Structural and functional insights into caseinolyticproteases reveal an unprecedented regulationprinciple of their catalytic triadEvelyn Zeilera,1, Anja Listb,1, Ferdinand Alteb, Malte Gerscha, Rudolf Wachtela, Marcin Porebac, Marcin Dragc,Michael Grollb, and Stephan A. Siebera,2

aCenter for Integrated Protein Science Munich (CIPSM), Institute of Advanced Studies, Department Chemie, Lehrstuhl für Organische Chemie II, TechnischeUniversität München, 85747 Garching, Germany; bCIPSM, Department Chemie, Lehrstuhl für Biochemie, Technische Universität München, 85747 Garching,Germany; and cDepartment of Bioorganic Chemistry, Faculty of Chemistry, Wrocław University of Technology, 50-370 Wrocław, Poland

Edited by Charles S. Craik, University of California, San Francisco, CA, and accepted by the Editorial Board May 3, 2013 (received for review November 2, 2012)

Caseinolytic proteases (ClpPs) are large oligomeric protein com-plexes that contribute to cell homeostasis as well as virulenceregulation in bacteria. Although most organisms possess a singleClpP protein, some organisms encode two or more ClpP isoforms.Here, we elucidated the crystal structures of ClpP1 and ClpP2 frompathogenic Listeria monocytogenes and observe an unprece-dented regulation principle by the catalytic triad. Whereas L.monocytogenes (Lm)ClpP2 is both structurally and functionallysimilar to previously studied tetradecameric ClpP proteins fromEscherichia coli and Staphylococcus aureus, heptameric LmClpP1features an asparagine in its catalytic triad. Mutation of this aspara-gine to aspartate increased the reactivity of the active site and led tothe assembly of a tetradecameric complex. We analyzed the hetero-oligomeric complex of LmClpP1 and LmClpP2 via coexpression andsubsequent labeling studies with natural product-derived probes. No-tably, the LmClpP1 peptidase activity is stimulated 75-fold in thecomplex providing insights into heterooligomerization as a regulatorymechanism. Collectively, our data point toward different preferencesfor substrates and inhibitors of the two ClpP enzymes and highlighttheir structural and functional characteristics.

The caseinolytic protease P (ClpP) is a highly conserved en-zyme present in bacteria and higher organisms (1–3). ClpP isresponsible for cell homeostasis and among other duties for theregulation of bacterial virulence in several pathogens includingStaphylococcus aureus and Listeria monocytogenes (4, 5). Earlystructural studies revealed the topology of the Escherichia coliClpP complex that consists of two heptameric rings building upa 300 kDa cylinder (Fig. 1A) (6). The interior of this proteolyticmachinery exhibits 14 active sites flanked by axial pores thatallow protein substrates to enter the hydrolytic chamber. ClpPgains its catalytic activity in complex with AAA+-chaperones(such as ClpC, ClpE, and ClpX in the case of L. monocytogenes).These ATP-dependent enzymes bind to the axial pores of ClpP,unfold the protein prone to degradation, and direct it into theproteolytic chamber (7–9).A close-up view of a single ClpP monomer reveals several

characteristic structural features that are conserved among thisclass of proteases. To harmonize the ClpP nomenclature for allsubsequent discussions, we use a general sequence numberingbased on the first determined crystal structure of ClpP fromE. coli [EcClpP, Protein Data Bank (PDB) ID code 1TYF] (10)(Fig. 1B). According to this nomenclature, a catalytic triad (Ser98,His123, Asp172) essential for proteolysis, a central E-helix witha Gly-rich loop region essential for interring contacts between thetwo heptamers, and a N-terminal region essential for interactionwith a AAA+-chaperone can be observed in all published X-raystructures to date (Fig. 1A, Fig. S1B) (10–18). Cocrystallization ofE. coli ClpP with an irreversible dipeptide chloromethylketoneinhibitor confirmed the reactivity of the catalytic triad residuesSer98 and His123 and illustrate a binding site for the dipeptidewithin the Gly-rich loop region that adopts an antiparallel beta-strand (19) (Fig. 2). Recently, two conformations of ClpP from

S. aureus have been reported that are thought to represent phys-iologically important states with an active and an inactive catalytictriad corresponding to an extended and a bent E-helix, respectively(Fig. S2) (11, 12). In addition, a highly conserved aspartate/argi-nine sensor (Asp170/Arg171) links oligomerization to the catalyticactivity and exhibits characteristic conformations in both states(Fig. S2) (12). In agreement with this model, ClpP heptamers lackthe interaction of the sensor residues with their counterparts onthe adjacent ring and thus have an inactive triad. In the tetrada-cameric state, the senor feedbacks the correct assembly to theactive sites, thereby ensuring controlled proteolysis.Although most organisms possess a single ClpP protein with

a conserved fold (6, 11, 13–16, 18, 20), the genomes of someorganisms encode two or more ClpP isoforms (21–24). For a cya-nobacterial system, heptameric rings of mixed composition havebeen reported that interact with different chaperones (22). Incontrast, ClpP proteins from L. monocytogenes (LmClpP1 andLmClpP2) as well as from Mycobacterium tuberculosis have beenfound to assemble into heterooligomeric complexes composed oftwo homoheptamers (25, 26). Inhibition of LmClpP2 with lactone-based inhibitors led to down-regulation of virulence without af-fecting viability (27). In contrast, both mycobacterial ClpP subunitsare essential for bacterial survival, emphasizing defined functionalroles of ClpP proteins among species (26, 28).Interestingly, LmClpP2 shares a high-sequence homology with

