Introduction

The peptidyl-prolyl cis-trans isomerases (PPIases or immunophilins) are enzymes that catalyze the isomerization of prolyl peptide bonds in proteins. PPIases conform a protein family with universal distribution, and can be classified into three structurally and biochemically highly divergent subfamilies: cyclophilins, FK506 binding proteins (FKBPs), and parvulins (for reviews see Göthel and Marahiel 1999; Ivery 2000). Specific inhibitors are capable of differentiating between each subfamily: cyclophilins are inhibited by cyclosporin A (CsA) and sanglifehrin (Zhang and Liu 2001); FKBPs are inhibited by FK506, and parvulins are inhibited by juglone (Göthel and Marahiel 1999). There is another unusual member of the family, known as the trigger factor, which has a protein domain similar to FKBPs but is not inhibited by FK506. Some authors (Stoller et al. 1995) have proposed classifying trigger factors as a new subfamily of PPIases.

Cyclophilins are abundant and ubiquitous proteins that are present in almost all living organisms analyzed to date, except in bacteria such as Mycoplasma geniculatum (Bang et al. 2000) and some Archaea (Maruyama and Furutani 2000). Cyclophilins are well characterized in many eukaryotes, where they show remarkable diversity, with up to 16 members described in human and 29 in Arabidopsis (Wang and Heitman 2005). In contrast, detailed characterization of these enzymes in prokaryotes is limited to few cases including Escherichia coli (Liu and Walsh 1990; Liu et al. 1991; Hayano et al. 1991; Compton et al. 1992; Clubb et al. 1994; Norregaard-Madsen et al. 1994), Bacillus subtilis (Herrler et al. 1992, 1994; Achenbach et al. 1997; Göthel et al. 1998), Streptomyces chrysomallus (Pahl et al. 1992, 1997), Acinetobacter calcoaceticus (Kok et al. 1994), Legionella pneumophila (Schmidt et al. 1996), Erwinia chrysanthemi (Pissavin and Hugouvieux-Cotte-Pattat 1997), Halobacterium cutirubrum (Nagashima et al. 1994), Streptomyces antibioticus ATCC11891 (Manteca et al. 2004), and Mycobacterium tuberculosis (Henriksson et al. 2004). The most obvious biochemical difference between bacterial and eukaryotic cyclophilins is a weaker binding of CsA to the former. The cyclophilins from Gram-negative bacteria are particularly resistant to inhibition by CsA (Schönbrunner et al. 1991; Manteca et al. 2004). Although the biological significance of this resistance to CsA inhibition remains unclear, some authors (Schönbrunner et al. 1991; Schmidt et al. 1996; Pissavin and Hugouvieux-Cotte-Pattat 1997) have suggested differentiating Gram-negative bacteria cyclophilins as a separate group on the basis of this feature.

The subclass Actinobacteridae is a subdivision of the phylum and class Actinobacteria. These are aerobic Gram-positive bacteria with a high G+C content that can be classified into three orders: Actinomycetales, Bifidobacteriales, and unclassified actinomycetes (Cole et al. 2005). Members of the Actinomycetales are mainly found in the soil, where they play important roles in decomposition and humus formation. However, a few of them, such as Mycobacterium, Corynebacterium, and Streptomyces, may be pathogens (McNeil and Brown 1994; Martin et al. 2004). Some, as in the case of Streptomyces, form branching filaments and present a “multicellular” behavior pattern with complex life cycles, in which cell death processes are developmental landmarks (Nicieza et al. 1999; Manteca et al. 2006). These features together with the production of vegetative and reproductive structures have made Streptomyces valuable in analyzing prokaryotic differentiation (Chater and Horinouchi 2003). Moreover, the origin of some of the domains characteristic of the apoptotic proteins of eukaryotes can be traced back in evolution to Streptomyces (Yarmolinsky 1995; Hochman 1997; Aravind et al. 1999; Koonin and Aravind 2002a). All these aspects, plus the capability to produce many commercial antibiotics (Baltz 1998), have motivated the extensive investigation of the genetic basis of Streptomyces developmental processes (Chater 2001; Paradkar et al. 2003). The availability of the two complete genome sequences of Streptomyces coelicolor A3(2), Streptomyces avermitilis (Bentley et al. 2002; Ikeda et al. 2003), and the partially sequenced genome of Streptomyces scabies (The Sanger Institute; http://www.sanger.ac.uk/Projects/S_scabies/) has greatly facilitated these studies.

