Pyrococcus horikoshii TET2 Peptidase Assembling Process and Associated Functional Regulation*

Background: TET aminopeptidases are 12-subunit complexes present in the three domains of life and are involved in important biological functions. Results: The TET assembling process has been characterized. The oligomerization triggers TET activity toward large polypeptidic substrates. Conclusion: The assembling of TET is a controlled process and regulates its activity in vivo. Significance: This work provides a new example of peptidase regulation driven by self-oligomerization. Tetrahedral (TET) aminopeptidases are large polypeptide destruction machines present in prokaryotes and eukaryotes. Here, the rules governing their assembly into hollow 12-subunit tetrahedrons are addressed by using TET2 from Pyrococcus horikoshii (PhTET2) as a model. Point mutations allowed the capture of a stable, catalytically active precursor. Small angle x-ray scattering revealed that it is a dimer whose architecture in solution is identical to that determined by x-ray crystallography within the fully assembled TET particle. Small angle x-ray scattering also showed that the reconstituted PhTET2 dodecameric particle displayed the same quaternary structure and thermal stability as the wild-type complex. The PhTET2 assembly intermediates were characterized by analytical ultracentrifugation, native gel electrophoresis, and electron microscopy. They revealed that PhTET2 assembling is a highly ordered process in which hexamers represent the main intermediate. Peptide degradation assays demonstrated that oligomerization triggers the activity of the TET enzyme toward large polypeptidic substrates. Fractionation experiments in Pyrococcus and Halobacterium cells revealed that, in vivo, the dimeric precursor co-exists together with assembled TET complexes. Taken together, our observations explain the biological significance of TET oligomerization and suggest the existence of a functional regulation of the dimer-dodecamer equilibrium in vivo.


Tetrahedral (TET) aminopeptidases are large polypeptide destruction machines present in prokaryotes and eukaryotes.
Here, the rules governing their assembly into hollow 12-subunit tetrahedrons are addressed by using TET2 from Pyrococcus horikoshii (PhTET2) as a model. Point mutations allowed the capture of a stable, catalytically active precursor. Small angle x-ray scattering revealed that it is a dimer whose architecture in solution is identical to that determined by x-ray crystallography within the fully assembled TET particle. Small angle x-ray scattering also showed that the reconstituted PhTET2 dodecameric particle displayed the same quaternary structure and thermal stability as the wild-type complex. The PhTET2 assembly intermediates were characterized by analytical ultracentrifugation, native gel electrophoresis, and electron microscopy. They revealed that PhTET2 assembling is a highly ordered process in which hexamers represent the main intermediate. Peptide degradation assays demonstrated that oligomerization triggers the activity of the TET enzyme toward large polypeptidic substrates. Fractionation experiments in Pyrococcus and Halobacterium cells revealed that, in vivo, the dimeric precursor co-exists together with assembled TET complexes. Taken together, our observations explain the biological significance of TET oligomerization and suggest the existence of a functional regulation of the dimer-dodecamer equilibrium in vivo.
Efficient and controlled intracellular polypeptide breakdown is a primordial requirement that controls many cellular processes (1). Aminopeptidases constitute a group of enzymes of critical importance to intracellular regulatory networks. In addition to their key role in energy and amino acid metabolism, they contribute in a crucial manner to the protein degradation pathways by trimming the peptides released by ATP-dependent proteases such as the proteasome (2)(3)(4)(5). Moreover, aminopeptidases remove post-translationally the N-terminal amino acid from precursor proteins thus directing their maturation and cellular localization, as well as controlling their halflives (6). These enzymes equally assume important roles in specific biological processes involving peptide signaling, such as the production of the major histocompatibility complex ligands (7). Therefore, altered intracellular aminopeptidase activities have been associated with a variety of pathologies, including aging, cataracts, cystic fibrosis, angiogenesis, and cancers (5, 8 -10).
Numerous intracellular energy-independent peptidases coexist in the cytosol, but only a few of them share the capacity for self-assembly as large homo-multimeric complexes. These include bleomycin hydrolase, leucine aminopeptidase, DppA, tripeptidyl peptidase II, Tricorn protease, protease1, pab87, and the TET 4 aminopeptidase (11)(12)(13)(14)(15)(16)(17)(18). All these systems confine the peptidase activity to inner cavities, accessible exclusively to unfolded polypeptides. Although most of them adopt a barrel-shaped architecture in which the active sites are lined up alongside a single central channel, the TET aminopeptidases form unique dodecameric edifices with a typical tetrahedral shape (18). The TET particle interior is accessible via the openings situated on each facet of the tetrahedron. The internal organization of TET peptidases revealed a highly self-compartmentalized system comprising a crossing network of four access channels extended by vast catalytic chambers in which three active sites are arranged in a circular fashion (19,20). This spatial arrangement is structurally different from all other self-compartmentalized protease complexes of known three-dimensional structure.
TET cytosolic enzymes belong to the M42 or M18 metallopeptidase families in the clan MH according to the MEROPS classification system (21). The typical TET dodecahedral quaternary structure was initially described in archaea (18,22,23). It was also found in bacteria (24), and recently, the crystallographic structure of bovine and human tetrahedral aspartylaminopeptidases have revealed that the TET complexes are also present in eukaryotic cells (25,26). The quaternary structure of archaeal, bacterial, and eukaryal TET assemblies is highly conserved. Their tertiary structures show that they all exhibit a two-domain architecture consisting of a catalytic and a dimerization domain (19,24,25).
The high evolutionary conservation of TET peptidases in the three kingdoms of life suggests that they perform important biological functions. They were found to process polypeptides up to 27 amino acids in length in the absence of ATP (18,27). Three different versions of TET complexes co-exist in the hyperthermophilic archaeon Pyrococcus horikoshii as follows: PhTET1, PhTET2, and PhTET3. These complexes are built up in an identical fashion and have comparable dimensions. PhTET2 can be defined as a leucyl-aminopeptidase that displays a preference for neutral and aliphatic substrates; PhTET3 is a lysyl-aminopeptidase that hydrolyzes preferentially basic residues, and PhTET1 is a glutamyl-aminopeptidase that shows high specificity toward acidic residues (19,27,28). The comparison of the surface electrostatic potential features of the proteolytic chambers and of the structures of the active site pockets suggest a mechanism of substrate (N-terminal amino acid) discrimination based on the PhTET internal surface electrostatic potential features (19,28).
