Did the Mitochondrial Processing Peptidase Evolve from a Eubacterial Regulator of Gene Expression?

The recent sequencing of the genome of Rickettsia prowazekii, a eubacterium that is more closely related to mitochondria than to any other known prokaryote, has provided exciting new insights into the evolution of mitochondria and their genomes (Andersson et al. 1998 ). Consistent with the endosymbiont hypothesis (Gray et al. 1989 ; Gray 1993 ; Yang et al. 1985 ), many mitochondrial proteins appear to be functionally conserved in R. prowazekii and other eubacteria. An important exception concerns the proteins which make up the machinery for the import of mitochondrial proteins from the cytosol. In this study, we addressed the function of a eubacterial homolog of mitochondrial processing peptidases (MPPs), which remove targeting signals from proteins imported into the mitochondrial matrix compartment. Most mitochondrial proteins are encoded by nuclear DNA and synthesized in the cytosol. Hence, they have to be imported into mitochondria, for which purpose they are synthesized with an amino-terminal targeting (pre-)sequence. After translocation into the mitochondrial matrix, the presequence is removed by MPP (Brunner, Klaus, and Neupert 1994 ). MPP consists of two nonidentical but homologous subunits, designated α-MPP and β-MPP, which were identified in the mitochondria of several organisms (Braun and Schmitz 1995 ). Only the β subunit has catalytic activity (Arretz et al. 1994 ). All known MPP proteins belong to the so-called pitrilysin, or insulinase, family of endoproteases (Barret, Rawlings, and Woesner 1998 ). This family includes (1) Escherichia coli pitrilysin, a periplasmic protease involved in the degradation of small peptides; (2) insulinases from mammals and insects; (3) the PqqF protein from Klebsiella, involved in biosynthesis of the coenzyme PQQ; and (4) the N-arginine dibasic convertase from the rat. All catalytically active members of the pitrilysin family contain the motif His-x-x-Glu-His-x76-Glu, in which the two histidine residues and the C-terminal glutamate are required for the binding of zinc (Barret, Rawlings, and Woesner 1998 ). This motif is not conserved in α-MPP. An unrooted tree depicting possible evolutionary relationships between known members of the pitrilysin/insulinase family of endoproteases is shown in figure 1A. It has to be noted, however, that the phylogenetic analysis of this protease family is complicated by the fact that the most distantly related proteins belonging to this family share amino acid sequence similarity only in the regions containing their zinc-binding motifs (data not shown). Interestingly, three genes for proteins of the pitrilysin/insulinase family (i.e., YhwN, YmfH, and YmxG), containing the zinc-binding motif which is characteristic for active proteases, were identified by sequencing of the genome of the eubacterium Bacillus subtilis (Kunst et al. 1997 ). One of these proteins, YmxG, is highly similar to mitochondrial MPPs, sharing 51% identical residues or conservative replacements over its entire length of 409 residues with β-MPP of the potato (Chen et al. 1993 ; GenBank accession number U27560). Therefore, we renamed this MPP-like protein MlpA. Notably, MlpA of B. subtilis shares a high degree of similarity with MPP-like proteins from the parasitic eubacteria Mycobacterium leprae, Mycobacterium tuberculosis, and R. prowazekii (52%–59% identical residues or conservative replacements). The potential phylogenetic relationship between the known MPP and MlpA proteins is shown in figure 1B. Even though the latter 10 proteins show significant amino acid sequence similarity (data not shown) and a strongly supported relationship, it has to be noted that the amino acid composition of some of the corresponding sequences did not pass the 5%-χ2 test that we used to determine whether the amino acid composition of each sequence is identical to the average amino acid composition of the whole alignment. Therefore, the outcome of the present phylogenetic analysis must be regarded as a tentative result that awaits the availability of information on conserved structural features in MPP and MlpA proteins (e.g., homologous α helices and β sheets) for further improvement. In contrast to mycobacteria and R. prowazekii, B. subtilis is highly amenable to genetic analyses. Thus, the identification of the mlpA gene of B. subtilis offered the exciting possibility to investigate the function of the MlpA protein of this organism in particular, and the evolution of the function of MPP-like proteins in general. For this purpose, the mlpA gene of B. subtilis