ClpP enzymes of various organisms that feature one ClpP (Fig. S1A and C). LmClpP1 exhibits only 41% sequence identity withLmClpP2, raising the question of how these two distinct isoformsinteract and how they differ functionally. Furthermore, there isa distinct difference between the two ClpP homologs in the com-position of their catalytic triad: Asp172 of LmClpP2 is replaced byan asparagine in LmClpP1, an unusual observation within serineproteases that is, however, conserved in several uncharacterizedhomologs (Fig. S1 A and C). Although the replacement of an as-partate with an asparagine represents only a moderate structuralalteration, it significantly influences the strength of the catalytictriad charge-relay system. The nucleophilicity of the active siteSer98 in LmClpP1 and LmClpP2 was previously monitored andcompared by β-lactone activity-based probes (25, 29). Although allmonocyclic β-lactones selectively labeled LmClpP2 either as

Author contributions: M. Groll and S.A.S. designed research; E.Z., A.L., F.A., M. Gersch,R.W., and M.P. performed research; M. Gersch, M.D., M. Groll, and S.A.S. analyzed data;and S.A.S. wrote the paper.

Conflict of interest statement: M. Gersch, M.P., M.D., and S.A.S. are named inventors ona patent application describing fluorogenic substrates suitable for ClpP activity measurements.

This article is a PNAS Direct Submission. C.S.C. is a guest editor invited by the EditorialBoard.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID codes 4JCQ, 4JCR, and 4JCT).1E.Z. and A.L. contributed equally to this work.2To whom correspondence should be addressed. E-mail: stephan.sieber@tum.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental.

11302–11307 | PNAS | July 9, 2013 | vol. 110 | no. 28 www.pnas.org/cgi/doi/10.1073/pnas.1219125110

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF1http://www.pdb.orghttp://www.rcsb.org/pdb/explore/explore.do?structureId=4JCQhttp://www.rcsb.org/pdb/explore/explore.do?structureId=4JCRhttp://www.rcsb.org/pdb/explore/explore.do?structureId=4JCTmailto:stephan.sieber@tum.dehttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1219125110

a homooligomer or as part of the heterooligomeric complex,a probe derived from the bicyclic natural product vibralactone(VLP) was able to interact with both LmClpP1 and LmClpP2catalytic sites. Importantly, binding of the ligand to LmClpP1was only observed in the presence of LmClpP2 (25).

ResultsIn our studies we performed a two-tiered strategy that first eval-uates the chemical reactivity of the enzyme active site followed bythe investigation of the corresponding molecular arrangement atthe atomic level.

Redesign of the Catalytic Triad in LmClpP1 by a N172D MutationLeads to Increased Reactivity and Tetradecameric Oligomerization.To investigate the reactivities of the two different triads, we per-formed several in-depth chemical and biochemical assays withLmClpP1 and LmClpP2. Considering the electronic layout ofa charge-relay system, it is expected that the natural Asp172Asnexchange in LmClpP1 reduces the strength of hydrogen bondswithin the active site, thus lowering the nucleophilicity of Ser98.This is in accordance with previous findings that this mutation

impairs LmClpP1 to open monocyclic β-lactones but still allowsbinding to bicyclic VLP (25). To test if the LmClpP1 triad can bereactivated, we mutated Asn172 to Asp and subsequently moni-tored the active site acylation by the following procedure (Fig. 3 Aand B): (i) Living E. coli cells recombinantly expressing the mutantLmClpP1 construct were incubated with various lactone probesin situ (Fig. S3A). (ii) Following lysis, Huisgen–Sharpless–Meldalclick-chemistry was used to attach a fluorescent rhodamine-azideto the alkyne handle (Fig. 3A) (30–32). (iii) The proteome wasseparated by SDS/PAGE and lactone modified ClpPs were visu-alized via fluorescent scanning. Whereas VLP only showed mod-erate labeling of wild-type LmClpP1, mutation of Asn172 toalanine prevented binding to VLP completely (Fig. 3B). Thisstrongly indicates that Asn172 is an essential member of the cat-alytic triad and that it is required for proper acylation reactivityduring peptide bond hydrolysis. Interestingly, the N172D mutantenzyme reacted not only with VLP but also with monocyclic probesG2 and E2 that did not bind to wild-type LmClpP1 (Fig. 3B). Thesefindings emphasize that the N172D redesign of the triad led to anincreased reactivity of LmClpP1. We thus performed assays withfluorogenic substrates to quantitatively compare the catalytic ac-tivities of the LmClpP1 wild-type and mutant enzymes. The fre-quently used substrate Suc–Leu–Tyr–7-amino-4-methylcoumarin isonly hydrolyzed by LmClpP2 but not by LmClpP1 (Fig. 3C). Wetherefore used a tripeptide substrate with leucine at the P1 site(Leu–7-amino-4-carbamoylmethylcoumarin; SI Methods), which iscleaved by both LmClpP enzymes (for structural details on thefluorogenic substrate Leu–ACC, see Fig. S5B). Remarkably, us-ing this substrate, LmClpP1–N172D showed a ∼20-fold increase inpeptidase activity compared with wild-type LmClpP1 (Fig. 3D).It was previously suggested that the activity of ClpP is coupled to

