (Bsu) and SipT (Bam), the SipT(Ban) protein was renamed SipX (Ban).

Chapter Five

66

Fig. 5.2. Functional identification of major SPases. Westernblotting analysis of B. subtilis 8G5 (BS), B. subtilis ∆S (BS ∆S),B. subtilis ∆ST (BS ∆ST) and E. coli MC1061 (EC) producingSip (Bli) as specified by plasmid pGDL90, or the hybrid SPaseSipSV as specified by plasmid pM0V. As a control, strainscontaining the empty vector pGDL48 (ev) were used. Thepresence of SipS (Bsu), SipT (Bsu) and Sip (Bli) was visualizedwith polyclonal antibodies against the catalytic domain of SipS.Note that the hybrid SPase SipSV does not cross-react withthese antibodies.

Furthermore, SipC of Bacillus caldolyticus (Bca)seems to be most closely related to SipV (Bsu) andSipV (Bam). Most importantly, the functionallydefined major SPases SipS, SipT and SipP1015 ofB. subtilis cluster together (encircled in Fig. 5.1).This clustering is supported by bootstrappercentages of 79/93 and 77/60 when completesequences or β-strands were used for the analyses,respectively. This observation suggests that theother SPases in this cluster, SipP1040, SipS (Bam),SipT (Bam) and Sip (Bli), should also be classifiedas major SPases. In contrast, all enzymes notincluded in this cluster would be minor SPases.

Functional identification of major BacillusSPasesAs the distinction between major and minor SPasesis based on functional differences, we tested theoutcome of the phylogenetic analysis incomplementation experiments with tworepresentative SPases: SipC of Bacilluscaldolyticus (118), which clusters with the minorSPases (Fig. 5.1), and Sip (Bli) of B. licheniformis(214), which clusters with the major SPases. Tothis purpose, the sipC gene was expressed in the B.subtilis strain ∆S, which lacks the sipS gene, bytransformation with the pGDL48-derived plasmidpGDL46.36. Subsequently, we tried to disrupt thesipT gene of the resulting strain with a Cmresistance marker by transformation withchromosomal DNA of B. subtilis ∆T-Cm. Eventhough this experiment was repeated several times,no Cm resistant transformants were obtained,

Fig. 5.3. Limited cross-reactivity of SipS-specific antibodies.Western blotting analysis of B. subtilis ∆S strains containingplasmids for the overproduction of type I SPases of B.subtilis. Plasmid pGDL41 was used for the overproduction ofSipS, pGDL100 for SipT, pGDL121 for SipU, pGDL131 forSipV, and pGDL140 for SipW.

indicating that SipC of B. caldolyticus behaves as aminor SPase in B. subtilis that cannot replace SipSand SipT. A completely different result wasobtained in parallel experiments with Sip (Bli). Inorder to test whether this SPase could replace SipSand SipT of B. subtilis, the sip (Bli) gene wasamplified by PCR and cloned. Next, B. subtilis ∆Sas transformed with the pGDL48-derived plasmidpGDL90, which carries the sip (Bli) gene, and thesipT gene of the resulting strain was disrupted witha Cm resistance marker by transformation withchromosomal DNA of B. subtilis ∆T-Cm. ViableCm resistant transformants were obtained, whichwere shown to have disrupted sipS and sipT genes(data not shown). As shown by Western blotting,these transformants produce Sip (Bli), which cross-reacts with the antibodies raised against thecatalytic domain of SipS of B. subtilis (Fig. 5.2).Taken together, these observations strongly suggestthat Sip (Bli) behaves as a major SPase and SipC(Bca) as a minor SPase in B. subtilis. The view thatall SPases clustering with SipS (Bsu), SipT (Bsu),SipP1015 and Sip (Bli) behave as major SPaseswas finally confirmed by similar complementationexperiments, demonstrating that SipS (Bam), SipT(Bam) and SipP1040 can replace SipS and SipT ofB. subtilis (J.D.H. Jongbloed and H. Tjalsma,unpublished observations).