The particular substrate specificities of the three TET variants in P. horikoshii suggest that they form a complementary set of enzymes (28). Because of their cooperative action, the archaeal TET peptidases can be designated as a "peptidasome" involved in the destruction of a vast variety of polypeptides. It has been suggested that, in peptide-fermenting organisms, the TET system plays an important role in the energy metabolism (19) or in the intracellular protein degradation by hydrolyzing the peptides produced by the proteasome endopeptidase activity (23). In addition, the TET peptidases could play more specific physiological roles as they can cleave physiologically relevant peptides. This hypothesis is supported by recent work on the eukaryotic tetrahedral aspartyl aminopeptidase that has been proposed to be a key player in the central nervous system, in particular by regulating the ocular and renin system (25).
The TET enzymes are co-catalytic metallopeptidases typically binding, by means of five amino acid ligands, two atoms of zinc or cobalt per monomer. The catalytic mechanism also implies a glutamate and an aspartate residue. Cobalt ions have a clear stimulatory effect on the amidolytic activity of PhTETs, and the co-catalytic metals have been found to be important to maintain the PhTET oligomerization state (19,(27)(28)(29). Unlike other self-compartmentalized peptidases, TETs are not processive enzymes that imply the detachment of the peptide moiety from the active site once the N-terminal residue has been cleaved (19,27,28,30). The mechanism of TET hydrolysis is extremely similar to the one of secreted monomeric aminopeptidase such as Vibrio aminopeptidase Ap1; it is the charge properties and the dimensions of the catalytic pocket of each monomer that trigger the specificity of the enzyme toward the N-terminal amino acid from the peptide chain (31). Thus, in the case of TET peptidases, the biological significance for oligomerization and active site self-compartmentalization is not clear. To address this question, a site-directed mutagenesis strategy was used to slow down the natural oligomerization process of the PhTET2 complex. The structural properties of the purified PhTET2 dimer and of various oligomeric form intermediates were characterized by combining small angle x-ray scattering (SAXS), native gel electrophoresis, analytical ultracentrifugation (AUC), and electron microscopy. This allowed the dissection of the TET assembling pathway. The relationship between the aminopeptidase activity and its multimeric structure was also assessed by functional assays. Finally, density gradient fractionations and immunodetection experiments performed with Halobacterium and Pyrococcus cell extracts suggested the existence of a regulatory mechanism controlling the TET oligomerization state in vivo.

EXPERIMENTAL PROCEDURES
P. horikoshii and Halobacterium salinarum Cell Cultivation-P. horikoshii was grown on SME YP medium, pH 7 (32), supplemented with PIPES (20 mM) and elemental sulfur (1 g/liter). The cultivation was performed overnight at 90°C in 1-liter serum vials containing 500 ml of medium under anaerobic conditions (N 2 gas phase). Cells were harvested by centrifugation, and cell pellets were immediately stored at room temperature in water/isopropyl alcohol (50:50 v/v) until utilization. H. salinarum NRC1 was grown in hypersaline medium and processed as described (33). Cells were harvested at the end of mid-log phase.
PhTET2 Cloning, Expression, and Mutagenesis-Wild-type and mutant (R217S, R220S, F224S, H248S, and I292A) Phtet2 genes were generated from synthetic DNA fragments optimized for codon usage in Escherichia coli and cloned in the overexpression plasmid pET41c by GeneCust Europe. The resulting constructs were transformed in the E. coli strain BL21 (DE3) for the recombinant expression of wild-type and mutated PhTET2 proteins as described in Durá et al. (27).
Protein Purification-Total protein extracts from E. coli cells expressing the various recombinant PhTET2 proteins were prepared as described in Durá et al. (27). After heating at 85°C for 15 min to eliminate most mesophilic proteins from the E. coli host, the lysates were clarified by centrifugation at 17,400 ϫ g for 1 h, and the supernatant was loaded onto a 6-ml Resource Q column (GE Healthcare) equilibrated in 100 mM NaCl, 20 mM Tris-HCl, pH 7.5. After washing with 3 column volumes, bound proteins were eluted with a linear salt gradient (0.1-0.35 M NaCl in 20 mM Tris-HCl, pH 7.5). In the case of the PhTET2 pentamutant, two well separated elution peaks (called A and B) were observed. The fractions of each peak were combined and concentrated to 5 mg/ml using a centrifugal filter unit (Millipore) with a molecular mass cutoff of 10 kDa. For further purification and equilibrium shift assays, the proteins were loaded onto a Superose 6 size exclusion column (GE Healthcare) equilibrated in the desired buffer. To study salt (from 20 to 300 mM NaCl) and pH (7-9) effects, 50 mM HEPES and TAPS buffers were used. The purified proteins were kept at 4°C after the size exclusion step. Aliquots were analyzed on native gel electrophoresis as described below to observe the evolution of the equilibrium with time. Native gel electrophoresis in combination with analytical size exclusion chromatography experiments (see below) were systematically performed to determine optimal buffer conditions for AUC, SAXS, and activity measurements and to monitor the oligomerization state of the PhTET2 protein samples before and after experiments.
Native Gel Electrophoresis Experiments-Native-PAGE experiments were carried out to accurately identify the different PhTET2 subspecies generated upon incubation at different times and buffer conditions. Protein samples were mixed with 1 volume of loading buffer (20 mM Tris-HCl, 62.5% glycerol, pH 6.9) just before analysis. Polyacrylamide gels containing 8% acrylamide/bisacrylamide, 20 mM NaCl, 50 mM Tris-HCl, pH 8.8, were run at 4°C for 90 min in a Tris/glycine buffer (25 mM Tris-HCl, 192 mM glycine, pH 7.5). The protein bands were visualized by Coomassie Brilliant Blue staining. The ratios between the different PhTET2 species that were detected on native gel electrophoresis were found to be in excellent agreement with those determined on the same samples and in the same buffer by size exclusion experiments on a Superose 6 column.