the oligomeric state with heptamers as inactive and tetradecamersas active species (12, 25). Recombinantly purified wild-typeLmClpP1 exists predominantly as a heptamer. In contrast, theLmClpP1–N172Dmutant formed a tetradecamer as shown by sizeexclusion chromatography (SEC) (Fig. S3B). These findings in-dicate that the active site mutation affects both the reactivity aswell as the oligomerization of the protease.Next, we elucidated the β-lactone binding preferences of

LmClpP1, LmClpP2, and LmClpP1–N172D with a collection ofprobes via in situ labeling experiments. Although wild-typeLmClpP2 accommodates aromatic rings as well as medium to longaliphatic chains (e.g., E2, G2, U1P, 120P, and D3, respectively),LmClpP1–N172D favors small P1 sites as present in G2 and E2(Fig. 3E), indicating significant differences in the substrate bindingchannel. Toquantitatively assess the amount of lactone binding, weapplied intact proteinmass spectrometry (MS) and determined theinfluence of all lactones on peptidase activity (Fig. S4 A–C). ForLmClpP1–N172D, solelyVLP and, to aminor extent,G2were ableto reduce peptidase activity by 50% and 12%, respectively. Cor-respondingly, VLP was identified by MS to react covalently withLmClpP1–N172D. In case of LmClpP2, lactone-induced inhibitionof peptidase activity corresponded well with the proportion of

Fig. 1. Main structural elements of ClpP. (A) Top and side view of the tet-radecameric ClpP complex from E. coli (10) (EcClpP, PDB ID code 1TYF, sur-face representation) with one subunit highlighted in dark gray. Each subunit(close-up, ribbon diagram) is made up of seven α-helices (denoted with let-ters) and 11 β-strands (denoted with numbers) and contains a catalytic triad(highlighted with red circles). Relevant secondary structures (α-helices E andF, β-strand 9) are highlighted in gold. (B) Sequence alignment of EcClpP withLmClpP1 and LmClpP2. The secondary structure elements are depicted forEcClpP. The catalytic triad is framed in red, the residues forming the E-helixare underlined in orange, the conserved proline and the glycins in the Gly-rich loop are colored blue, and the Asp/Arg sensor is shown in green.

Fig. 2. Stereo-representation of ClpP monomers. Structural superpositionof LmClpP1 (gold), LmClpP2 (green), SaClpP (PDB ID code 3V5E, pale red),and EcClpP (PDB ID code 2FZS, gray) with covalently bound CMK inhibitor.

Zeiler et al. PNAS | July 9, 2013 | vol. 110 | no. 28 | 11303

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF4

covalently modified active sites as well as with results of gel labeling(Fig. 3E, Fig. S4B). Additionally, we examined the kinetics for thereaction of LmClpP2 with the representative lactones G2 and U1Paccording to our previously established protocol (33) (Fig. S5). Theresults revealed a one-step reaction for G2 and a reversible pre-complex of enzymeandU1P before irreversible covalent attachment.

Heterooligomeric Complex Formation Stimulates Hydrolytic Activityof LmClpP1. Labeling of coexpressed and copurified LmClpP1 andLmClpP2 (in which only one of the two proteins containsa Strep-tag) with VLP yielded bands for both isoforms suggestingthe presence of a heterooligomeric complex (Fig. 4A) (25).However, this complex represents only a minor species in vitro asshown by analytical SEC experiments and native PAGE (Fig.S6). We therefore investigated the activity of several LmClpP1and LmClpP2 active site mutants by in situ proteome labeling.As expected, the LmClpP1–S98A and N172A mutants were notmodified by VLP due to disruption of the catalytic triad (Fig. 4B).The corresponding mutations in LmClpP2 (S98A and D172A)resulted in inactive enzymes as well (Fig. 4C). Interestingly, wild-type LmClpP1 retained acylation activity in complexes with bothLmClpP2 mutants, demonstrating that LmClpP2 serves as an

effector for LmClpP1. These observations are in agreement withrecent findings on MtClpP2 (26) and emphasize the mutuallystimulating influence of both LmClpP isoforms.To elucidate this aspect in more detail we analyzed a putative