Major and minor Bacillus signal peptidases

67

Fig. 5.4. Alignment of the substrate binding regions of known Bacillus SPases. The conserved domains B, C, and D from P-typeSPases of B. subtilis, B. amyloliquefaciens, B. licheniformis, B. caldolyticus, B. anthracis and E. coli, which contain residues thatform the S1 and S3 substrate binding regions, were aligned. Residues, predicted to belong to the S1 or S3 pockets are labeled with 1or 3, respectively. Residue numbers below the alignment are derived from SipS (Bsu).

Because the antibodies raised against SipS of B.subtilis cross-reacted with Sip (Bli) (these studies),and the major SPase SipP1015 (108), weinvestigated whether these antibodies could beused to discriminate between major and minorSPases of B. subtilis. To this purpose, Westernblotting experiments were performed with strainscontaining plasmids for the overproduction of therespective SPases. Only SipT was shown to cross-react with the antibodies raised against SipS, whichimplies that the major SPases SipS, SipT andSipP1015 share at least one antigenic determinantwhich is absent from the minor SPases SipU, SipVand SipW (Fig. 5.3). This idea is supported by theobservation that the major SPases SipS (Bam),SipT (Bam) and SipP1040 cross-reacted with theantibodies against SipS (Bsu) (data not shown). Ithas to be noted, however, that the antibodiesagainst SipS also cross-reacted with the catalyticdomain of SipC (Bca) upon overproduction in E.coli (data not shown), indicating that theseantibodies do not allow the discrimination betweenmajor and minor Bacillus SPases in general.

SPase active site modeling by homologyTo investigate whether the active site geometries ofthe known major and minor P-type SPases of B.subtilis are significantly different, three-dimensional models of these SPases wereconstructed on the basis of the crystal structure ofthe E. coli SPase as determined by Paetzel et al.(29). For this purpose, the sequences of theseSPases were aligned with the ClustalW Program(Fig. 5.4). SipS (Bsu) and the E. coli SPase showan overall sequence identity of 26%, which is low

for modeling by homology. However, the fourconserved domains B to E of these SPases show62% sequence identity. Notably, the active site ofthe E. coli SPase is almost entirely composed ofthese four conserved domains that are typical forall P-type SPases (29). In what follows, we have,therefore, based our conclusions exclusively onmodeled active site regions of Bacillus SPases. Thehomology modeling program WhatIf was used togenerate the three-dimensional models of variousknown SPases of bacilli. As shown for SipS of B.subtilis (Fig. 5.5), Met44 and Leu48 (marked inblue), Val39 and Val82 (marked in green) andLys83 form the S1 substrate binding pocket, whileTyr37, Val54, Val73 and His80 (marked in yellow)together with the residues marked in green formthe S3 pocket. These findings are in goodagreement with the structure of the S1 and S3substrate binding pockets of the SPase of E. coli(132). Furthermore, the idea that the latter residuesmake contact with the substrate (i.e. the SPaserecognition sequence in a precursor protein) wassupported by a molecular dynamics analysis inwhich a pentapeptide of five Ala residues in a β-strand conformation was modeled into the substratebinding pockets of SipS (Fig. 5.5). Thispentapeptide was placed at the position whichcorresponds to that of the PENEM inhibitor in thecrystal structure of the E. coli SPase I (29).

The comparison of our models for the P-typeSPases of B. subtilis showed that the substratebinding pockets of SipS, SipT, SipP1015 andSipP1040 were highly similar, whereas that ofSipV was significantly smaller (Fig. 5.5).