AUC-Sedimentation velocity experiments were performed at 20°C on different samples of the purified mutated PhTET2 protein with an Optima XL-1 analytical ultracentrifuge (Beckman) at 42,000 rpm with a 50-Ti eight-hole rotor (Beckman Instruments). The analyzed samples were prepared in 50 mM HEPES, 300 mM NaCl, pH 7.5 buffer, at a protein concentration of 0.4 mg/ml as described above. Two-channel centerpieces with an optical path of 12 mm were used, and all experiments were performed using sapphire windows. Scans were recorded at 280 nm with radial spacing of 0.005 cm. The program Sednterp was used to estimate the partial specific volume from amino acid composition as well as the density and viscosity . Data were analyzed with the program Sedfit (34) using a continuous c(s) distribution model. SAXS-All experiments were performed on the beamline BM16 at the European Synchrotron Radiation Facility (Grenoble, France). Scattering curves of the samples were measured at 80°C. The scattering patterns were corrected for background scattering and the geometry of the experimental arrangement. Three samples were measured as follows: the mutated dodecamer of PhTET2 (PhTET2-12s), purified to homogeneity just before the experiment and concentrated up to 2.9 mg/ml in 50 mM HEPES, 20 mM NaCl, pH 7.5; the mutated dimer of PhTET2 (PhTET2-2s), purified to homogeneity just before the experiment and concentrated up to 3 mg/ml in 50 mM TAPS, 20 mM NaCl, pH 9; and the wild-type dodecamer of PhTET2 (PhTET2 WT), purified to homogeneity just before the experiment and concentrated up to 3.5 mg/ml in 50 mM HEPES, 20 mM NaCl, pH 7.5. The ab initio envelope was generated using the program GASBOR (35). Fifteen models were averaged and filtered with the help of the program DAMAVER (36). The resulting bead model was transformed into a volumetric map using the program pdb2vol from the Situs program package (37) and the crystallographic model was placed into this map by rigid body docking with the program Colores (38). The theoretical SAXS curves were calculated with the program CRYSOL (39) in default mode. The missing fragments were modeled (internal loop, 120 -132, and the N-terminal 1-5) in the PhTET2 12s structure and were added using PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC, and Coot (40).
Electron Microscopy-The dimeric fraction of PhTET2 was incubated in 50 mM HEPES, 150 mM NaCl, pH 7.5, after the size exclusion column step of the purification. The reoligomerization was followed by native gels and gel filtration. The intermediate fractions were isolated from the dimer and dodecamer of PhTET2 and pooled in a "low molecular weight intermediate" sample and a "high molecular weight intermediate" sample.
4 l of the PhTET2 sample (ϳ0.1 mg/ml) were loaded between the mica-carbon interface as described in Franzetti et al. (18). The sample was stained using 2% sodium silicotungstate, pH 7.5, and air-dried. Images were taken under low dose conditions in a CM12 Philips electron microscope working at 120 kV and with a nominal magnification of 40,000 using an Orius SC1000 CCD camera.
The different PDB files generated by molecular modelization were loaded into SPIDER (41) and filtered to 20-Å three-dimensional structures. Those structures were reprojected in all directions with a 10°spacing, and the reprojections were compared by cross-correlation with raw negative staining images selected from the CCD frames. For each structure, the best cross-correlating images were carefully examined by eye. The best matching and unambiguous images and the corresponding three-dimensional model orientation were selected and included in Fig. 5. Special care was taken to include only modelspecific representative views (reprojections and views that are not overlapping with other model reprojection views).
Activity on Monoacyl-pNA Peptides-PhTET2 hydrolytic activity on different synthetic chromogenic compounds was determined using aminoacyl-pNAs as described (27). Four replicates and four enzyme blanks were assayed for each experimental point. Leu-pNA was used for half-life calculations as described (27). Half-lives were calculated from a first-order exponential decay fit to the experimental points.
Activity on Polypeptides-PhTET2 hydrolytic activity on synthetic peptides was determined through a procedure inspired from Frottin et al. (42). Different commercial polypeptides containing a poorly hydrolyzable residue in position P 1 Ј were assayed. Reactions were initiated by addition of the enzyme (0.1 mg/ml) to 140 l of a pre-warmed reaction mixture containing 0.1 mg/ml o-dianisidine (Sigma), 3 units of horseradish peroxidase (Sigma), 0.5 units of L-amino acid oxidase (Sigma), and 0.1-6 mM peptide in 50 mM PIPES, 150 mM KCl, 1 mM CoSO 4 , pH 7.5. The hydrolytic activity was assayed by monitoring the absorbance of oxidized o-dianisidine at 440 nm, coupling the PhTET2 activity to both L-amino acid oxidase and peroxidase activity, according to the reaction sequence described by Frottin et al. (42). Assays were performed in 1-mm-thick quartz cuvettes. Absorbance evolution was measured using a Beckman spectrophotometer DU 7400 equipped with a thermostat and a 6-position sample changer. The hydrolytic activity of PhTET2 was measured at 40°C. To calculate the concentration of oxidized o-dianisidine in solution, the molar extinction coefficient used was ⑀ ϭ 10,580 M Ϫ1 cm Ϫ1 . Catalytic constants K m and k cat were estimated using the enzymology tools of SigmaPlot version 11.0 from Systat Software, Inc. For the same mass concentration, the molecular concentration of dimer is six times higher than the one of the dodecamer. The difference between the mass and the molar concentrations has been taken into account for the estimation of the kinetic parameters. Four replicates and four enzyme blanks were assayed for each experimental point.
