A Lon-Like Protease with No ATP-Powered Unfolding Activity

Lon proteases are a family of ATP-dependent proteases involved in protein quality control, with a unique proteolytic domain and an AAA+ (ATPases associated with various cellular activities) module accommodated within a single polypeptide chain. They were classified into two types as either the ubiquitous soluble LonA or membrane-inserted archaeal LonB. In addition to the energy-dependent forms, a number of medically and ecologically important groups of bacteria encode a third type of Lon-like proteins in which the conserved proteolytic domain is fused to a large N-terminal fragment lacking canonical AAA+ motifs. Here we showed that these Lon-like proteases formed a clade distinct from LonA and LonB. Characterization of one such Lon-like protease from Meiothermus taiwanensis indicated that it formed a hexameric assembly with a hollow chamber similar to LonA/B. The enzyme was devoid of ATPase activity but retained an ability to bind symmetrically six nucleotides per hexamer; accordingly, structure-based alignment suggested possible existence of a non-functional AAA-like domain. The enzyme degraded unstructured or unfolded protein and peptide substrates, but not well-folded proteins, in ATP-independent manner. These results highlight a new type of Lon proteases that may be involved in breakdown of excessive damage or unfolded proteins during stress conditions without consumption of energy.


Introduction
Proteases play a crucial role in cellular protein quality control to degrade aberrant or unwanted proteins, which may be prone to toxic aggregate formation, into short peptides. Most of these specialized proteases are oligomeric and form an enclosed chamber, inside which the proteolytic sites are located and accessible only through a narrow axial pore [1]. In most cases, selected protein substrates are fed through the axial pore into the degradation chamber of the protease in an energy-dependent process by a ring of built-in motor subunits or domains containing conserved motifs for ATP binding and hydrolysis found in the superfamily of AAA + (ATPases associated with various cellular activities) [2].
Lon proteases are key components of the protein quality-control systems in bacteria and eukaryotic organelles responsible for degrading abnormal or unfolded proteins; they are also involved in regulation of many cellular processes by selective degradation of regulatory proteins [3]. Lon was originally identified from Escherichia coli (hereafter EcLonA) and was the first known ATPdependent protease [4]. Almost all Lon proteases possess two important functional domains in a single polypeptide chain: the AAA + domain, which contains conserved motifs (Walker motifs) for ATP binding and hydrolysis, and the C-terminal protease domain with a conserved serine-lysine catalytic dyad in its proteolytic active site [5]. Based on the consensus sequences at the proteolytic sites in the protease domains, two subfamilies of Lon proteases, LonA and LonB, have been classified previously [6]. In general, the two subfamilies are associated with two distinct consensus sequences of the Walker motifs in the AAA + domains and specific domain organization features in which an N-terminal domain is attached to the AAA + modules of all LonA but not LonB members, the latters are found only in archaebacteria and possess instead additional membrane-spanning segments integrated into their AAA + domains that anchors the proteins to the membrane [7]. Such proteolytic site-based classification approach, however, has left out a large group of Lon-like proteases encoded in the genomes of many Gram-negative bacteria (also in certain Gram-positive bacteria and archaea), which contain LonB-like consensus sequences at the proteolytic sites but have no transmembrane regions nor detectable AAA + consensus motifs [6,8].
Recently, the lack of ATPase activity has been confirmed for a Lon-like protease from Thermus thermophilus [9]; however, the biochemical properties and functional roles of these Lon-like proteases have remained largely uncharacterized.
For the present work, we performed multiple sequence comparisons of LonA, LonB, and the Lon-like proteases, which revealed additional sequence features in the protease domain unique only to the latter but not present in the ATP-dependent LonA/B members. Phylogenetic analysis based on sequence alignment results indicated that these Lon-like proteins with unique conserved insertions belong to a clade of Lon proteases distinct from the LonA and LonB groups. This group of Lon-like proteases may be assigned to a new subfamily, LonC, following previous classification scheme [6]. Here we report the biochemical and biophysical characterizations of a Lon-like protease from M. taiwanensis [10].