The number and distribution of cyclophilins within the class Actinobacteria have not yet been studied in detail. A cyclophilin from Mycobacterium tuberculosis has recently been crystallized (Henriksson et al. 2004). Two cyclophilins have been analyzed in depth in Streptomyces chrysomallus. One, named ScCypA, has amino acid sequence and biochemical characteristics similar to those of eukaryotic cyclophilins and was shown by hybridization experiments to be present also in the genome of Streptomyces lividans, S. coelicolor, Streptomyces griseus, and Streptomyces parvulus (Pahl et al. 1992). The other, ScCypB, was suggested to have a specialized function in the cell (Pahl et al. 1997). In a previous study (Manteca et al. 2004), we characterized a cyclophilin of Streptomyces antibioticus that showed higher sequence similarity to cyclophilins from Gram-negative bacteria than to other cyclophilins of Streptomyces or other Gram-positive bacteria, being the first and only example described so far in the latter.

In this work, we performed a comprehensive evolutionary study of Actinobacteria cyclophilins. Phylogenetic relationships of Actinobacteria cyclophilins were reconstructed in order to establish orthologous and paralogous relationships within this protein subfamily. The study provides a robust framework to interpret evolution of the distinct localizations and functions of the different paralogues, as well as to identify those derived amino acid residues that confer each protein its unique properties and responses to inhibitors. Furthermore, the phylogeny also allowed us detection of horizontal gene transfer (HGT) events.

In addition, and to complement the phylogenetic analysis, Southern and Western blot experiments were carried out with some of the most relevant cyclophilin genes/proteins as probes in order to determine the taxonomic distribution of the different cyclophilins in Actinobacteria. The studies show the presence of a remarkable diversity of cyclophilins within the genus Streptomyces, including novel, previously unreported cyclophilins located in the membrane.

Materials and Methods

Bacterial Strains, Growth Conditions, and Plasmids

The Streptomyces strains used in this study were S. antibioticus ATCC11891, S. antibioticus ATCC10382, S. antibioticus ETH7451, S. griseus ATCC10137, S. griseus B-2682, S. coelicolor A3(2), S. chrysomallus ATCC3338, S. lividans 1326, S. albus G (Chater and Wilde 1976), S. spectabilis ATCC27465, S. achromogenes ATCC12767, S. griseocarneus ATCC 12628, S. avermitilis ATCC 31267, S. aureofaciens B96 (Dr. A. Godany, Institute of Molecular Biology, Slovak Academy of Science, Bratislava), and S. glaucescens ETHZ22794. E. coli XL1-Blue was used as host for the preparation of plasmid DNA. E. coli C600 and Pseudomonas aeruginosa PAO1 were used as controls in the Southern and Western blot experiments. The Streptomyces species were cultured on solid or in liquid GAE medium (glucose, asparagine, yeast extract, and salts) (Méndez et al. 1985) and YEME medium (Hopwood et al. 1985) at 30°C and 200 rpm in the case of submerged conditions. In solid medium, plates were covered with sterile cellophane disks before incubation. Plates and liquid cultures were directly inoculated with a spore-dense suspension. E. coli strains were grown in Luria broth medium with specific antibiotics. The pGEM-T plasmid (Promega) was used to clone PCR products and to sequence them.

Chemicals and Enzymes

Restriction enzymes, T4 DNA-ligase, Klenow DNA polymerase, synthetic oligonucleotides, and nylon filters were acquired from Amersham Pharmacia Biotech. Ampicillin was obtained from Sigma and used at a concentration of 100 μg/ml. Thermostable DNA polymerase (DyNAzyme) was from Finnzymes. Peroxidase chemiluminescence blotting substrate and the digoxigenine DNA labeling kit were supplied by Roche. PVDF (Polyvinylidene Difluoride) membranes were from Millipore.

PCR Amplification of Streptomyces cyclophilins

Degenerated oligonucleotides (Table 1) designed from highly conserved sequences of cyclophilins from Streptomyces (Scoe-B—S. coelicolor cyclophilin B accession CAC42137M; Scoe-A—S. coelicolor cyclophilin A, accession CAB45223; Schr-B—S. chrysomallus cyclophilin B, accession Q06118; Schr-A—S. chrysomallus cyclophilin A, accession P77949; and Sant-C—S. antibioticus ATCC11891 cyclophilin, accession P83221) and other organisms (EcoA: E. coli cyclophilin A, accession P20752) were used to amplify cyclophilins from different species belonging to the genus Streptomyces.

Table 1 Oligonucleotide sequences used in this work
Full size table

Recombinant DNA Techniques

Chromosomal and plasmid DNAs were isolated using standard procedures (Sambrook et al. 1989). Separated DNA fragments were isolated from agarose gels using the QUIAEX IIR Gel Extraction System (Qiagen) and ligated into pGEM-T plasmids (Promega). E. coli cells were transformed using the CaCl2 method. Sequencing was performed with an ABI Prism 310 Genetic Analyzer following the manufacturer’s instructions and using M13 universal primers. Sequences were processed and analyzed using GCG software (Genetics Computer Group) (Devereux et al. 1984).