activation of LmClpP1 in the presence of LmClpP2 by peptido-lytic assays. To ensure that the activity exclusively originates fromLmClpP1, we used a copurified mixture of inactive LmClpP2–S98A and LmClpP1. Quantification through MS and coomassiegel yielded a ratio of 3:1–9:1 of LmClpP2–S98A to LmClpP1varying with different preparations (Fig. 4D). Interestingly, thecomplex of LmClpP2–S98A and LmClpP1 showed a 75-fold in-crease in peptidase activity compared with pure LmClpP1 (Fig.4E). Notably, LmClpP1 was even ninefold more active than theLmClpP1–N172D mutant in the heterooligomeric complex (Fig.3D). Because Suc–Leu–Tyr–AMC is not cleaved in the assay, sideactivity by E. coli ClpP can be ruled out (Fig. 4E). As expected forLmClpP1-mediated activity, only VLP but not U1P reduced thepeptidase turnover by 50% (Fig. 4F). In agreement with β-lactonelabeling (Fig. 4G), these results point toward ClpP1 regulationdependent on the heterooligomer formation and, furthermore,display substrate preferences between LmClpP1 and LmClpP2.

Crystal Structures of LmClpP1 and LmClpP2. To gain insight into themolecular interactions that account for the observed differences inchemical reactivity, we determined the crystal structures ofLmClpP1 and LmClpP2. Both enzymes were purified, crystallized,and the structures were determined by molecular replacement(Table S1). LmClpP2was elucidated at 2.6Å resolution and consistsof two heptameric rings that form a tetradecameric barrel (Fig. 5A).

Fig. 3. Activity based protein profiling experiments with β-lactone probesand peptidase activity of different LmClpP constructs. (A) Schematic illus-tration of a labeling experiment with a ClpP subunit containing the Ser–His–Asp catalytic triad. For R1 and R2, please refer to Fig. S3A. (B) FluorescentSDS/PAGE analysis of LmClpP1 separately as wild-type and with mutatedN172 residue, labeled with VLP and the monocyclic β-lactones (VLP, G2, E2,U1P) in situ. (C) Peptidase activity of LmClpP proteins (3 μM) measured with200 μM Suc–Leu–Tyr–AMC at 32 °C. (D) Peptidase activity of LmClpP proteins(3 μM) recorded with 100 μM of a custom-synthesized, fluorogenic Leu–ACC(for structural details on the fluorogenic substrate Leu–ACC, see Fig. S5A)substrate at 32 °C. Note, LmClpP1–N172D is 20-fold more active than wild-type LmClpP1. (E) Fluorescent SDS/PAGE analysis of ClpP1 and ClpP1–N172Din comparison with ClpP2, labeled in situ with VLP and monocyclic β-lactones(VLP, U1P, E2, G2, D3, 120P, A1, M1, N1, P1, Q1).

Fig. 4. ClpP1P2 heteroconstructs illustrating distinct substrate specificities.(A) Labeling profile of the isolated tetradecameric complex of LmClpP1(strep)/LmClpP2 with VLP. (B and C) Fluorescent SDS/PAGE analysis ofrecombinantely coexpressed ClpP1P2 either with ClpP2 wild-type (wt) anda ClpP1 mutant (B) or ClpP2 wild-type and a mutated ClpP1 (C). The labelingwas performed with VLP and the monocyclic β-lactone U1P. (D) CopurifiedStrep-tagged LmClpP2–S98A and tag-free LmClpP1 (ratio 5:1) identified byintact protein MS. (E) Heterooligomerization of LmClpP1 stimulates itspeptidolytic activity 75-fold. Rate constants of the heterooligomeric complexwere calculated with respect to the concentration of the LmClpP1 subunits.(F) Peptidase activities of the heterooligomeric complex after incubationwith 50 μM of inhibitor at room temperature. (G) Binding of isolated LmClpP1and in the presence of LmClpP2–S98A with β-lactones (VLP, G2, U1P).

11304 | www.pnas.org/cgi/doi/10.1073/pnas.1219125110 Zeiler et al.

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF5www.pnas.org/cgi/doi/10.1073/pnas.1219125110

These two rings are interconnected by E helices, resulting in a cy-lindrical shape with a height of 86 Å. The helix is similar to pre-viously reported structures in the active and extended form (10, 12,17, 18); the complex height, however, resembles the inactive andbent form (11, 13, 15, 34). The residues forming the catalytic triad(Ser98, His123, and Asp172) of LmClpP2 are misaligned in all 14subunits (Fig. S7A). This is further supported by the tilted positionof the arginine sensor (Arg171) that engages in an intramolecularcontact with Asp168 (Fig. S7B). In analogy to inactive ClpP fromS. aureus (SaClpP; Fig. S2), Arg171 hence lacks binding to Asp170across the heptamer interface, which likely induces the mis-alignment of the proximal catalytic triad (Fig. S7) (12).The observation of an extended E-helix accompanied by an

inactive triad is not present in many structures. A structuraloverlay between single subunits of extended SaClpP (active state),chloromethylketone-bound E. coli ClpP (CMK–EcClpP), andLmClpP2 reveals significant differences in the orientation andlength of the Gly-rich region that precedes the central E-helix(Fig. 2). In LmClpP2 the Gly-rich segment forms a long, un-structured loop accompanied by a partially unfolded and thus sig-nificantly shorter E-helix. In SaClpP and CMK–EcClpP, however,the Gly-rich region adopts beta strand conformation leading toa well-defined substrate binding channel (Fig. 2) (10, 19). Of note,the beta strand mediates ring–ring contacts in the active structure ofSaClpP and thus stabilizes tetradecamer formation (Fig. 2, Fig. S7).The structure of LmClpP1 was elucidated at 2.0 Å resolution