Chapter Five

68

Fig. 5.5. Active site models for SipS and SipV. (The illustration on the back of the cover is a color representation of thisfigure). Stereo pictures of residues in the modeled active site regions of SipS (top) and SipV (bottom). Residues rendered in spheresare likely to be involved in the formation of the S1 pocket (Met44 and Leu48 of SipS; blue), S1 and S3 pockets (Val39 and Val82of SipS; green) or S3 pocket (Tyr37, Val54, Val73 and His80 of SipS; yellow) for substrate binding. The SipS model contains apentapeptide of Ala residues, docked in the substrate binding pocket. The P1 and P3 residues of this model substrate are shown inorange. The active site Ser43 and Lys83 residues, as well as Tyr81 of SipS and the corresponding Leu residue of SipV, are shownas “ball and stick” models. The latter residues are probably involved in substrate stabilization by interaction with P2 residues of thesubstrate.

Conversely, the substrate binding site of SipUappeared to have a wider S1 pocket than theequivalent sites of the other P-type SPases of B.subtilis. Upon close examination, the volume of thesubstrate binding pocket of SipV is relatively smallbecause the side chains of Leu73, Ile82 andpossibly Leu54 (SipS numbering) protrude into theS3 pocket (Fig. 5.5). The latter side chains arelarger than those of the equivalent residues in SipSof B. subtilis (Val73, Val82 and Val54,respectively) and other SPases (Fig. 5.4). Takentogether, these observations indicate that the activesite geometries of the minor SPases SipU and SipVof B. subtilis are different from the active sitegeometries of the known major SPases.

A SipS-SipV fusion protein is a major SPaseTo investigate whether the active site geometry is acritical determinant for major and minor BacillusSPases, a SipS-SipV hybrid protein (denoted

SipSV) was constructed, which is specified byplasmid pM0V. Notably, this fusion between SipSand SipV of B. subtilis was made at the catalyticSer residue of these SPases. Consequently, SipSVconsists of the first 43 residues of SipS, and thecarboxyl-terminal part of SipV. The majoradvantage of this approach is that the active sitegeometry of SipSV is nearly identical to that ofSipV (data not shown). Next, we tested whetherSipSV is a major or minor SPase by introducingpM0V into B. subtilis ∆ST, as described above forthe sipC (Bca) and sip (Bli) genes. Strikingly,viable ∆ST transformants containing pM0V wereobtained, showing that SipSV can replace SipS andSipT. As shown in Fig. 5.2, SipSV is notrecognized by the antibodies raised against SipS. Inconclusion, these observations show that SipSV isa major SPase, and that SipV is not a minor SPasedue to the geometry of its catalytic site but, rather,that some residues of its amino-terminal stretch

Major and minor Bacillus signal peptidases

69

determine SipV to belong to the class of minorSPases. Furthermore, the antibodies against SipSdo not distinguish between major and minorSPases.

Discussion

On the basis of their importance for cell viability,we have previously classified the type I SPases ofB. subtilis as major (SipS, SipT, SipP) and minorSPases (SipU, SipV, SipW) (106;108). Thus far, itwas not clear which properties of these SPases areimportant for this functional distinction, inparticular with respect to the P-type SPases.Consequently, simple amino acid sequencealignments could not be used to predict the groupto which certain Bacillus P-type SPases wouldbelong. In the present studies, we show for the firsttime that major and minor SPases can bedistinguished via phylogenetic analyses.Surprisingly, the subsequent molecular analysesdemonstrate that the distinction between major andminor SPases does not specifically relate to thecatalytic domain of a Bacillus P-type SPase, butrather to its amino-terminal domain which containsthe membrane anchor. The latter result wasunexpected, because it was recently shown byCarlos et al. (221) that the transmembrane domainsof P-type SPases are not important determinantsfor cleavage fidelity in vitro.