Phylogenetic Analysis

Databases at the National Center for Biological Information (http://www.ncbi.nlm.nih.gov) were screened for cyclophilins using the BLAST search program (Altschul et al. 1997). In addition, some cyclophilins were sequenced anew in this work (detailed under Results; see Table 2). The partial sequences of S. scabies cyclophilins were obtained from The Sanger Institute (http://www.sanger.ac.uk/Projects/S_scabies/). Only amino acid sequences of cyclophilins were used as queries in order to prevent any bias that could be introduced by the broad range of GC contents and codon preferences in the different taxa.

Table 2 Cyclophilin sequences used in the phylogenetic analysis ordered by taxonomical classification
Full size table

A total of 97 complete or almost-complete cyclophilin proteins were retrieved (the partial sequence of cyclophilin B from S. scabies was not included because it is relatively short). The sequences used and their abbreviations are listed in Table 2. Sequences were aligned using the MUSCLE software version 3.6 (Edgar 2004) and multiple alignments were subsequently refined by eye with MacClade 4.05 (Maddison and Maddison 1992). Ambiguous alignments in highly variable (gap-rich) regions were excluded from the phylogenetic analyses (aligned sequences and exclusion sets are available from the authors upon request). We used PROTTEST version 1.2.6 (Abascal et al. 2005) to select the substitution model that best fit the empirical data set (WAG+I+Γ; α=1.45, I=0.01) and PHYML version 2.4.4 (Guindon and Gascuel 2003)to find the maximum likelihood (ML) tree. The robustness of the inferred tree was assessed with bootstrapping (BP; 500 pseudoreplicates) using PHYML. Bayesian inferences were conducted using MrBayes v3.1 (Huelsenbeck and Ronquist 2001) by Metropolis coupled Markov Chain Monte Carlo (MCMCMC) sampling for 2 × 106 generations (two runs in parallel; four simultaneous MC chains; sample frequency, 100) under the WAG+I+Γ model. Based on the number of generations needed to reach convergence, burn-in was set to 450 trees. Robustness of the inferred trees was evaluated using Bayesian posterior probabilities (BPPs).

The recovered ML tree was used as framework to identify those protein residues that are shared derived by the members of each of the cyclophilin main paralogues. Each of the protein residues was mapped onto the phylogeny, and the ancestral character states and shared-derived amino acid residues were inferred with parsimony using PAUP* v4.0b10 (Swofford 2002) and MacClade v4.0. Inferred shared-derived amino acids identified in each paralogue are not necessarily highly conserved across all members of the paralogue. Hence, only those shared-derived characters that had a consistency index (as a measure of the fit of each character to the tree) above 0.5 were considered.

Analysis of Cyclophilin Gene Distribution in the Genus Streptomyces

Southern blot analyses were performed on the chromosomal DNA from different Streptomyces species. All chromosomal DNAs were digested with BamHI and the resulting fragments were run in a 0.8% agarose gel. Four cyclophilin genes were used as probes: Scoe-A, Scoe-B, Sant-C, and Scoe/liv-M (see Table 2). The genes were amplified by PCR with the primers listed in Table 1 (Scoe-A forward-reverse, Scoe-B forward-reverse, Scoe/liv-M forward-reverse, and Sant-C forward-reverse), labeled with digoxigenine, and used as probes. Southern blots were performed under high-stringency conditions, as described by Hopwood et al. (1985), so as that strong signals should indicate a very high identity with the probe.

Analysis of the Presence of Proteobacterium-like Cyclophilins in the Genus Streptomyces

The Streptomyces strains were grown on the surface of cellophane disks. Young vegetative cultures were scraped out with a plain spatula, resuspended in buffer A (20 mM Tris-HCl, pH 8, 1 mM EDTA, 7 mM ß-mercaptoetanol, and 0.5 mM PMSF), and ruptured in an MSE soniprep 150, in four cycles of 10 sec, on ice. The ratio between fresh weight of mycelium and buffer A volume was the same in all samples (64 mg mycelium/ml buffer A). After centrifuging at 10,000 rpm in an Eppendorf microcentrifuge for 30 min at 4°C, the supernatants were used for Western blot analysis. The anti-Sant-C antibodies used in these experiments had been previously obtained in our laboratory (Manteca et al. 2004). Immunoblotting was carried out after protein SDS-PAGE and transfer to PVDF membranes (Immobilon-P; Millipore). Antibody reaction was allowed to take place for 1 hr at room temperature with a serum dilution of 1:1000. Immunodetection was performed with anti-rabbit peroxidase IgG (Sigma) using a peroxidase chemiluminescence reaction (BM Chemiluminescence Blotting Substrate- POD; Roche).