Sucrose Density Gradient Fractionations and Protein Immunodetection Experiments-P. horikoshii cell pellets (0.2 g) were suspended in 1 ml of lysis buffer (150 mM KCl, 50 mM Tris-HCl, 40 mM MgCl 2 , DNase I grade I (Roche Applied Science, 0.05 mg/ml), pH 7.6). The disruption of the cells was achieved by sonication (10 pulses of 20 s at medium power with a Branson Sonifier 50). The crude extracts were clarified by centrifugation at 30,000 ϫ g for 30 min at 4°C (Ultracentrifuge Beckmann Optima-TL 100), and 250 l of the post-membrane protein cytoplasmic extracts (S30) was gently loaded onto 5-25% continuous sucrose density gradients made in 10 ml of Beckman ultraclear centrifuge tubes. The sucrose density gradients were centrifuged at 210,000 ϫ g during 20 h (Beckman L-80 ultracentrifuge, SW-41 Ti rotor). H. salinarum cellular extracts and sucrose density gradient experiments (5-20%) were performed in native hypersaline conditions as described in Chamieh et al. (33).
Purified dodecameric and dimeric forms of the mutated PhTET2 protein were run simultaneously on separate tubes. The gradients were fractionated into 750-l aliquots. In P. horikoshii S30 fractionation experiments, the proteins were precipitated from the fractions during 30 min at Ϫ20°C with trichloroacetic acid (15% final concentration). After centrifugation at 16,000 ϫ g for 30 min at 4°C, the pellets were washed with glacial acetone, and the samples were immediately centrifuged for 15 min at 16,000 ϫ g at 4°C. The pellets were air-dried for 10 min before resuspension in ice-cold 20 mM Tris base. One volume of SDS-PAGE loading buffer (50 mM Tris-HCl, 8 M urea, 2 M thiourea, 100 mM DTT, 3% SDS, 0.1% bromphenol blue, 10% glycerol, pH 6.8) was then added to the samples. These were heated at 95°C for 5 min, and their proteins were separated by SDS-PAGE using 12% gels. Staining was done with Coomassie Brilliant Blue or the proteins were transferred onto Hybond-P PVDF-membranes, and immunoreactive bands were visualized by chemiluminescence as detailed by the supplier (ECL detection kit; GE Healthcare). The signal was detected using a Kodak Image station 4000 mm. Specific PhTET2 antibodies were raised against the following synthetic peptide chosen in the protein primary sequence DERDVDAT-VELMTKALENIHELKI. The anti-HsTET antibodies were raised against two peptides, TRGSQVRIETDDGPV and AHAGDRDSFGVSV. The antisera were used at 1:2500 and 1:10,000 dilution for PhTET2 and HsTET, respectively. The experiment was repeated at least three times with P. horikoshii and H. salinarum cell batches from different cultures.

Purification and Structural Characterization of a Dimeric
PhTET2 Precursor-For this study, the PhTET2 complex from P. horikoshii was chosen. The protein exhibits high thermal stability, and compared with other types of TET complexes, it possesses broader substrate specificity. Additionally, it can be purified in large amounts without adding tags that may create artifacts when studying the oligomerization process. When recombinant PhTET2 proteins are expressed in E. coli, only the dodecameric complex was detected in the soluble post heatshock fractions. Moreover, once assembled, the TET particles are extremely robust in vitro and can only be broken down into smaller oligomeric forms by using extreme pH conditions and EDTA treatments (29). These harsh physicochemical treatments do not allow the functional characterization of the low molecular weight species; moreover, the process is only partially reversible and generates aggregates. We have therefore used a site-directed mutagenesis approach to study the TET assembling process and the influence of the enzyme's oligomerization state on its catalytic and thermal stability properties. The examination of the subunit interfaces in PhTET1, PhTET2, and PhTET3 according to their crystallographic structures, as well as the structural study of a 24-subunit PhTET1 complex, suggested that the dimer is the building block of the PhTET edifices (19,28). Point mutations were therefore designed to weaken the interactions at the interfaces between the dimers within the dodecamer. Some of them were changed to serine instead of alanine to completely eliminate the possible formation of any hydrophobic interaction of the side chain methyl group of the alanine and also to make the interface more polar. However, neither single nor double mutations could appreciably alter the rate of TET dodecamer formation in the post heatshocked supernatant of the transformed E. coli cells, and only after a combination of five mutations (R217S, R220S, F224S, H248S, and I292A) was a significant slowdown of the oligomerization process achieved. The five mutations are mostly located in the prominent intersubunit interaction regions located at the apices of the tetrahedron on residues participating in hydrophobic clusters but also in polar contacts (for details see supplemental Fig. 1).
When the post-heat shock protein extracts of E. coli cells expressing the PhTET2 pentamutant were resolved on a Resource Q ion exchange column, two well separated elution peaks were obtained, whereas the expression of the wild-type PhTET2 protein produces only one peak. The corresponding fractions were pooled (pool A and pool B). Native gel electrophoresis analysis revealed that pool A contained a homogeneous population of low molecular weight oligomers, and pool B contained predominantly large oligomers that migrated as the wild-type TET dodecamer (Fig. 1A). The two oligomeric forms of the mutated PhTET2 proteins were further purified by size exclusion chromatography (Fig. 1B). The A pool contained PhTET2 oligomers with an apparent molecular mass of Ͻ100 kDa, whereas the B pool principally contained a large complex. According to the molecular weight calibration of the size exclusion column and to negative staining electron microscopy images, the low molecular weight form would correspond to a free TET dimer (78 kDa), although the larger complex was a dodecamer (468 kDa), as clearly shown by the negative staining electron microscopy pictures that revealed its tetrahedral shape, identical to that of the wild-type TET (Fig. 1B).