Sequence Alignment and Phylogenetic Analysis
We have recently sequenced the total genome of Meiothermus taiwanensis WR-220 and identified three Lon homologs in this indigenous thermophilic Gram-negative bacterium. While two were LonA-like, the third gene encoded a novel Lon-like protein of 719 amino acids (hereafter MtaLonC) with a conserved Cterminal Lon protease domain but containing neither LonA-like N-terminal domain, nor transmembrane regions associated only with LonB or the conserved features of AAA + modules (see below). MtaLonC expressed in M. taiwanensis at 55uC, the organism's growth temperature, and the expression was induced at 65uC [ Fig.  S1]. We conducted BLAST searches of the protein databases and identified a large number of bacterial sequences with BLAST Evalues suggesting homology with MtaLonC. Most of these Lon-like proteases are found in the proteomes of Gram-negative bacteria, of which many belong to the medically important class of cproteobacteria, which includes the genera of Acinetobacter, Klebsiella, Pseudomonas, Salmonella, Vibrio, and Yersinia. We performed multiple sequence alignment for these Lon-like proteins, along with studied members from the LonA and LonB subfamilies, and compared the results using three different programs: MUSCLE, T-Coffee, and ClustalW2 [11][12][13]. A representative phylogenetic tree calculated from the MUSCLE alignment of Lon homologs is presented in Figure 1. It is clear that three major clades, evident in all phylogenies calculated using each of the three different alignments, are present and two of them correspond to the previously classified LonA and LonB subfamilies [6]. Using the same naming scheme, here the third group was tentatively assigned as LonC. It is noteworthy that LonB and LonC may have derived from a common ancestor, which in turn shares a common ancestor with LonA ( Fig. 1, Table S1).
Alignment of the conserved Lon protease domains from all three subfamilies revealed two insertions unique to the LonC group (K634-G640 and V660-V664 of MtaLonC)( Fig. 2A, B and C). According to the crystal structure of the protease domain of EcLonA, one insertion has a consensus sequence of CX 3 GLX(G/ E) and is located in the loop connecting the helix a3 and strand b9, two structural elements conserved in all known structures of isolated Lon protease domains [14][15][16][17][18]. The other insertion with a consensus sequence of wX 2 Aw (w = hydrophobic amino acid) is within helix a5, which adapts an alternative loop conformation in two structures [14,18]. Based on the recent structure of a hexameric LonB (TonLonB; PDB code 3K1J) [19], all structural elements (a3, a5, and b9) are on the outer surface of the degradation chamber (Fig. 2D). Interestingly, they are located at the contact interface between the a subdomain of the ATPase module and the protease domain; hence the LonC-specific insertions may affect the communication between the two functional units.

Hexameric Assembly without Mg 2+ or Nucleotides
Full-length MtaLonC was cloned and expressed in E. coli as a C-terminal His-tagged protein, which appeared to be wellfolded (Fig. S2). The protein had a tendency to form aggregates eluted as a sharp peak at the void volume when concentrated but appeared as a single hexameric species at 1 mg/mL in pH 5-8, with a molecular mass of 420-480 kDa as judging by size-exclusion chromatography. Formation of MtaLonC hexamers was not dependent on Mg 2+ or ATP, a property different from EcLonA [20]. The hexameric assembly was also confirmed by analytical ultracentrifugation, from which an apparent molecular weight (Mw.) was estimated of 460 kDa, corresponding to a hexamer ( Fig. 3A and Fig. S3; monomer Mw. = 79 kDa). To examine the oligomeric assembly in further detail, we performed electron microscopic (EM) analysis. Electron micrographs of the hexamer and void-volume fractions from Superose 6 all showed a uniform size of negatively stained particles with two general forms (Fig. 3B) and no larger particles with a higher ordered structure could be observed in the voidvolume fraction. Most of the particles with a likely en face orientation appeared as symmetric hexamers (for example, images 1, 15, 20, and 23), with a dimension of 13 nm comparable to that of hexameric TonLonB (120 Å ). Also observed were egg-shape particles with central dark region implying a hollow chamber (images 5, 10, 26 and 27). In some cases, tunnel-like feature could be identified as well, which together with the hollow chamber were also found in the hexameric structure of TonLonB [19].