Protein Analysis

Proteins were analyzed by SDS-PAGE in 12% poliacrylamide gels (Laemmli 1970) and stained with Coomassie blue. The molecular markers used were from Bio-Rad (low range). Protein concentrations were determined by Lowry assay (Lowry et al. 1951), using bovine serum albumin as standard (Sigma).

Three-Dimensional Protein Modeling

The modeled three-dimensional structure of Scoe/liv-M was calculated on the basis of its amino acid sequence with the Internet-based protein structure homology modeling server SWISS-MODEL (http://swissmodel.expasy.org). The ‘‘first approach” mode was used with default settings (Guex and Peitsch 1997; Guex et al. 1999; Schwede et al. 2003). The model obtained was visualized using the SWISS-Pdb-Viewer program.

Results and Discussion

Streptomyces Cyclophilins: Gene Cloning and Analysis

An alignment of all known Streptomyces cyclophilins, together with one human and one E. coli cyclophilins is shown in Fig. 1. Three cyclophilin genes from Streptomyces were obtained anew in the present work using two degenerated oligonucleotides, corresponding to the conserved amino acid motif SAGRIV that is present in the amino-terminal end of the Sant-C cyclophilin (Manteca et al. 2004) and to several highly conserved amino acid sequences found at the carboxy end of the protein, respectively (see Materials and Methods and Table 1). The sequences obtained lack about 10–15 amino acids from both ends, as the oligonucleotides used were not an exact match to the upstream and downstream ends of the genes (Fig. 1). Nonetheless, all the amino acid residues relevant for PPIase activity were sequenced in all of them (see arrows in Fig. 1). The new sequenced cyclophilin genes are Sant-B from S. antibioticus ATCC11891 (obtained by using SAGRIV-VVVETR oligonucleotides), Sant7451-B from Streptomyces antibioticus ETH7451 (obtained with SAGRIV-AESGAL oligonucleotides), and Sachr-C from S. achromogenes (obtained with Sant-C forward and Sachr-C reverse oligonucleotides) (Tables 1 and 2).

Fig. 1
figure 1

Amino acid sequence alignment of all known Streptomyces cyclophilins, human (Hsap) cyclophilin and a proteobacterium (EcoA) cyclophilin. The Streptomyces sequences included are Sscab-A, Save-A, Scoe-A, Schr-A, Sgri-B, Scoe-B, Schr-B, Sant-B, Sant7451-B, Sant-C, Sachr-C, Save-M, Scoe/liv-M, Sscab-M, and Scoe/liv-D. The accession numbers of these proteins are given in Table 2. Sscab-A and -M are incomplete sequences obtained directly from the Sanger Institute (http://www.sanger.ac.uk/Projects/S_scabies). Amino acid residues involved in CsA (arrows above the alignment) and substrate (arrows below the alignment) binding of HsA (Janowski et al. 1997; Göthel and Marahiel 1999; Fanghanel and Fischer 2003) are indicated. Shared derived amino acids of each paralogue are boxed (see Fig. 2 and Table 3). An asterisk indicates those residues conserved in all cyclophilins. The putative membrane domain of M cyclophilins is indicated by a bracket.

Full size image
Table 3 Shared derived characters of the different paralogues inferred using parsimony
Full size table
Fig. 2
figure 2

Phylogenetic relationships of all known Actinobacteria cyclophilins, representative cyclophilins of other bacteria, and some eukarya. The recovered ML phylogram is shown. Numbers above and below nodes represent ML bootstrap (>50%) and Bayesian posterior probabilities (>90%), respectively. The main cyclophilin paralogues are shown. Solid lines indicate four main cyclophilin paralogues that include Actinobacteria representatives. Dashed lines delimitate two further groups, one of them represented by Cyanobacteria and the other a heterogeneous group (see text). Sequence designations and sources are given in Table 2.

Full size image

In a previous study, we already cloned and sequenced one cyclophilin of S. antibioticus ATCC11891 (SanCyp18 [Manteca et al. 2004]) that showed higher sequence similarity to cyclophilins from Gram-negative bacteria such as Ralstonia solanacearum (66% identity) or E. coli (59.1% identity) than to other reported Streptomyces cyclophilins (38.8% identity with respect to S. coelicolor Scoe-A, 35.2% with respect to S. avermitilis Save-A, 31% identity with respect to S. chrysomallus cyclophilin A). In addition, we retrieved from NCBI databases another 12 cyclophilins of Streptomyces (Table 1). In order to simplify the nomenclature of the cyclophilin groups described in this work (Figs. 1 and 2), the S. antibioticus cyclophilin SanCyp18 (Manteca et al. 2004) was renamed Sant-C. Moreover, S. chrysomallus cyclophilin ScCypB (Pahl et al. 1997) was renamed Schr-A in the present study, because cyclophilins closely related to this protein are present in all Streptomyces and in all Actinobacteria (see Figs. 1–3). Conversely, the SchrA cyclophilin (similar to eukaryotic cyclophilins [Pahl et al. 1992]) was renamed Schr-B, a less extended group within Streptomyces (see Figs. 1–3).