TET Low Molecular Weight Complexes Co-exist with the Assembled Dodecamers in Vivo-To gain insight into the oligomerization state of the PhTET2 aminopeptidase complex in vivo, post-membrane protein cytoplasmic extracts (S30) from P. horikoshii cells were resolved through continuous sucrose density gradients. Fractions were collected, and the PhTET2 proteins were immunodetected in each fraction by using a specific antibody ( Fig. 2A). Two major populations of native PhTET2 proteins complexes were found to accumulate in the cellular extracts, a high and a low molecular weight species. To assess their respective oligomerization state, purified pentamutant PhTET2 dodecamer and dimers were analyzed on separate gradients, in the same experimental conditions. The dimeric form was obtained from a PhTET2 mutant that is described in this study. It showed that the high molecular weight complex detected in P. horikoshii sedimented as the purified recombinant dodecamer, although the low molecular weight species would correspond to a PhTET2 dimer ( Fig. 2A). This indicates that the five mutations introduced in the PhTET2 interfaces indeed lead to the stabilization of the low molecular weight form observed in vivo ( Fig. 2A). This result is interesting because, when wild-type PhTET2 proteins are expressed in E. coli, only the dodecameric complex is detected in the soluble post heat shock extract.
To confirm the presence of unassembled TET complexes in the cytosol of archaea, we performed similar cell fractionation and immunodetection experiments in H. salinarum, an extreme halophilic organism that accumulates multimolar salt concentration in its cytosol (43). For this reason, all experiments on H. salinarum samples, including sucrose gradients sedimentation, were performed in physiological hypersaline conditions (3.4 M KCl) as described in Chamieh et al. (33). The results are shown in Fig. 2B. As in the P. horikoshii extracts, a low molecular weight species of TET with an apparent molecular mass of Ͻ100 kDa is present together with the fully assembled TET particle. We therefore concluded that, in vivo, a significant population of low molecular weight TET FIGURE 1. Purification of a PhTET2 dimeric complex. A, native gel electrophoresis analysis of the PhTET2 protein that was obtained after expression of the recombinant wild-type and the pentamutant (R217S, R220S, F224S, H248S, and I292A) proteins in E. coli, heat shock clarification and separation by Resource Q ion exchange chromatography. WT is the purified wild-type PhTET2 dodecamer; Pool A corresponds to a low molecular weight complex, although the major constituent of Pool B is a high molecular weight complex that migrates as the wild-type PhTET2 dodecamer. B, Superose 6 column chromatography purification of dimeric and dodecameric mutant PhTET2 complexes. The A 280 absorption profiles are presented together with negative staining electron microscopy images taken from aliquots of the pooled peak fractions. The elution volume and the shape of the purified complexes from pools A and B correspond to dimers and TET dodecamers, respectively. This information was further confirmed by AUC (Fig. 5) and SAXS (Figs. 3 and 4).  . horikoshii and H. salinarum cells. A, immunodetection of the native TET2 complexes from P. horikoshii (Ph). Post-membrane extracts (S30) were loaded on a 5-25% sucrose density gradient and resolved by ultracentrifugation. The fraction numbers are indicated on top, and the arrow represents the sedimentation orientation. The proteins from each fraction were separated on a 12% SDS-polyacrylamide gel, and the presence of the PhTET2 protein was revealed by Western blot by using specific antibodies. The bottom panels show Coomassie Blue staining of 12% SDS-polyacrylamide gels. In this experiment, purified dimeric and dodecameric PhTET2 complexes were resolved on a sucrose gradient by using the same protocol as for the S30 extracts. These complexes were obtained from a mutant protein in which amino acids have been changed to slow down the particle assembling process (see Fig. 1). The positions of the dimer and the dodecamer in the gradients are indicated and correspond to the immunodetected PhTET2 complexes in the cell extracts. B, immunodetection of the native TET complexes from the extreme halophilic strain H. salinarum. S30 extracts were fractionated in native hypersaline conditions on a 5-20% sucrose gradient. The gradient was calibrated with halophilic protein complexes of known molecular sizes (indicated above) as described in Chamieh et al. (33). A low molecular weight TET complex co-exists with the assembled TET dodecamer in the cytosols of P. horikoshii and H. salinarum.

Quaternary Structure and Thermal Stability Comparisons between the Wild-type and the Pentamutant PhTET2
Dodecamers-The presence of low molecular weight forms of the TET complex in vivo prompted us to study their structural and biochemical properties as well as the TET assembling mechanism. The structural and biophysical properties of the mutant dodecameric particle were therefore studied to determine the potential effects of the mutations on its final quaternary edifice. To this end, SAXS curves were recorded on the mutant dodecamer in solution and compared with those of the PhTET2 wild-type protein. Both sets superposed nicely with the SAXS data from the wild-type TET, even when the experiments were performed at 80°C, thus showing that the quaternary structure of the mutant is identical and equally stable as one of the wild-type TETs at the extreme physiological temperatures in which P. horikoshii thrives (Fig. 3). The agreement of the theoretical SAXS curve (calculated with CRYSOL) with the experimental data is good, in particular when the missing loop and the N terminus (not visible by x-ray crystallography) were modeled. The remaining discrepancy at the side minima could be explained by the conformational flexibility of these two fragments that produce a smearing of side minima in SAXS curves of globular objects (Lindner and Zemb (44)).
These experiments show that WT and mutant PhTET2 particles are therefore extremely stable once they are formed, and the deleted interactions located at the dimer-dimer interfaces do not contribute to the stability of the dodecameric structure at high temperatures.