(Fig. 5B). Surprisingly, the overall structural fold of LmClpP1matches by 88% and a root-mean-square deviation of Cα-atoms of0.9 Å despite the low sequence homology with LmClpP2 (Fig.S7B). This finding suggests that both enzymes evolved underphylogenetic pressure to conserve their overall architecture that is

likely a prerequisite for recognition and heterooligomer formation.Again, the Arg171 sensor motif is tilted and the catalytic triad in-active (Fig. S7B). Interestingly, structural alignments show an evenshorter E-helix of LmClpP1 compared with LmClpP2 as well as anextended loop that is similar to the one observed for LmClpP2(Fig. 2, Fig. S7A). Within this loop alternative conformations arevisible in the electron density maps, indicating structural flexibilityin this region. Furthermore, the sequence comparison betweenLmClpP1 and LmClpP2 shows no identity in its primary structureand length within the N-terminal amino acids (Fig. 1B). LmClpP1thus exhibits a larger axial pore comparedwith LmClpP2 and otherbacterial ClpPs (Fig. 5), which is in line with results of previouselectron microscopy studies (25).

Crystal Structure of the LmClpP1–N172D Mutant. To understand themechanistic basis for the reactivation of LmClpP1, we crystal-lized the LmClpP1–N172D mutant and determined its crystalstructure to 2.0 Å resolution (Table S1). The enzyme assemblesinto a tetradecameric complex (Fig. 5C), however with severalstriking differences compared with the wild-type (wt) LmClpP1structure, (i) the catalytic triad is aligned as in the active SaClpPor EcClpP structures (Fig. S7A), (ii) the E-helix of LmClpP1–N172D is extended by four amino acids, and (iii) the loop regionis four amino acids shorter as in the wild type, which is due toa transition of loop residues I136–A140 into the correspondingE-helix. As in the wild-type structure, no electron density wasobtained for the Gly-rich loop.The conformation of the catalytic triad—that is, the position

of Asn/Asp172—is directly linked to an Asp–Arg sensor aspreviously described for SaClpP (12). The interactions of thesensor residues Asp170 and Arg171 with their counterparts onthe opposite ring (Fig. 6 A and B) as illustrated in LmClpP1–N172D enable the interaction of the catalytic residues His123and Asp172 in contrast to LmClpP1 wild type (Fig. 6B). Hereby,several hydrogen bonds within the equatorial plane between thetwo heptameric rings are formed, stabilizing the tetradecamericcomplex. Interestingly, a conserved Pro125 undergoes majorstructural rearrangements that correspond to an aligned anda misaligned triad. The distinct states of Pro125 keep its transconfiguration. This tilted position is indicative for an active triad,while rotation of the proline by 58°(backbone rearrangementC-α, 1.3 Å; N, 1.9 Å) as observed in wild-type LmClpP1 depictsan inactive state. Furthermore, this proline is at a central posi-tion that links the alignment of the active site via the flexibleglycine-rich loop with the E-helix length.

Structural Expansion of Wild-Type and Mutant LmClpP1. Anotherstriking difference between wild-type and mutant LmClpP1 is thelateral dimension of both complexes with 7% (6 Å) expansion ofthe mutant (Fig. S8). The basis for the increased length is theprolonged E-helix in the mutant enzyme, which displaces bothheptamers apart from each other, preventing a clash with helix F(residues Tyr162–Thr169; Fig. S8). Interestingly, the lateral di-mension of LmClpP2 is just in between these two states, con-tributing to the view that ClpP proteases are able to sample manydifferent conformations (35).

DiscussionWhereas ClpPs from bacteria that exhibit only one isoform havebeen studied in great detail, the function and regulation of systemswith more than one ClpP isoform are still poorly understood.Pioneering work in mycobacteria revealed a heterooligomeric as-sembly of ClpP1 and ClpP2 in which the interaction of both hep-tameric rings stimulated activity and exhibited specialized substratepreferences (26). Here we elucidate the high-resolution X-raystructures of LmClpP1, LmClpP1–N172D, as well as LmClpP2 andevaluate their activity by customized probes as well as substrateturnover assays. These chemical and structural experiments pro-vide unprecedented insight into the regulation principle of thecatalytic triad, which influences proteolytic activity, oligomerization,

Fig. 5. Structure of LmClpP tetradecamers. Top and side view of the tet-radecameric complex LmClpP2 (A, green), LmClpP1 (B, gold), and LmClpP1–N172D (C, blue). The E-helix is highlighted with a dark color.