Fig. 3
figure 3

Southern blot hybridization of BamHI-digested chromosomal DNA from various Streptomyces species, with probes of representative cyclophilins from the four Streptomyces cyclophilin groups. Each lane contains 10 μg of total chromosomal DNA. (1) Escherichia coli C600; (2) S. antibioticus ATCC11891; (3) S. antibioticus ATCC10382; (4) S. antibioticus ETH7451; (5) S. griseus B-2682; (6) S. coelicolor A3(2); (7) S. chrysomallus; (8) S. lividans; (9) S. albus; (10) S. spectabilis; (11) S. achromogenes; (12) S. griseocarneus; (13) Pseudomonas aeruginosa PA01. λ, phage λ DNA digested with PstI; fragment sizes are indicated (kb) at the left. (a) Scoe-A probe; (b) Scoe-B probe; (c) Scoe/liv-M probe; (d) SantC probe. The stringency of the final wash was 0.5× SSC, 65°C.

Full size image

Phylogenetic Relationships of Actinobacteria Cyclophilins

Phylogenetic relationships among cyclophilins from the Class Actinobacteria (Streptomyces, Mycobacterium, Corynebacterium, Saccharopolyspora, Bifidobacterium, Nocardia, Propionibacterium, and Frankia) were reconstructed within the context of their broader distribution and homology with other prokaryotic and eukaryotic cyclophilins (Table 2 and Fig. 2).

Although cyclophilins are typically about 170 amino acids in length, phylogenetic analyses were based on a sequence data set that initially included 594 positions because the alignment of the 97 analyzed proteins required the postulation of numerous gaps. A total of 469 gap-rich positions were excluded due to ambiguity in the positional homology assignment. Of the remaining 125 positions, only 2 were invariant (a proline and a phenylalanine, at positions 30 and 112 of Hsap, respectively) and 119 were parsimony-informative. The ML tree that was recovered based on this sequence data set is shown in Fig. 2. Up to six main paralogue groups of cyclophilins are clearly identified based on the reconstructed phylogeny. These groups show high statistical support, but phylogenetic relationships among them remain unresolved (Fig. 2). Four of the groups include Actinobacteria and were named using correlative letters (A, B, C, and M) according to their abundance in the class Actinobacteria.

Group A corresponds to highly conserved cytosolic cyclophilins (about 80% amino acid identity), and it is present in all analyzed Streptomyces (five species), as well as in other Actinobacteria (Bifidobacterium, one species; Mycobacterium, three species, Corynebacterium, two species; Brevibacterium, one species; Propionibacterium, one species; and Frankia, one species). All Actinobacteria analyzed in detail contain at least one orthologue of this cyclophilin. Within this group, there is a S. coelicolor/S. lividans cyclophilin, named Scoe/liv-D, which shows an extremely long branch (Figs. 1 and 2). Analysis of the Scoe/liv-D sequence reveals that some amino acids critical for the PPIase activity (arrows in Fig. 1) are mutated. These data and the fact that S. coelicolor (and S. lividans) harbors another orthologue of group A cyclophilins (Fig. 2) strongly suggest that the Scoe/liv-D gene encodes a mutated protein that either is void of biological function or has been recruited for a totally different function. The widespread distribution of group A cyclophilins in Actinobacteria and their highly conserved sequence suggest an important function for these proteins in these bacteria. Yet their precise biological roles remain unknown.