Lack of ATPase Activity
One of the main sequence features of the LonC proteases is their lack of canonical AAA + motifs. Previously, a similar Lonlike protein TTC1975 from Thermus thermophilus, which belongs to the LonC subfamily ( Fig. 1), was shown to have no ATPase activity and its proteolytic activity was not stimulated by the addition of ATP [9]. Similar to TTC1975, we found no ATPase activity associated with MtaLonC at either 37uC or 55uC, the optimal growth temperature for the thermophilic bacteria, using a colorimetric assay [21,22]; by contrast, the activity was detected for EcLonA, as well as a LonA protease from the thermophilic Brevibacillus thermoruber (BtLonA) (Fig. 4A and B). The ATPase activity was also tested by monitoring release of inorganic phosphate (Pi) from ATP hydrolysis by the enzyme by 31 P-NMR at 37uC (Fig. 4 C-F) [23]. As shown, no peak for P i was detected even after 5 days of incubation of ATP with MtaLonC (Fig. 4D), during which the protein appeared to be intact as judging by SDS-PAGE and gel filtration analyses. On the contrary, incubation of an excess amount of ATP with EcLonA resulted in hydrolysis of ATP and the appearance of a peak for P i (Fig. 4F). This result confirmed that MtaLonC has no ATPase activity.
Despite the lack of ATPase activity, we found that MtaLonC still retains an ability to bind nucleotide. Isothermal titration calorimetry (ITC) was used to analyze the binding isotherms of ATPcS and MtaLonC (Fig. 4G). The mixture of ATPcS and MtaLonC yielded exothermic titration curves indicative of Figure 1. Phylogenetic tree calculated from amino acid sequence alignment of Lon-like proteases (LonC) with representative members of LonA and LonB subfamilies. Three distinct clades can be recognized through phylogenetic reconstruction. Indicated are the names of species where at least one member of the corresponding subfamily of Lon proteases is present. Note that LonA proteases are ubiquitous in all species shown; thus some bacteria, including E. coli, possess more than one subfamilies of Lon. In addition, multiple paralogs related to LonA have been found in certain bacteria. The Lon genes are indicated by binomial names of the species. The accession codes for the genes can be found in Table S1. doi:10.1371/journal.pone.0040226.g001 binding to the nucleotide. The titration isotherm curve of the complexes was fitted with a one-site model and gave a binding stoichiometry of n = 1.15, suggesting an equal binding of 6 nucleotides per hexamer, with a K d of 7.8 mM. By contrast, TonLonB exhibited a binding of 3 ATPcS per hexamer with a higher affinity (2 mM) [19]. In fact, a maximum occupancy of four nucleotides to the six available ATP/ADP binding sites in the hexameric ring of AAA + modules were observed from solution studies [24][25][26], as well as in some crystal structures [19,27,28]. As proposed previously, such unequal binding of nucleotides may be necessary for the ATPase cycle to drive conformational changes in the hexameric AAA + ring to generate mechanical power for substrate unfolding and translocation [1]. The fact that MtaLonC has no ATPase activity but binds six nucleotides in hexameric assembly points out a possibility that it may contain six degenerate AAA-like modules, which may have not lost nucleotide-binding ability but cannot generate power strokes to unfold protein substrates as canonical AAA + proteases do. A strong support to this hypothesis comes from sequence alignment of TonLonB and MtaLonC, from which we identified a Walker A-like motif near the N-terminus (GPPSLGKHE; conserved residue underlined) and a Walker B-like motif (''Walker B'': GYLILDA) in the middle portion of the sequence where the conserved Glu (preceding the conserved aspartate) essential for ATP hydrolysis is missing and replaced with a leucine. The conserved AAA + residues, sensor-1 and sensor-2, are replaced by non-conserved residues; and the arginine finger is disrupted by a short insertion in MtaLonC (Fig. 2C). Missing all these important AAA+ residues required for sensing and performing ATP hydrolysis in MtaLonC is consistent with its lack of ATPase activity. A structural analysis is needed to validate this hypothesis.