SAXS Structure of the Free TET Dimer-Because the dimer is the basic building unit in the dodecameric TET particle (19), it is important to characterize its structure further to understand the mechanisms underlying the TET oligomerization process and to address the question of the physiological significance of unassembled dimers in vivo. Although crystallographic studies of the mutated free dimer are not possible due to its self-assembling properties, its oligomerization kinetics are slow enough to allow SAXS studies on monodisperse dimeric samples, even at 80°C. The SAXS experimental curve obtained from the free dimer superposed nicely, up to a scattering vector Q ϭ 0.15 Å Ϫ1 , with that calculated (with CRYSOL) from the crystallographic dimer structure within the wild-type dodecameric assembly (PDB entry 1Y0R) (Fig. 3), in particular when the missing internal loop and the N terminus were added. The minor mismatch around Q ϭ 0.20 Å Ϫ1 can probably be explained by a conformational flexibility of the loop and N terminus as in the case of the dodecameric structure in Fig. 3. Moreover, the crystallographic structure of the dimer can be embedded very nicely into the ab initio envelope (Fig. 4B). A dimer of dimers (tetrameric V-shaped structure) and a trimer of dimers (hexameric triangularly shaped structure) did not fit the experimental SAXS data. We therefore conclude that the structure of the free dimer in solution is very similar to that of the wild-type dimer within the dodecamer. Thus, no major conformational changes of the TET dimer precursor seem to be necessary to constitute the 12-subunit tetrahedral particle.
Biophysical Characterization of the PhTET2 Oligomerization Process-The results described above showed that the mutated PhTET2 is fully able to self-assemble into a bona fide TET machine that maintains the same quaternary structure and thermal stability properties as the wild-type TET. Consequently, the pentamutant represents a model to study the oligomerization process. To identify the parameters controlling the equilibrium between the different oligomeric forms, samples were incubated in different buffers after the size exclusion purification step (see "Experimental Procedures"). At physiological pH, the formation of dodecamers from the dimer can be stimulated by a slight increase in salt concentration, although the dimer is stabilized at high pH. Native gel electrophoresis analysis performed at different times after incubation of the purified dimer revealed well defined bands corresponding to intermediate oligomeric states between the dimer and the dodecamer (Fig. 5A). The different oligomeric species were characterized by AUC (Fig. 5, B and C). Native gels electrophoresis and AUC analyses showed the same number of intermediates. Sedimentation profiles displayed multiple different peaks at 4.70, 6.78, 9.29, 11.91, and 15.75 S, indicating the presence of five species in the sample. The same number of species was observed on native gel electrophoresis. Using the Svedberg equation (45), we calculated theoretical sedimentation coefficients for dimeric and dodecameric PhTET2 as well as for putative intermediates (Fig. 5C). Accordingly, the peaks were assigned to, respectively, dimer, tetramer, hexamer, octamer, and dodecamer of PhTET2.
The native gel analysis of the different assembling intermediates revealed that the tetrameric form is the first one to be detected. Interestingly, the accumulation of the octamer precedes the apparition of the hexamer; we could not detect the decamer, and no intermediate forms with odd subunit numbers were identified (Fig. 5C). These experiments demonstrated that the TET oligomerization process is a nonrandom stepwise process.  AUGUST 2, 2013 • VOLUME 288 • NUMBER 31

JOURNAL OF BIOLOGICAL CHEMISTRY 22547
Electron Microscopy and Structural Modeling of the Different PhTET2 Intermediates-The PhTET2 assembling intermediates identified by AUC and native gels were separated from the dimer precursors and from the final TET dodecamer by using gel filtration chromatography. The column fractions were analyzed by electron microscopy. Three types of well structured edifices were identified in the samples as follows: a tricornshaped complex, an open chain, and a complex with a 4-fold symmetry. The sizes of these complexes were consistent with the hexameric and octameric intermediates complexes.
Smaller, less structured species that would correspond to the tetramers could also be seen. To propose structural models for the octameric and hexameric TET assembling intermediates, we generated the dimeric building blocks from the crystal structures of PhTET2 (PDB code 1Y0R). By using the oligomerization interfaces that were identified in the PhTET dodecamers, it was possible to generate a tricorn complex made of three dimers (Fig. 6). Alternatively, three dimers can also assemble as an open chain with a Z shape (Fig. 6). These two assemblies can be recognized in the dodecamer.  The existence of complexes with a 4-fold symmetry cannot be modeled from the 12-subunit dodecamer structure. We used instead the 24-subunit structure that has been described for PhTET1 (19). Indeed, this tetraicosameric edifice is not fashioned through the assembly of two dodecamers, but the octamer is one vertex of the 24-mer assembly. The octamer uses different intersubunit contacts than the dodecamer. In particular, the interface with the five proposed mutations is not involved in the octamer. As the root mean square deviations between the PhTET1 (PDB code 2CF4) and PhTET2 dimers were 5.1 Å over 660 C-␣ atoms, we used the PhTET2 dimer for modeling octamer. Accordingly, we extracted an octameric structure from the PhTET1 24-subunit one and superimposed the PhTET2 dimer structure to produce a PhTET2 octamer model. The structures corresponding to the octameric and hexameric models are presented in Fig. 6. Low resolution envelopes were generated from the corresponding coordinates files for back projection using Spider (41). The experimental electron microscopy images were compared with the different views of the complexes (see "Experimental Procedures") ( Fig.  6). As a control, the same work was performed with the final TET particle to clearly distinguish the differences between the shapes and dimensions of the tricorn and the TET dodecamer, respectively. Model-specific images could be found for each orientation of the complexes. The shapes and dimensions of the single particles identified by electron microscopy are consistent with those arising from the modeling, the AUC, and the native gel experiments. These results indicate that the proposed octameric and hexameric (tricorn and open chain) structures correspond to the octameric and hexameric intermediates that were identified in the PhTET2 assembling pathway.
The octameric square complex is built upon different interfaces to the two hexameric forms and the final TET dodecamer. The same interactions can be used to generate the tetramer intermediates. The five mutations that were designed to slow down the TET assembling process do not impact the alternative contact area present in the tetramers and octamers. Accordingly, the mutations favor the accumulation of tetramers and octamers, although the efficient formation of hexamers is delayed. This explains why the tetramers and octamers preceded the apparition of the hexamers during the assembling process of the PhTET2 pentamutant. It is noteworthy that the formation of the TET dodecameric apices that possess a 3-fold symmetry cannot arise directly from the tetrameric or octameric structures. This indicates that the tetramers and octamers are the products of an alternative pathway induced by the mutations.