Group B is comprised of both bacterial and eukaryotic cyclophilins. Three bacterial genera are included in this group: Streptomyces, with four species, and Synechocystis (Phylum Cyanobacteria) and Rhodopirellula baltica SH1 (Phylum Aquificae, Order Aquificales), with one species each (Fig. 2). Streptomyces is thus the only bacterial genus in which these cyclophilins are distributed in several species. The wide taxonomic distribution of this paralogue could suggest that all its members are true orthologues that share a common ancestor. This scenario, however, would require loss of the orthologue in many other species where is not found. Alternatively, it has been suggested that early HGT between primordial eukaryotic cells and Streptomyces could be responsible for the presence of Schr-B cyclophilin in S. chrysomallus (reported as SchrA in that study [Pahl et al. 1992]). This process is reminiscent of the one proposed for the acquisition of putative eukaryotic-type protein phosphatases (PPs) and other sensory and regulatory activities in bacteria with complex life cycles, such as Streptomyces and Cyanobacteria (Koonin and Aravind 2002a; Petrickova and Petricek 2003; Shi and Zhang 2004). Interestingly, Synechocystis, a genus of the phylum and class Cyanobacteria, also harbors this type of cyclophilins (Fig. 2). Another possible explanation for the common presence of these cyclophilins in Streptomyces and eukaryotes would be early HGT from Actinomycetes to ancestral eukaryotic cells, in the same way as proposed for the acquisition of protein domains involved in the signaling of apoptotic processes in eukaryotic cells: eukaryote-type protein kinases of the PNK2 family, apoptotic Toll-interleukin-receptor domain (TIR) adaptors, caspase-like proteases, and apoptotic (AP-) ATPases or NTPases (Leonard et al. 1998; Koonin and Aravind 2002a; Petrickova and Petricek 2003). Rhodopirellula baltica, a representative of the phylum Plantomycetes, also harbors cyclophilins belonging to group B. The Planctomycetes form an independent and deep-branching bacterial phylum. They lack the otherwise universal bacterial cell wall polymer peptidoglycan and present an internal cytoplasmic compartment defined by a single intracytoplasmic membrane that holds a condensed fibrillar nucleoid as well as ribosome-like particles (Lindsay et al. 2001). These structures, together with the unusual fatty acid composition of the phospholipids, are more similar to eukaryotes than to any other representative of the bacterial domain (Kerger et al. 1988). Accordingly, they might represent one of the ancient groups from (or to) which the “B” cyclophilins could have been transferred. Overall, the interesting taxonomic distribution of group B makes these cyclophilins worthy of more in-depth study in the future.

A third group (named “M” after membrane) consists of cyclophilins presenting a putative membrane domain (Fig. 1) and, hence, a higher molecular mass (30–34 kDa; an exception being the Saccharopolyspora erythraea Sery-M cyclophilin, with a molecular mass of 17 kDa). The M paralogue group is exclusively present in species belonging to the order Actinomycetales: Streptomyces (four species), Mycobacterium (four species), Corynebacterium (three species), Saccharopolyspora (one species), Nocardioides (one species), and Propionibacterium (one species) (Fig. 2). All the Actinomycetales whose genome has been fully sequenced have one single cyclophilin from this group (see below). They are the only membrane cyclophilins described to date in bacteria. Nonorthologue membrane cyclophilins have been described in eukaryotic cells (mitochondrial cyclophilin F [Clark et al. 2003], the NK-tumor recognition protein [Anderson et al. 1993], NinA [Schneuwly et al. 1989]).

Group C is mostly formed by Proteobacteria-phylum cyclophilins. In addition, the Gram-positive S. antibioticus ATCC11891, S. achromogenes, and Nocardia farcinica also harbor cyclophilins from this group (Fig. 2), which suggests HGT of these genes from a Gram-negative bacterium. In fact, the cyclophilins from Gram-negative bacteria that grouped together with Streptomyces and Nocardia cyclophilins in the recovered tree were those corresponding to soil bacteria that share their habitat with Streptomyces/Nocardia and, therefore, could have transmitted genetic information to the latter. Enzyme activity properties of Gram-positive bacteria cyclophilins from group C further support their origin through HGT. For instance, the Sant-C cyclophilin shows an inhibition constant for CsA in the micromolar range (K i, 21 μM), which is a typical value for cyclophilins of Gram-negative bacteria, and significantly different from those of Gram-positive bacteria cyclophilins, which are in the nanomolar range (Manteca et al. 2004). The presence of cyclophilins from groups B and C in Streptomyces confirms previous evidence (Wiener et al. 1998; Ueda et al. 1999; Egan et al. 2001; Metsä-Ketelä et al. 2002; Garcia-Vallve et al. 2003; Nishio et al. 2004; Kawase et al. 2004; Manteca et al. 2004) on the relevance of HGT in the acquisition of foreign genes by this bacterium.

The phylogenetic analysis carried out with the prokaryotic cyclophilins also uncovered another two cyclophilin groups (marked with dashed lines in Fig. 2), one of them (D) comprising exclusively Cyanobacteria and the other (E) including representatives of phylogenetically diverse groups (Archaea, Proteobacteria, Cyanobacteria, and Firmicutes). As mentioned above, the wide taxonomic distribution of this paralogue may indicate that HGT is a common evolutionary mechanism of acquisition of cyclophilins.