Zeiler et al. PNAS | July 9, 2013 | vol. 110 | no. 28 | 11305

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF8http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF8

helix length, and complex size via several coordinated conforma-tional changes.Isolated LmClpP1 is almost inactive in peptidolytic and acyl-

ation assays. However, an active site redesign by mutagenesisinduced both oligomerization and significant activation in pep-tide hydrolysis as well as lactone binding. In addition, LmClpP1is significantly stimulated in a heterooligomeric complex withLmClpP2. These two observations of our biochemical studiesidentify an important regulatory role of the catalytic triad. Wethus determined the key players of a complex structural networkby comparing the wild-type and mutant LmClpP1 structures thatrevealed pronounced differences in the arrangement of thecatalytic triad residue Asn172. The orientation of this amino acidinfluences two adjacent structural core features, the Asp170/Arg171 sensor, important for oligomerization, and the Pro125switch. This switch links the N-terminal end of the E-helix withHis123 of the catalytic triad and thus helix length with enzymeactivity (Fig. 7). In the N172D mutant, this region is in a relaxedmode with the central Pro125 residue in a tilted position (58°)(Fig. 6A). This structural rearrangement provides the helix moreflexibility to expand. If this Pro125 rotates back, as observed inthe wt structure, the loop residues pull at the helix end, leading toits partial unfolding. In turn, the position of the proline is directlycoupled to the orientation of the catalytic triad His123 thatadopts in two conformational states as well as reflecting the ac-tivity status of the enzyme. Therefore, the mutant structurecorresponds to an active ClpP species with an aligned catalytictriad. By contrast, the LmClpP1 wild-type enzyme, which con-tains an Asn172, exhibits a weaker H-bond network with His123,likely causing a conformational change toward an unfolded E-helix as well as an inactive triad (Fig. 7).Further properties of LmClpP1 are the lack of all conserved

residues that are important for the interaction with ATP-dependentchaperones such as ClpX as well as an enlarged entry channel thatmight allow chaperone-independent accessibility of protein sub-strates (36). This has been observed with N-terminal deletionmutants and with acyldepsipeptide-bound ClpPs that exhibit anenlarged entry pore and therefore unregulated proteolytic activity(18, 36–38). However, a ClpX-independent proteolytic machinerywould require tight regulation in vivo. Thus, the naturally occurringAsp to Asn substitution accounting for the reduced activity ofLmClpP1 in combination with the restricted ligand specificity, as

shown by our acylation and peptidolytic assays, seems to be anevolutionary invention ensuring controlled degradation processes.Additionally, the proteolytic activity of LmClpP1 is governed

by the need for heterooligomerization with LmClpP2. Accordingto our data, we propose that activation of LmClpP1 is achievedby the interaction of its relaxed E-helix with the matching E-helixof the adjacent LmClpP2 heptameric ring across the tetrade-camer interface. This interaction extends both helices, pushesthe Pro125 switch, aligns the Arg/Asp network, and activates theLmClpP1 triad for substrate processing.Oneprerequisite for LmClpP1/LmClpP2 heterooligomerization

is dissociation of LmClpP2 into heptamers and an appropriaterecognition by LmClpP1. Although LmClpP1 and LmClpP2 sharelow sequence identity, converged evolution of the two enzymes isa likely explanation that ensures a conserved overall structural foldand facilitates recognition of different heptamers for coordinatedcomplex formation. We modeled a heterooligomer structure (Fig.S7C), confirming that the interaction between the two differentrings provides sufficient contacts and stability, which is in line withour biochemical as well as EM studies (25).Taken together, we elucidate striking features of the two ClpP

isoforms of L. monocytogenes, revealing an unprecedentedregulation principle of LmClpP1 activity. Despite low sequencehomology, LmClpP1 and LmClpP2 have evolved a similar overall

Fig. 6. Mechanism that links activity to oligomerization. (A) Tetradecameric complex of LmClpP1–N172D (blue) structurally superimposed with wt (gold, onlytwo opposite monomers are shown). The monomers superposition of ClpP1 wt (gold) and ClpP1–N172D (blue) show the key elements involved in the catalytictriad. When the active site is aligned in the case of LmClpP1–N172D, proline 125 induces a conformational switch toward the E-helix (A140–T158), whichresults in the extension of the E-helix (I136–T158) and thus in an extended complex. Furthermore, the alignment of the triad induces the down position of theAsp/Arg (D170/R171) sensor. (B) The close-up displays the Asp/Arg sensor (D170/R171) and the interaction to the sensor of the adjacent monomer of theopposite ring (D′170/R′171), which is absent in wt ClpP1.

Fig. 7. Extension of both heptameric rings in the ClpP tetradecamer.Depicted is a model displaying all key residues within the inactive (com-pressed) and active (extended) tetradecamer. Activation proceeds through anextension of the E-helix, proline rearrangement, active site alignment, andinteraction of the Asp/Arg residues across the heptamer interface.

11306 | www.pnas.org/cgi/doi/10.1073/pnas.1219125110 Zeiler et al.