Selective Degradation of Unfolded Peptide or Protein Substrates in ATP-and Divalent Cation-independent Manner
The peptidase/protease activity of EcLonA is known to rely strictly on Mg 2+ (or other divalent cations) and ATP [29]. We found that in the absence of both, MtaLonC readily cleaved F-b20-Q, a fluorogenic peptide derived from an internal segment of b-galactosidase (QLRSLNGEWRFAWFPAPEAV), which contains all the elements required for recognition and degradation by EcLonA in the presence of ATP [30]. Using the b20 peptide, we found that MtaLonC had an optimal peptidase activity between 52 and 65uC, at pH 7.0 (Fig. S4). As shown in Fig. 5A, MtaLonC cleaved b20 at similar rates without or with added ATP, ADP, or nonhydrolyzable AMPPNP. The K M for the ATP-independent reaction by MtaLonC was 14 mM, threefold higher than EcLonA (4.6 mM), whereas the V max was 50fold lower than EcLonA (0.2 vs. 10 mM/min), the latter however had no cleavage activity at all towards b20 in the absence of ATP ( [30] and data not shown). Mass spectrometry analysis of digested b20 products revealed a major cleavage site between the residues F11 and A12. The lower cleavage rate of b20 indicated that this peptide is not as good substrate to MtaLonC as to EcLonA, which may reflect their nonhomologous N-terminal domains mediating substrate recognition/entry to the proteolytic chamber. Nevertheless, MtaLonC still recognizes b20 as a degron. The fact that MtaLonC was able to cleave b20 at the position where the P1 (F11) and P3 (W9) residues are hydrophobic, a result consistent with previous cleavage analysis of ribosomal S2 protein by EcLonA [31], suggests a conserved Lon proteolytic domain present in these two types of Lon. We also examined the proteolytic activity of MtaLonC using four substrates with different thermostabilities.  Bovine a-caseins comprise a S1 -casein and a S2 -casein; both are unfolded phosphoproteins without any secondary structure. We found that MtaLonC preferably degraded a S2 -casein but did so less well for a S1 -casein in the absence of ATP and Mg 2+ (Fig. 5B). Interestingly, MtaLonC did not completely degrade caseins into small peptides; a peptide product with an intermediate size was in fact accumulating in the course of digestion (Fig. 5B). a S2 -Casein is distinguished from a S1 -casein by the presence of clusters of negative and positive amino acids at the N-and C-termini, respectively. Intriguingly, MtaLonC could not degrade dephosphorylated a-caseins; by contrast, EcLonA can efficiently degrade all forms/isoforms of a-casein (Fig. S5) [32]. These results indicated that MtaLonC might possess a unique axial pore, which mediates substrate recognition distinct from that of EcLonA. E. coli Hub has been shown to be a substrate of EcLonA (Fig. 6A) [33]. CD spectra of Hub indicated that the protein was folded at both 25 and 37uC, but exhibited little secondary structural feature at 55uC (Fig. 6B). MtaLonC degraded the protein at 55uC, where Hub was likely unfolded. By contrast, folded Hub was degraded only by EcLonA but not by MtaLonC at 37uC (Fig. 6A). In a related experiment, we found that MtaLonC could not degrade hen lysozyme, which contains four internal disulfide bonds, at 55uC. However, unfolded lysozyme, denatured by TCEP treatment, was completely degraded by MtaLonC (Fig. 6C). These results strongly suggested that MtaLonC could only degrade unfolded protein substrates. As a further test, we engineered a construct, designated as POP1-b8, in which the hydrophobic core of b20 (WRFAWFPA) was cloned to the C-terminus of pyrin-only protein 1 (POP1), a stable 12-kDa a-helical protein. Although only the octapeptide core of the b20 degron was used in the fusion construct, the resulting C-terminal sequence is highly similar to b20. A similar heptapeptide core peptide of b20 has been demonstrated to inhibit degradation of F-b20-Q by EcLonA [30]. While native POP1 was not recognized and degraded by EcLonA (Fig. 7A), the enzyme degraded POP1-b8 into shorter species (Fig. 7B). On the contrary, MtaLonC could not degrade POP1-b8, which exhibited thermal stability from 25 to 55uC (Fig. 7C). Overall, these results indicated that MtaLonC has no unfolding power and thus cannot degrade folded proteins.

Processive Protein Degradation by MtaLonC
Canonical ATP-dependent Lon proteases are known to degrade protein substrates in processive manner [31,34]. To find out whether MtaLonC carried out degradation process in similar fashion, purified E. coli Hub and a S2 -casein were used as substrates and analyzed by reverse-phase HPLC and mass spectrometry ( Fig. 8; Fig. S6). Similar to previous findings, the HPLC profiles of degradation products showed time-dependent increase of most product peptide peaks, which persisted throughout the incubation period [31,34]. These results suggested that processivity is an inherent feature of Lon proteolytic domain and is not dependent on ATP hydrolysis.