From the phylogenetic analysis, we can conclude that the order Actinomycetales (and especially the genus Streptomyces) has an unusual diversity of cyclophilins. The putative membrane cyclophilins (group M) are exclusive of the Actinomycetales order; the Proteobacteria-like cyclophilins are not present in any Gram-positive organism sequenced to date, with the exception of some Streptomyces species and Nocardia farcinica; finally, eukaryotic-like B cyclophilins are found exclusively in several Streptomyces, Rhodopirellula baltica, and Synechocystis. The specific functions of the different cyclophilin paralogues described in this study are largely unknown. In order to further understand the functional diversification of cyclophilins, site-directed mutagenesis experiments on the different paralogues are needed. In this regard, we performed additional evolutionary analyses in trying to determine which residues may be responsible for the potentially different (novel) functions of each paralogue. Shared-derived amino acid residues that are specific to each paralogue were inferred by ancestral character reconstruction using parsimony based on the ML tree (Fig. 1 and Table 3). Under the parsimony criterion, groups A, B, C, D, E, and M are defined by 6, 18, 18, 10 (or 16 if Pma6 is not included in group D), 11, and 15 synapomorphies each, respectively (not shown). However, there is considerable variation of these shared-derived amino acid residues within each paralogue as indicated by the estimated low consistency indexes (CIs). It is likely that the shared-derived residues that may confer each cyclophilin its peculiar properties should be among those with a higher degree of conservation, i.e., with a higher CI (Table 3). It is also important to note here that phylogenetic relationships within each paralogue are generally poorly resolved (low bootstrap and posterior probability support) and, hence, that ancestral character state reconstruction should be interpreted with caution.

Southern and Western Blot Analysis of the Taxonomic Distribution of Cyclophilins in the Genus Streptomyces

The above analysis of the diversity of Streptomyces cyclophilins was completed by Southern blot experiments. Representative genes encoding cyclophilins pertaining to the four groups A, B, C, and M (Scoe-A, Scoe-B, Sant-C, and Scoe/liv-M; see above and Figs. 1 and 2) were used as probes. Hybridizations were performed under highly stringent conditions, which select for genes with high identities (at least 70% [Hopwood et al. 1985]). The results are presented in Fig. 3. Southern blot analysis performed with a Scoe-A probe showed positive results in the 11 Streptomyces species analyzed (Fig. 3a). This indicates that all the Streptomyces analyzed harbor one cyclophilin from group A, as evidenced in the Streptomyces species whose genome has been sequenced, and that these cyclophilins are highly conserved (about 80% identity between the known sequences; see above and Figs. 1–3).

When the Scoe-B gene was used as the probe, four positive signals were detected in S. antibioticus ATCC10382, S. antibioticus ETHZ7451, S. lividans 1326, and S. coelicolor A3(2) (Fig. 3b). These bands are different from those that tested positive reaction with the Scoe-A probe (compare molecular weights in Figs. 3a and b). Obviously, this does not exclude the presence of cyclophilins from the eukaryotic-like group, with an identity lower than 70% to the assayed Scoe-B cyclophilin in other Streptomyces. In fact, this does occur with S. chrysomallus cyclophilin Schr-B (Figs. 1–3), which does not show any signal in the Southern blot under the conditions described.

The Southern blot analysis performed with a Scoe/liv-M probe showed the presence of putative orthologous genes in 8 of the 11 Streptomyces analyzed: S. antibioticus ATCC11891, S. antibioticus ATCC10382, S. antibioticus ETH7451, S. griseus B-2682, S. coelicolor A3(2), S. lividans 1326, S. achromogenes, and S. griseocarneus (Fig. 3c). It is conceivable that the three species that did not give a signal (S. chrysomallus, S. albus, and S. spectabilis) may nonetheless have a somewhat less similar membrane cyclophilin, in which case the presence of the membrane cyclophilins might be considered to be a general characteristic within Streptomyces.

As expected, the Southern blot analysis performed using the “C” cyclophilin Sant-C probe showed a strong signal in the two species harboring this cyclophilin (S. antibioticus ATCC11891 and S. achromogenes; Fig. 3d). As in the case of the Scoe-A probe described above, this does not exclude the presence of proteobacteria-like cyclophilins with a lower identity to the assayed Sant-C(18) cyclophilin in other Streptomyces. This is clearly exemplified by the E. coli C600 and Pseudomonas aeruginosa PA01 cyclophilins used as controls, which harbor cyclophilins with an identity percentage of about 60% with Sant-C, which nonetheless did not give a positive signal in the Southern experiment (Fig. 3d).

Variation in the stringency conditions provoked cross hybridization between CypA and CypB probes, as well as between CypC and CypM, albeit to a lesser extent (not shown). Despite the fact that the Southern blot technique showed some limitations under the conditions assayed in the present work, the results obtained were consistent (as indicated by the fact that different genes hybridize with the different probes; compare molecular weights in Fig. 3). Furthermore, the positive and negative controls (sequenced Streptomyces cyclophilins and Gram-negative bacteria) included in the experiments provided additional support.