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=SF7www.pnas.org/cgi/doi/10.1073/pnas.1219125110

structural fold. However, a major difference is the nature ofLmClpP1’s unusual Asn–His–Ser catalytic triad (39). Our muta-tional and structural experiments emphasize that this triad repre-sents a key regulatory element that keeps LmClpP1 heptameric andinactive in absence of LmClpP2. In contrast, LmClpP2 is an activetetradecamer that is able to operate independently of LmClpP1.Moreover, our biochemical data show that both isoforms exhibitpreferences for specific ligands and substrates in vitro.

MethodsDiffraction datasets were collected using synchrotron radiation of λ = 1.0 Å atthe beamline X06SA, Swiss Light Source (SLS). Datasets were processed using

the program package XDS (40). Table S1 summarizes data collection andrefinement statistics. Crystal structure analysis was performed by molecularreplacement using the program PHASER (41). For further crystallographicdetails, synthesis, cloning, mutagenesis, protein expression, purification andbiochemical assays, see SI Methods.

ACKNOWLEDGMENTS. E.Z. was supported by the Sonderforschungsbereich(SFB)749 and by the Technische Universität München Graduate School (TUM-GS). M. Gersch is a member of the TUM-GS and was supported by the Fondsder chemischen Industrie. S.A.S. was supported by the Deutsche Forschungs-gemeinschaft [SFB749, SFB1035, and Forschergruppe (FOR)1406], a EuropeanResearch Council starting grant, and the Center for Integrated Protein Sci-ence. M.D. was supported by the Foundation for Polish Science.

1. Gottesman S, Roche E, Zhou Y, Sauer RT (1998) The ClpXP and ClpAP proteases de-grade proteins with carboxy-terminal peptide tails added by the SsrA-tagging system.Genes Dev 12(9):1338–1347.

2. Gottesman S, Maurizi MR, Wickner S (1997) Regulatory subunits of energy-dependentproteases. Cell 91(4):435–438.

3. Gottesman S, Maurizi MR (1992) Regulation by proteolysis: Energy-dependent pro-teases and their targets. Microbiol Rev 56(4):592–621.

4. Frees D, Qazi SN, Hill PJ, Ingmer H (2003) Alternative roles of ClpX and ClpP inStaphylococcus aureus stress tolerance and virulence. Mol Microbiol 48(6):1565–1578.

5. Frees D, Sørensen K, Ingmer H (2005) Global virulence regulation in Staphylococcusaureus: Pinpointing the roles of ClpP and ClpX in the sar/agr regulatory network.Infect Immun 73(12):8100–8108.

6. Saito S, et al. (1984) Combination of borane-dimethyl sulfide complex with catalyticsodium tetrahydroborate as a selective reducing agent for α-hydroxy esters. ChemLett 13(8):1389–1392.

7. Saito S, Bunya N, Inabe M, Moriwake T, Torii S (1985) A facile cleavage of oxirane withhydrazoic acid in DMF: A new route to chiral β-hydroxyl-α-amino acids. TetrahedronLett 26(43):5309–5312.

8. Maglica Z, Kolygo K, Weber-Ban E (2009) Optimal efficiency of ClpAP and ClpXPchaperone-proteases is achieved by architectural symmetry. Structure 17(4):508–516.

9. Reid BG, Fenton WA, Horwich AL, Weber-Ban EU (2001) ClpA mediates directionaltranslocation of substrate proteins into the ClpP protease. Proc Natl Acad Sci USA98(7):3768–3772.

10. Wang J, Hartling JA, Flanagan JM (1997) The structure of ClpP at 2.3 A resolutionsuggests a model for ATP-dependent proteolysis. Cell 91(4):447–456.

11. Geiger SR, Böttcher T, Sieber SA, Cramer P (2011) A conformational switch underliesClpP protease function. Angew Chem Int Ed Engl 50(25):5749–5752.

12. Gersch M, List A, Groll M, Sieber SA (2012) Insights into structural network responsiblefor oligomerization and activity of bacterial virulence regulator caseinolytic proteaseP (ClpP) protein. J Biol Chem 287(12):9484–9494.

13. El Bakkouri M, et al. (2010) The Clp chaperones and proteases of the human malariaparasite Plasmodium falciparum. J Mol Biol 404(3):456–477.

14. Gribun A, et al. (2005) The ClpP double ring tetradecameric protease exhibits plasticring-ring interactions, and the N termini of its subunits form flexible loops that areessential for ClpXP and ClpAP complex formation. J Biol Chem 280(16):16185–16196.

15. Ingvarsson H, et al. (2007) Insights into the inter-ring plasticity of caseinolytic pro-teases from the X-ray structure of Mycobacterium tuberculosis ClpP1. Acta CrystallogrD Biol Crystallogr 63(Pt 2):249–259.

16. Kang SG, Maurizi MR, Thompson M, Mueser T, Ahvazi B (2004) Crystallography andmutagenesis point to an essential role for the N-terminus of human mitochondrialClpP. J Struct Biol 148(3):338–352.

17. Kim DY, Kim KK (2008) The structural basis for the activation and peptide recognitionof bacterial ClpP. J Mol Biol 379(4):760–771.