was disrupted with a kanamycin resistance marker, resulting in a truncation of the MlpA protein at residue 214. Using the mlpA mutant strain (ΔmlpA), it was shown that MlpA is not required for viability, growth, or sporulation. Unexpectedly, ΔmlpA cells showed about fivefold increased levels of proteolytic activity in their growth medium. As shown with specific inhibitors for serine proteases (PMSF) and metalloproteases (EDTA), this increase concerned the activities of both types of proteases (fig. 2A), several of which are secreted by B. subtilis (Wong 1995 ). In particular, the secretion of subtilisin (AprE), the major secreted serine protease of B. subtilis, was strongly stimulated, whereas the secretion of the neutral protease E (NprE), the major metalloprotease in the medium, was not affected (fig. 2B). Furthermore, neither the levels of α-amylase nor those of levansucrase in the medium of ΔmlpA cells were affected (not shown). This suggests that the disruption of mlpA stimulated the expression and/or secretion of a subset of proteins, including AprE and at least one unidentified metalloprotease. To test whether the expression of the aprE gene was stimulated, a transcriptional aprE-lacZ gene fusion was introduced (ectopically) in the chromosome of ΔmlpA cells and the parental strain. The resulting strains were grown in SSM medium for high production of AprE, and samples, withdrawn at hourly intervals, were assayed for β-galactosidase activity. As shown in figure 2C, in the postexponential growth phase, the activity of the aprE promoter was strongly increased in ΔmlpA cells, showing that the MlpA protein acts as a negative regulator of aprE gene expression. This effect was independent of DegU (data not shown), a key regulator among the nine known regulators of aprE transcription (Smith 1993 ). Interestingly, the expression of mlpA itself peaks at the onset of the transcription of the aprE gene (fig. 2D). Two possible modes of action of MlpA can be envisaged. First, MlpA may bind to the upstream sequences of the aprE gene, thereby acting as a repressor. We consider this possibility unlikely because MlpA lacks known DNA-binding motifs, such as a helix-turn-helix motif or the DEAD-box (see http://www.expasy.ch/sprot/prosite.html). Second, MlpA could act indirectly by modulating the activity of a transcriptional regulator of aprE. This could be achieved by activation of a repressor or inactivation of an activator. Because MlpA shows a high degree of similarity to proteases of the pitrilysin/insulinase family, particularly MPPs, we hypothesize that MlpA exerts its regulating effect on aprE expression through proteolysis. This hypothesis is plausible because proteolysis is an important theme in many regulatory pathways (Gottesman 1996 ). Our observation that MlpA is involved in gene regulation is particularly exciting, because it raises the intriguing possibility that the MPPs of mitochondria, which are essential for the biogenesis of mitochondria and, consequently, the life of the eukaryotic cell, have evolved from a eubacterial gene regulator. Antony Dean, Reviewing Editor Keywords: Bacillus subtilis, mitochondrial processing peptidase, gene regulation, subtilisin. Present address: VTT Biotechnology and Food Research, Espoo, Finland. Present address: Department of Pharmaceutical Biology, University of Groningen, Groningen, the Netherlands Address for correspondence and reprints: Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, the Netherlands. E-mail: j.m.van.dijl@farm.rug.nl. Fig. 1.—A, Phylogenetic analysis of the pitrilysin/insulinase family of proteases. First, a multiple alignment of amino acid sequences was made using CLUSTAL X, a Windows interface for the CLUSTAL W multiple-sequence alignment program (Thompson, Higgins, and Gibson 1994 ). For this purpose, the protein weight matrix PAM 350 was used, and a gap extension of 0.05 was defined. Furthermore, the gap separation distance was set to zero, and the end gap separation was switched on. For the subsequent analysis, multiple-substitution areas and hypervariable regions were excluded from the data set, resulting in a stretch of 844 conserved positions. Next, the data set was analyzed using the Blossom 62 model of evolution as implemented in PUZZLE, version 4.2 (Strimmer and von Haeseler 1996 ), because this method is most suitable for the analysis of distantly related proteins. Quartet puzzling values, which are indicated at the nodes of the resulting unrooted tree, were used to test the robustness of the tree topology. A Kishino-Hasegawa test did not reject trees with different topologies with regard to the