The most important outcome of the phylogeneticanalyses of the Bacillus type I SPases is that themajor SPases form a distinct cluster, which is well-supported by the maximum parsimony andmaximum likelihood methods. Moreover, thesephylogenetic analyses have predictive value, asexemplified by the complementation experimentswith SipC (Bca) and Sip (Bli), showing that thesebehave as minor and major SPases, respectively,when the corresponding genes are expressed in B.subtilis. This, however, does not exclude thepossibility that SipC is a major SPase in B.caldolyticus. Furthermore, the phylogeneticanalyses indicate the existence of two clusters of“minor” SPases: the SipC/SipV and the SipWclusters. The latter cluster was identifiedpreviously as it contains the known ER-type

SPases of bacilli (106;110). Only two BacillusSPases, SipU (Bsu) and SipX (Ban), do not belongto the three clusters of major SPases, SipC/SipV orSipW. This suggests that these two SPasesrepresent possible evolutionary intermediatesbetween different clusters, which is particularlyinteresting in the case of SipX of B. anthracis, asthis P-type SPase might represent a link betweenthe P- and ER-type Bacillus SPases.

The present observation that a SipSV hybridprotein, containing the largest part of the catalyticdomain of the minor SPase SipV, behaves as amajor SPase indicates that the catalytic domain ofthe P-type SPases is not the most importantdeterminant for the difference between major andminor SPases. This view is supported by the factthat, according to our models, the active sitegeometry of SipSV is identical to that of the minorSPase SipV. In this respect it is important to bearin mind that the S3 substrate binding pockets ofSipV and SipSV are relatively small compared tothose of other P-type SPases of B. subtilis, SipU inparticular. Nevertheless, the possibility that subtlechanges in the active site geometry of SipSV,caused by the fusion between the SipS and SipVmoieties, result in the conversion of a minor SPaseinto a major SPase can presently not be excluded.The idea that the catalytic domain is not importantfor the difference between major and minor SPaseswould explain why the antibodies raised againstthe catalytic domain of SipS (Bsu) can not be usedto distinguish between these two functionallydefined groups of SPases.

What could be the role of the amino-terminalresidues of SipS in determining its role as a majorSPase? Carlos et al. (221) have recently providedcompelling evidence that the transmembranesegments of type I SPases, such as SipS, are notimportant for substrate cleavage site selection.Furthermore, we have recently shown that themembrane anchor of SipS is not required for itsactivity (118). Together with our present results,these observations imply that the major-minordifference of SPases is not based on the recognitionof residues at the -1, -3 positions, relative to thescissile peptide bond per sé. This leaves open atleast three alternative possibilities. First, the

Chapter Five

70

amino-terminal residues might position thecatalytic site of a major SPase in such a way in themembrane that it can interact with the cleavage siteof one or more, as yet unidentified, pre-proteinsthat have to be processed for cell viability. Second,the amino-terminal residues might be required foran, as yet unidentified, essential interaction of amajor SPase with pre-protein translocases. Third,the amino-terminal residues might target therespective SPases to topologically distinct regionsof the membrane, such as the septa of dividingcells. Notably, the regions preceding the active siteSer residues of Bacillus SPases, which includetheir membrane anchor, show a relatively highdegree of sequence variation (118). To elucidatethe role of the amino-terminal region in SPasefunction, we are presently investigating the role ofthe first 42 residues of SipS by site-directedmutagenesis.

Acknowledgements

We thank Prof. Gert Vriend and Dr. Rob Veltmanfor critical discussions on the modeling of BacillusSPases, Dr. Danilo Roccatano en Dr. GiorgioColombo for the molecular dynamics simulations,and Dr. Harold Tjalsma, Prof. O. Kuipers and othermembers of the European Bacillus Secretion Groupfor stimulating discussions. M.L.v.R. wassupported by the Dutch Ministry of EconomicAffairs through ABON (Associatie BiologischeOnderzoeksscholen Nederland); J.D.H.J wassupported by a grant (805-33.605) from SLW(Stichting Levenswetenschappen); and S.B,J.D.H.J., J.Y.D., and J.M.v.D. were supported bygrants (QLK3-CT-1999-00415 and QLK3-CT-1999-00917) from the European Union.