Oligomerization Triggers PhTET2 Activity toward Large Polypeptidic Substrates-To study the role of the oligomerization with respect to the functional state of the aminopeptidase, the hydrolytic activities of the mutated PhTET2 dimer and dodecamer were first measured by using short chromogenic aminoacyl compounds as described by Durá et al. (27). These experiments were performed on a time scale of several minutes after the size exclusion purification step. As a control, aliquots were taken from the reaction mixture and analyzed by size exclusion chromatography and native gel electrophoresis to be sure that the dimers and dodecamers had not formed other multimers during the experiments. The rate of Leu-pNA cleavage performed by the enzyme dimeric forms and its comparison with the one of the mutant dodecamers revealed that, at 40 and 80°C, the activity of the two types of oligomers is nearly identical. The K m and k cat values were found to be very similar for the two types of TET assemblies (Table 1). This study shows that the dimeric form of TET is active and that the assembling of the protein into a 12-subunit tetrahedral superstructure does not modify significantly its amidolytic activity toward small substrates. The aminopeptidase capability of the dimer raised the question of the biological significance of the TET supramolecular complex. We previously showed that TET peptidases display a significant hydrolytic activity toward the N-terminal residue of oligopeptides up to 27 amino acids in size (27). When the activities of the mutant PhTET2 dodecamer and dimer were tested against longer peptides such as AAA-pNA and AAAA-pNA, a significant lag phase was observed in the case of the dimer. This suggests that the size of the peptide is an important parameter for substrate hydrolysis when comparing dimers to the fully assembled TET complex. To obtain kinetic constants from long polypeptides (about 10 amino acids in length), substrates that contained a favorable N-terminal residue and an unfavorable residue in position P 1 Ј were used. In this manner, the quantity of released amino acids reflects only the hydrolysis of the first residue, allowing the calculation of the kinetic parameters. The enzymatic assay employed was inspired from that developed by Frottin et al. (42). Again, the oligomeric state of the TET aminopeptidase was checked before and after the experiments. Two oligopeptides were tested, Met-Lys-bradykinin (MKRPPGF-SPFR) (Fig. 7) and MBL peptide 176 VDLTGNRLTY 185 (data not shown). In both cases, we observed that the dodecamer was more active than the dimer, and unlike for the short peptide experiment, an important difference in the reaction speed between the two PhTET2 oligomeric forms was observed. We measured the K m and k cat values of the dimer and dodecamer of the PhTET2 pentamutant for the Met-Lys-bradykinin peptide ( Table 1). The K m values of the dimer and the dodecamer were found to be identical. However, the k cat value of the dimer was significantly lower than the one of the dodecamer. Thus, the TET reaction efficiency toward long peptides is compromised when the enzyme is not assembled as a dodecameric complex. These results demonstrate that oligomerization allows the TET peptidase to better process long polypeptides. Therefore, the dimeric and dodecameric forms that we detected in vivo in two archaeal strains (Fig. 2, A and B) may perform different physiological roles.
PhTET2 Dimer Is Stable under High Temperature Conditions-The TET dimer represents the minimal assembling unit in the TET oligomerization pathway (19), and this observation raises the question of the dimer stability under the extreme temperature conditions that prevail in the natural environment in which Pyrococcus cells thrive. The SAXS experiments described here have demonstrated that, in vitro, the dimer population does not show any sign of structural alteration at 80°C, within 1 h (Fig. 4A). To further explore the thermal stability of the TET dimer, the residual aminopeptidase activity after different times of incubation at 80°C was measured as described previously (27). Leu pNA was used as a substrate. Aliquots were  Table 1.

TABLE 1 Kinetics constant of mutated PhTET2 dimer and dodecamer against a short (Leu-pNA) and an 11-residue (Met-Lys-bradykinin) peptide
Kinetics constant of mutated PhTET2 dimer and dodecamer against a short (Leu-pNA) and an 11-residue (Met-Lys-bradykinin) peptide. The k cat is the catalytic efficiency of the enzyme; the K m is the Michaelis constant of the enzyme reflecting the affinity of the enzyme for its substrate. The catalytic efficiency of the dimer on a long peptide is lower than the one of the dodecamer. The dodecameric form of PhTET2 is more efficient at processing long peptides than the dimeric form.

DISCUSSION
A key question regarding self-compartmentalized peptidase complexes are the mechanisms by which such well organized edifices assemble. In this study, a site-directed mutagenesis strategy has been used to slow down the self-assembling process of TET, a large dodecameric aminopeptidase present in the three domains of life. The disruption of stabilizing interactions located in the interface area between the subunits at the apices of the tetrahedron had a significant effect on the kinetics of the TET assembling process. This way, a stable dimeric species could be purified that was fully able to self-assemble into active dodecamers. SAXS studies showed that the dodecameric particle obtained from the mutant dimer displayed the same stability and quaternary structure as the wild-type TET complex, even under extreme temperatures.
Native gel electrophoresis, AUC, and electron microscopy analyses proved that the mutant PhTET2 oligomerization does not proceed by an incremental addition of dimers but involves well defined intermediate species that are generated with time from the PhTET2 dimer precursor. Although tetramers and octamers were detected at the beginning of the oligomerization process, they are the products of an alternative pathway induced by the mutations. Based on SAXS analysis of the free PhTET2 dimer, electron microscopy study, and structural modeling from existing crystallographic structures of the PhTETs (19,23), we propose that the natural pathway involves two hexameric intermediates, a tricorn and an open (Z-form) chain complex. Indeed, the tricorn represents a stable oligomeric form in solution. However, the Z-form is able to self-associate, leading to the extended interaction zone involving the catalytic domains of three subunits present in the highly stable biological dodecamers. Consequently, we propose that the open and closed hexamers are in equilibrium and represent the intermediates in the PhTET2 dodecamer assembling process (Fig. 8). The oligomerization interface that is present in the PhTET2 hexamers and dodecamers is well conserved in the three PhTETs (data not shown). This suggests that the oligomerization process and the associated functional activation that we described for PhTET2 may also be valid for PhTET1 and PhTET3.