relative positions of MlpA of Rickettsia prowazekii within the groups concerned in the polytomy. The GenBank ID numbers corresponding to the proteins used in the analysis are as follows: YwhN (Bacillus subtilis), g1565249; YmfH (B. subtilis), g2634058; SC9B10.04 (Streptomyces coelicolor), g2661690; Sll2009 (Synechocystis), g1652795; MlpA (MLCB22.26c; Mycobacterium leprae), g2342618; MlpA (MTV002.47c; Mycobacterium tuberculosis), g2624304; MlpA (YmxG; R. prowazekii), g2073473; β-MPP (Saccharomyces cerevisiae), g127290; β-MPP (Solanum tuberosum), g587564; β-MPP (Homo sapiens), g3115348; α-MPP (S. tuberosum), g266567; α-MPP (H. sapiens), g1709089; α-MPP (S. cerevisiae), g3889; BB0536 (Borrelia burgdorferi), g2688453; Ptr (pitrilysin; Escherichia coli), g882713; NRD (N-arginine dibasic convertase; Rattus norvegicus), g1352519; NRD (H. sapiens), g3914155; IDE (insulinase; H. sapiens), g124157; IDE (Drosophila melanogaster), g157168; PqqF (Pseudomonas fluorescens), g1709751; PqqF (Klebsiella pneumoniae). B, The phylogenetic analysis of known MPP and MlpA proteins was performed as described in A. A stretch of 579 conserved positions, remaining after the exclusion of autapomorphies and hypervariable regions, was used for the analysis. Fig. 2.—A, The mlpA gene of Bacillus subtilis 1012 (leuA8, metB5, r−M, m+M; Saito, Shibata, and Ando 1979 ) was disrupted with a kanamycin resistance marker, making use of a unique StuI site as previously described (Bolhuis et al. 1996 ). Cells of the resulting strain, denoted B. subtilis ΔmlpA, or cells of the parental strain (wt) were grown overnight in Schaeffer’s sporulation medium (SSM medium; Schaeffer, Millet, and Aubert 1965 ). Protease activity in the growth medium was visualized by spotting samples of 50 μl in wells on agarose–skim milk plates (1% agarose, 1% skim milk, 50 mM Tris-HCl [pH 8], and 4 mM CaCl2) and incubating at 37°C for 1 h. The halo size reflects the level of proteolytic activity in the medium. Protease activities (expressed as absorbance changes at 440 nm/h) in the absence (no addition, NA), or presence of 1 mM phenylmethanesulphonyl fluoride (PMSF) or 10 mM EDTA were determined with azocasein (Sigma). For this purpose, 250 μl growth medium was mixed with 150 μl 2% azocasein suspension in 50 mM Tris-HCl (pH 7.5) and 4 mM CaCl2, and after 60 min at 25°C, the reaction was stopped by addition of 1.2 ml 10% TCA. Next, the samples were centrifuged, and absorbance changes (440 nm) of the supernatant were determined. B, Western blotting analysis of the secretion of subtilisin (AprE) and neutral protease (NprE) by ΔmlpA cells. Cells were grown overnight in SSM medium. AprE and NprE in the growth medium were detected by SDS-polyacrylamide gel electrophoresis, blotting, and immunodetection with specific antibodies, as described priviously (Bolhuis et al. 1996 ). C, To analyze aprE transcription, the transcriptional aprE-lacZ gene fusion of B. subtilis BG4057 (Henner et al. 1988 ) was introduced in B. subtilis 1012 and ΔmlpA. Cells of the resulting strains, denoted B. subtilis 1012 aprE-lacZ () and ΔmlpA aprE-lacZ (▾) were grown in SSM medium, and samples were taken at hourly intervals for optical density (OD) readings at 600 nm and β-galactosidase activity determinations (Bolhuis et al. 1996 ). Zero time (t = 0) indicates the transition point between the exponential and the postexponential growth phases. D, To analyze mlpA transcription, a transcriptional mlpA-lacZ fusion was introduced in B. subtilis 1012. For this purpose, a 567-bp fragment containing the ribosome-binding site and start codon of the mlpA gene was amplified by PCR with the primers AB04mpp (aagaattcCGAGCAGCTCGACAAGAC; nucleotides identical to genomic template DNA are in capital letters, and restriction sites used for cloning are underlined) and AB05mpp (ttggatccGATTCTGCTATCTCGCGT). The amplified fragment was cloned in front of the promoterless lacZ gene of plasmid pLGW300 (van Sinderen et al. 1990 ), and the resulting plasmid (pLGM301) was integrated by single-crossover recombination into the mlpA gene on the chromosome of B. subtilis 1012. The resulting strain was denoted B. subtilis 1012::pLGM301. Cells of this strain were used to analyze transcription of mlpA-lacZ as described in C. We thank Drs. L. Hamoen, H. Tjalsma, J. Jongbloed, and M. van Roosmalen for valuable discussions, Dr. E. Ferrari for providing B. subtilis BG4057, and Dr. H. Paulus for providing a plasmid containing the 3′ end of mlpA. A.B. was supported by European Union (EU) Biotechnology Grant Bio2-CT93-0254, and S.B and J.M.v.D were supported by EU Biotechnology Grants Bio2-CT93-0254 and Bio4-CT96-0097.