Conclusion
In this work, we presented biophysical and biochemical characterizations of MtaLonC, which belongs to a novel class of ATP-independent Lon-like proteases. As revealed by EM, MtaLonC likely forms a hexameric assembly with a hollow chamber. Although devoid of ATPase activity, based on sequence alignment and an ability to bind six nucleotides we propose that MtaLonC hexamer may contain a layer of six AAA-like domains and thus may exhibit an overall threedimensional architecture similar to the recent crystal structure for the non-membrane portion of TonLonB [19]. However, in contrast to the obstructed hydrophobic pore (,5 Å ) in TonLonB, MtaLonC may possess a distinct open pore wide enough to allow only unfolded peptide or protein substrates to pass through, but disallow well-folded proteins to access the degradation chamber. The energy-independent degradation of unfolded substrates by MtaLonC reported here is reminiscent of the proposed role of 20S proteasome core particle (CP) to degrade unfolded oxidized proteins in an ATP-independent and ubiquitin-independent pathway [35]. The outer rings of CP form a 13-Å entrance known as the a-annulus, through which all substrates must pass to enter the proteasome. In this regard, MtaLonC may have a similar open pore. Currently, the functional roles of the Lon-like proteases are unknown. As the energy consumption for degrading polypeptide substrates by EcLonA may be substantial [30,36], it is tempting to speculate that the ATP-independent Lon-like proteases may offer fuel-free housekeeping solution for eliminating excessive unfolded or damaged proteins induced by various stress and oxidative conditions. However, as MtaLonC can degrade the specific acasein isoform, they may also have regulatory roles by degrading specific short-live substrates that cannot be recognized by canonical Lon proteases. Insight into this new type of Lonlike proteases awaits identification of the specific substrates.

Molecular Cloning and protein purification
The thermophilic bacterium M. taiwanensis WR-220 was previously isolated from a hot spring in Wu-rai, Taipei County, Taiwan [10]. M. taiwanensis WR-220 was grown in medium containing 0.3% peptone, 0.1% yeast extract, 0.1% glutamic acid, and Castenholz salts, at pH 7.8 and 55uC. The complete genome of M. taiwanensis was sequenced by Genomics BioSci & Tech (Taipei, Taiwan), and the nucleotide sequence with similarities to the gene encoding the Lon protease was annotated and cloned into expression vector pET21a. The forward primer contained an ATG initiation codon and an NdeI restriction site. The reverse primer complementarily overlapped the 39-end of the coding region and contained a HindIII site. After double-digestion with NdeI and HindIII, the product was purified and ligated into the NdeI/HindIII sites of the isopropyl b-D-thiogalactoside (IPTG)-inducible E. coli expression vector pET21a (+) (Novogen, USA). E. coli strain BL21(DE3) (Novogen, USA) was transformed with the recombinant plasmids. The nucleotide sequence of the gene encoding the Lon protease homolog MtaLonC was deposited in GenBank (accession number: GQ423488). POP1 with the FAWFP core octapeptide (WRFAWFPA) within b20 appended at the Cterminus (POP1-b20) was created by a PCR-based mutagenesis protocol [37] and the construct was verified by DNA sequencing. Proteins were expressed in BL21(DE3) cells by induction with IPTG at 20-25uC for overnight and purified from cleared cell lysates to homogeneity by sequential chromatography on Ni-Sepharose 6 fast flow, Mono Q 5/50 GL, and Superose 6 10/300 GL columns (GE Healthcare). Protein concentration was quantified with Bradford Reagent (Bio-Rad). Recombinant EcLonA and BtLonA (see main text) was expressed and purified as described previously [17,37].

Phylogenetic Analysis of MtaLonC
Protein sequence of MtaLonC was submitted to the BLAST network service of the Swiss Institute of Bioinformatics (SIB). Selected sequences that are highly similar to M. taiwanensis, together with five and four sequences from the LonA and LonB subfamilies, respectively, were downloaded for phylogenetic analysis. Sequences were aligned using the program MUSCLE [38] with default parameters. The aligned sequences were imported to MEGA 5.03 [39] for further analysis. Relationships among groups of proteases were inferred through the neighbour-joining method and the Jones-Taylor-Thornton (JTT) model implemented in MEGA. Rates among sites were uniform and gaps and missing data were completely deleted. The tree reliability was evaluated by performing 1,000 replicates of bootstrap re-sampling.

Electron Microscopy
Diluted MtaLonC solution (, 50 mg/ml) was negatively stained by 2% uranyl acetate (UA). A FEI Tecnai G2 Spirit Twin transmission electron microscope operating at 120 kV was employed to serve the EM data collection. The specimen was imaged on a Gatan 1k61k CCD at a magnification of 30000x, conferring a pixel size of 5.6 Å . The corresponding defocus values were in the range of 22.0 , 22.3 mm (estimated by CTFFIND3 [40]).