Self-compartmentalization is a hallmark of cytosolic peptidases that degrade peptides in an unspecific manner (46). In the proteasome, tricorn protease, and bleomycin hydrolase, the sequestration of the catalytic sites from the bulk environment into chambers represents a peptide filtering system that is necessary to prevent unwanted damage on folded polypeptides within the cytosol (47-49). Oligomerization has also been shown to control different aspects of the peptidase functions. In the 20 S proteasome and its bacterial homologue ClpP, the priming of the peptidase enzymatic activity is coupled with sub- The dimer constitutes the building block and self-assembles into a closed hexamer, which involves the association of three dimers (red, magenta and blue). Two open "mirror" conformations of this complex are in equilibrium and finally associate to form the super-stable dodecameric particle. unit association (50). In ATP-independent peptidases, such as protease 1 and Pab87, the active sites are formed at the subunit junctions (16,17). In tripeptidyl peptidase II, a giant aminopeptidase found in eukaryotes, the activity increases in a nonlinear fashion with the oligomerization rate (11), and in the case of bovine lens leucyl aminopeptidase, it has been suggested that the activity depends on the stabilization of each monomer catalytic site by the structure of the oligomer (51). On the contrary, in ␣-1 tryptase or kallikrein-related peptidases, self-oligomerization provides an inhibition of the enzyme activity (52,53). With respect to these findings, it was surprising to find out that the dimeric form of the TET complex still carries out the same catalytic activity on small peptides as the 12-subunit complex. This indicates that, in the TET dimers, the catalytic sites and pockets are already positioned in proteolytically active conformations. The SAXS structure of the free dimer is consistent with these findings, because it shows that the relative positions of the two monomers of each dimer within the tetrahedral complex are already imposed by the interactions between dimerization domains. In the case of this small peptide, the calculated K m values are similar, meaning that the substrate is recognized by the dimer and the dodecamer equally well. The similar calculated k cat value reflects that the active site remains unchanged upon oligomerization. Thus, in the case of PhTET2, the oligomerization does not affect the active site and catalytic pocket configurations and has little effect on the recognition and trimming of N-terminal amino acids. However, when studying the catalytic activity of TET dimers and dodecamers as a function of substrate length, we found that the dodecamers are more efficient in hydrolyzing long polypeptide substrates as compared with the dimers. The kinetic parameters indicated that the dodecamer possesses a better efficiency than the dimer toward long substrates as follows: the k cat value of the amidohydrolytic reaction, reflecting the catalytic efficiency of the system, is considerably reduced for the free dimer, although the K m value, reflecting the affinity of the active site and the catalytic pocket for the N-terminal amino acid, remains the same between the dimer and the dodecamer. Thus, during the reaction, the same number of peptidase-substrate complexes are formed, but these complexes are more productive in the case of the dodecamer. In contrast, the peptidase-substrate complexes formed by the dimer are unproductive and are not true Michaelis-Menten complexes and therefore cannot be turned over. Therefore, in the case of the PhTET complex, the self-compartmentalization provides a way to enhance the enzyme efficiency when the substrate size increase. This represents another type of peptidase functional regulation driven by self-oligomerization. The interior of the TET peptidases consists of four polypeptide navigation channels that cross the particles from the entry pores situated on the facets of the tetrahedrons toward the catalytic chambers located within each apex (19,28). A model for polypeptide processing was proposed based on the structural and enzymatic comparisons of the three TET enzymes from P. horikoshii. In this model, a series of mobile loops and electrostatic attractions/repulsions in the entry channels and in the catalytic chambers would orient the N terminus of the polypeptides toward the negatively charged active sites (19,28). In unassembled TET dimers, the polypeptide naviga-tion system is not present. This would explain why the TET dimer is disabled to break down long polypeptides.
P. horikoshii is a hyperthermophilic microorganism that grows optimally at 95°C (54). In this work, it has also been shown that, in vitro, the free dimers are as stable at physiological extreme temperatures as the assembled TET particles, with half-lives of several hours at 80°C. Different strategies are adopted by thermozymes to stabilize their native conformation at extreme temperatures (55). Among these, oligomerization has been proposed to represent an important determinant (56). In the case of the TET system, our results show that the dodecameric quaternary structure has little role in the high thermal stability of the enzyme. In fact, the monomer-monomer interfaces in the TET dimer consist of extended networks of ionic bonds that are likely to contribute to the high thermal stability of the enzyme as a free dimer. Therefore, dimers can be accumulated as stable precursors of the TET dodecamers in vivo.
There is little information available about the in vivo oligomeric states of large energy-independent peptidase complexes. We showed here that the PhTET2 dimer co-exists with the dodecamer in cellular extracts. A similar observation was also made under hypersaline conditions in the extreme halophilic archaeon H. salinarum. Unlike Pyrococcus, the Halobacterium genome contains only one copy of the TET peptidase. Thus, it is reasonable to propose that the dimer-dodecamer equilibrium is a hallmark for TET peptidases and that a specific regulation occurs in vivo to control the oligomerization state of TET. Because oligomerization affects PhTET2 activity with respect to the size of the polypeptidic substrates, a regulated oligomerization would therefore allow the TET particle to process a broad variety of peptides, from dipeptides to long polypeptides, in response to varying physiological demands; although the TET dimer would act preferentially on small peptides for energetic and anabolic purposes, its assembly into dodecamers and the concomitant formation of the polypeptide navigation system would trigger the intracellular aminopeptidase activity toward longer peptides such as those produced by the 20 S proteasome or toward peptides possessing specific biological activities. Thus, the control of the amount of assembled TET in vivo may represent an important regulatory step in protein degradation and in many specific biological functions based on polypeptide activity. Regarding that, in vitro, the TET oligomerization appears to be a rapid process, and given that the assembled TET particles are extremely robust, it is likely that the in vivo regulatory mechanisms involve the stabilization of the dimeric species. Studies are in progress in our laboratory to identify this mechanism.