18. Lee BG, et al. (2010) Structures of ClpP in complex with acyldepsipeptide antibioticsreveal its activation mechanism. Nat Struct Mol Biol 17(4):471–478.

19. Szyk A, Maurizi MR (2006) Crystal structure at 1.9A of E. coli ClpP with a peptidecovalently bound at the active site. J Struct Biol 156(1):165–174.

20. Kim SJ, Matsuoka S, Patti GJ, Schaefer J (2008) Vancomycin derivative with damagedD-Ala-D-Ala binding cleft binds to cross-linked peptidoglycan in the cell wall ofStaphylococcus aureus. Biochemistry 47(12):3822–3831.

21. Schelin J, Lindmark F, Clarke AK (2002) The clpP multigene family for the ATP-dependent Clp protease in the cyanobacterium Synechococcus. Microbiology 148(Pt 7):2255–2265.

22. Stanne TM, Pojidaeva E, Andersson FI, Clarke AK (2007) Distinctive types of ATP-dependent Clp proteases in cyanobacteria. J Biol Chem 282(19):14394–14402.

23. Viala J, Mazodier P (2002) ClpP-dependent degradation of PopR allows tightly reg-ulated expression of the clpP3 clpP4 operon in Streptomyces lividans. Mol Microbiol44(3):633–643.

24. Viala J, Rapoport G, Mazodier P (2000) The clpP multigenic family in Streptomyceslividans: Conditional expression of the clpP3 clpP4 operon is controlled by PopR,a novel transcriptional activator. Mol Microbiol 38(3):602–612.

25. Zeiler E, et al. (2011) Vibralactone as a tool to study the activity and structure of theClpP1P2 complex from Listeria monocytogenes. Angew Chem Int Ed Engl 50(46):11001–11004.

26. Akopian T, et al. (2012) The active ClpP protease from M. tuberculosis is a complexcomposed of a heptameric ClpP1 and a ClpP2 ring. EMBO J 31(6):1529–1541.

27. Böttcher T, Sieber SA (2009) Beta-lactones decrease the intracellular virulence ofListeria monocytogenes in macrophages. ChemMedChem 4(8):1260–1263.

28. Raju RM, et al. (2012) Mycobacterium tuberculosis ClpP1 and ClpP2 function togetherin protein degradation and are required for viability in vitro and during infection.PLoS Pathog 8(2):e1002511.

29. Böttcher T, Sieber SA (2008) β-lactones as specific inhibitors of ClpP attenuate theproduction of extracellular virulence factors of Staphylococcus aureus. J Am Chem Soc130(44):14400–14401.

30. Huisgen R (1984) 1,3 Dipolar Cylcoaddition Chemistry (Wiley, New York), pp 1–176.31. Rostovtsev VV, Green LG, Fokin VV, Sharpless KB (2002) A stepwise huisgen cyclo-

addition process: Copper(I)-catalyzed regioselective “ligation” of azides and terminalalkynes. Angew Chem Int Ed Engl 41(14):2596–2599.

32. Tornøe CW, Christensen C, Meldal M (2002) Peptidotriazoles on solid phase: [1,2,3]-triazoles by regiospecific copper(i)-catalyzed 1,3-dipolar cycloadditions of terminalalkynes to azides. J Org Chem 67(9):3057–3064.

33. Gersch M, et al. (2013) The mechanism of caseinolytic protease (ClpP) inhibition.Angew Chem Int Ed Engl 52(10):3009–3014.

34. Lee BG, Kim MK, Song HK (2011) Structural insights into the conformational diversityof ClpP from Bacillus subtilis. Mol Cells 32(6):589–595.

35. Kimber MS, et al. (2010) Structural and theoretical studies indicate that the cylin-drical protease ClpP samples extended and compact conformations. Structure 18(7):798–808.

36. Bewley MC, Graziano V, Griffin K, Flanagan JM (2009) Turned on for degradation:ATPase-independent degradation by ClpP. J Struct Biol 165(2):118–125.

37. Brötz-Oesterhelt H, et al. (2005) Dysregulation of bacterial proteolytic machinery bya new class of antibiotics. Nat Med 11(10):1082–1087.

38. Li DH, et al. (2010) Acyldepsipeptide antibiotics induce the formation of a structuredaxial channel in ClpP: A model for the ClpX/ClpA-bound state of ClpP. Chem Biol17(9):959–969.

39. Snijder HJ, et al. (2001) Structural investigations of the active-site mutant Asn156Alaof outer membrane phospholipase A: Function of the Asn-His interaction in thecatalytic triad. Protein Sci 10(10):1962–1969.

40. Kabsch W (1993) Automatic processing of rotation diffraction data from crystals ofinitially unknown symmetry and cell constants. J Appl Cryst 26(6):795–800.

41. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Cryst 40(Pt 4):658–674.

Zeiler et al. PNAS | July 9, 2013 | vol. 110 | no. 28 | 11307

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

15,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219125110/-/DCSupplemental/pnas.201219125SI.pdf?targetid=nameddest=STXT