Analytical Ultracentrifugation Analysis
Sedimentation velocity was performed at 20,000 rpm with standard double-sector centerpieces at 20uC in a Beckman XL- The data were analysed with the software SEDFIT [41]. The protein sample (7 mM) was in the buffer containing 20 mM sodium citrate, 100 mM NaCl, 1 mM DTT, pH 5.5 with a concentration of 7.8 mM. All samples were visually checked for clarity after ultracentrifugation; no precipitation was observed. Sedimentation equilibrium experiment was performed using an Optima XL-I analytical ultracentrifuge (Beckman Instruments, Palo Alto, CA, USA) in an An-60 Ti rotor equipped with a standard six-channel cell. The protein sample was in the buffer containing 20 mM sodium citrate (pH 5.5), 100 mM NaCl, 1 mM DTT, with a concentration of 7.8 mM.
The protein samples were rotated between speeds 4,000, 5,000 and 6,000 r.p.m. at 20uC until equilibrium was attained. Absorbance was monitored at 280 nm. The apparent molecular weight (MWapp) was obtained by global fitting of multiple scans acquired at 4,000 r.p.m. using the sedimentation analysis software supplied by the manufacturer. ATPase Activity Assay The ATPase activity of MtaLonC was measured as described previously [21,22]. The reaction mixtures contained 50 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 1.0 mM ATP, and 5 mg MtaLonC in a total volume of 100 ml. Solutions were incubated at 37 or 50uC for 40 min, and the color of the reaction was developed by adding 800 ml malachite/molybdate solution. The reactions were terminated by adding 100 ml 34% sodium citrate. The absorbance of the final reaction was determined at 660 nm on a Beckman Coulter DU730 UV/Vis spectrophotometer.

Peptidase Activity Assay
The fluorescence peptide assays were performed with the F-b20-Q peptide {ortho-aminobenzoic acid (Abz)-QLRSLNGEWRFAWFPAPEAV[Tyr(3-NO 2 )]A}, flanked by a pair of fluorophore (Abz) and quencher (nitrotyrosine) [30]. The reaction mixtures contained b20 peptide (5 mM), 2 mM HEPES, 10 mM NaCl, pH 7.5 with or without MtaLonC (0.4 mM) were used as positive and negative control. The samples in the buffer containing b20 peptide (5 mM), 2 mM HEPES, 10 mM NaCl, 10 mM MgCl 2 , pH 7.5 with 1 mM ATP, ADP, or AMPPNP were used as experimental groups. The reactions were carried out at 55uC. The fluorescence spectra were recorded with a Jobin Yvon Fluorolog-3 fluorescence spectrophotometer. Excitation was at 320 nm, and emission was measured at 420 nm.

Secondary Structure Analysis
Far-UV CD spectra were recorded with JASCO J-715 spectropolarimeter. a S2 -Casein was in the buffer containing 0.1 mM NaCl, 20 mM HEPES pH 7.5, whereas MtaLonC was in the buffer containing 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , 3 mM KF and 70 mM NaF, pH 7.4. The POP1-b20 (0.5 mg/ml) was in 50 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 and measured at different temperature. E. coli Hub (4.0 mg/ml) was in 0.1 mM NaCl, 20 mM HEPES pH 7.5 and measured at different temperature. The far-UV CD spectra were the mean of three accumulations with a 0.1-cm light path.

Western Blotting
M. taiwanensis was grown in medium containing 0.3% peptone, 0.1% yeast extract, 0.1% glutamic acid, and Castenholz salts, pH 7.8, at 55uC and 65uC. Bacteria cells were lysed in B-PER bacterial protein extraction reagent (PIERCE, USA). Bacteria lysates (100 mg) were separated by SDS-PAGE and electrotransferred to a PVDF membrane. The membranes were incubated at room temperature for 1 h in blocking solution (5% nonfat milk in phosphate-buffered saline (PBS) containing 0.1% Tween 20) and then at room temperature for 2 h with anti-MtaLonC antibody (1:40000). After three washes in PBS with 0.1% Tween 20, the membrane was incubated with an HRP-conjugated secondary antibody (1:50000). The membrane was developed using an enhanced chemiluminescence Western blot detection system.

Analysis of Protein Degradation by Reverse-phase Chromatography
Bovine casein (200 mg) was incubated with MtaLonC (15 mg) in the buffer (60 ml) containing 50 mM Tris HCl, 10 mM Na 2 HPO 4 , pH 8.0 for 0, 15, 30, and 60 min. After incubation, equal volume (60 ml) of 7.4 M guanidine hydrochloride was added to stop enzyme reactions. The samples were analyzed by reverse phase HPLC (C 18 column) with line gradient (0 to 100% acetonitile, with 0.05% trifluoroacetic acid) for 60 min. The sample of each peak was collected and analyzed by Synapt G2 HD mass spectrometer. The MS/MS data was analyzed by Mascot search.