Further data on the distribution of cyclophilins from group C within the genus Streptomyces were obtained by Western blot analysis with the antibodies obtained against Sant-C protein (Manteca et al. 2004). Cellular extracts used for the test were obtained from solid cultures of different Streptomyces species in the substrate mycelium phase, as detailed under Materials and Methods. The results (Fig. 4) showed that some of the species analyzed contained proteins with a molecular mass of about 18 kDa which present a strong reaction against the Sant-C antibodies. These species were S. achromogenes, S. griseus ATCC10137, S. griseus B-2682, S. antibioticus ETH7451, S. antibioticus ATCC10382, and S. antibioticus ATCC11891. S. antibioticus ATCC11891 and S. achromogenes were previously shown to harbor a cyclophilin gene of the proteobacteria type C cluster (Table 2 and Figs. 1–3); these results correlate with the clear signal shown in the Western blot analysis. This signal is equivalent to that shown by the E. coli C600 and Pseudomonas aeruginosa PA01 samples (Fig. 4). In contrast, S. coelicolor, S. avermitilis, and S. chrysomallus, which harbor cytosolic cyclophilins from the Actinobacteria class group (A; Figs. 1 and 2), did not react with Sant-C antibodies. Interestingly, E. coli cell-free extracts gave two signals that conceivably correspond to the previously described periplasmic and cytosolic cyclophilins, respectively (Hayano et al. 1991). All these data strongly indicate that, under the experimental conditions employed, the Sant-C antibodies are highly selective for cyclophilins from the proteobacteria type C cluster. The results are consistent with the phylogenetic analysis data, which indicate that, to date, Nocardia farcinica and the Streptomyces species included in the C cluster are the only Gram-positive bacteria harboring cyclophilins normally encountered in Gram-negative bacteria.

Fig. 4
figure 4

Western blot analysis of different Streptomyces extracts with the Sant-C polyclonal antibodies (see Materials and Methods). M, marker (1 μg of purified recombinant Sant-C [Manteca et al. 2004]); (1) S. chrysomallus; (2) S. achromogenes; (3) S. griseus ATCC10137; (4) S. avermitilis; (5) S. aureofaciens; (6) S. coelicolor A3(2); (7) S. lividans; (8) S. glaucescens; (9) S. griseus B-2682; (10) S. antibioticus ETH7451; (11) S. antibioticus ATCC10382; (12) S. albus; (13) S. antibioticus ATCC11891; (14) S. griseocarneus; (15) E. coli C600; (16) Pseudomonas aeruginosa PA01.

Full size image

Structural Similarity of the Membrane Cyclophilin Scoe/liv-M and Other Bacterial Cyclophilins

As stated above, the membrane cyclophilins (group M) form a clearly divergent group in the molecular phylogeny (Figs. 1 and 2). The peculiarity of this group of proteins is also reflected in their structure. A three-dimensional model of the structure of Scoe/liv-M was obtained as described under Materials and Methods. Five (1w74A, 1w74B, 1j2aA, 1valB, and 1v9tB) of >50 cyclophilin structures were selected from the protein data bank (PDB) by the program used to calculate the Scoe/liv-M model (Fig. 5). The resulting model has a root mean square deviation of 0.68 Å calculated for the backbone superposition with Mtub-A (1w74A) and EcoB (1j2aA). The root mean square deviation with Hsap (1CWA) is even higher, 1.38 Å, thus indicating the uniqueness of the Actinomycetales cyclophilin M group. However, the overall backbone superposition presents a good fit with the general structure of cyclophilins, especially for the most secondary structure elements of α-helices and β-sheets. The differences are particularly obvious in the loop regions that have been described as necessary to build up contacts with CsA (Mikol et al. 1993) (Fig. 5). Moreover, the Scoe/liv-M cyclophilin has changed 6 of the 13 particular amino acids that intervene in CsA binding (Norregaard-Madsen et al. 1994) (Fig. 1). The most relevant changed amino acids that are changed are histidine 125 and tryptophan 121 (both Hsap numeration), which are essential for CsA binding and hence for PPIase activity inhibition. An intermolecular hydrogen bond is formed between the side chain of this tryptophan and the CsA molecule (Liu et al. 1991). The corresponding amino acid in Scoe/liv-M is leucine (Fig. 1). All these data strongly suggest that the membrane cyclophilin M group constitutes a new family of cyclophilins with specific structural peculiarities. Further biochemical and genetic characterization of these proteins will be necessary to corroborate their cellular location and elucidate their biological function.

Fig. 5
figure 5

Three-dimensional structure comparison between the membrane cyclophilin Scoe/liv-M (blue) and members of other groups of Streptomyces cyclophilins (green). (A) Mtub-A (1w74A), rmsd 0.60 Å; (B) EcoB (1j2aA), rmsd 0.60 Å; (C) Hsap (1CWA), rmsd 1.38 Å. The backbone structures are presented as ribbons, including secondary structure elements (loops, helices, and β-sheets) (upper panels). The loop regions, which have been proven to play an important role in csA binding (Konno et al. 1996), are highlighted in the lower panels.

Full size image