Identification and Nomenclature of Immunophilin Genes in the O. sativa Genome
In an attempt to detect IMM genes in the rice genome, we searched the GRAMINE database. As a result, 56 putative IMM genes, consisting of 29 putative FKBPs and 27 putative CYPs, were identified. Additional file
1 shows the 56 genes with their names, accession numbers, putative subcellular localization, isoform numbers, amino acid numbers, number of ESTs, molecular weight, pI, expression level, and similarities with orthologous genes in
Arabidopsis thaliana and
Chlamydomonas reinhardtii IMMs. Like other photosynthetic organisms encoding the putative 52 genes of
Arabidopsis, 49 genes of
Chlamydomonas, putative 27 genes for FKBPs and 36 genes for CYPs of
Sorghum bicolor, and putative 26 genes for FKBPs and 25 genes for CYPs of
Vitis vinifera (our unpublished data), rice also has one of the largest IMM families in organisms whose genomes have been completely sequenced [
4,
6]. In addition, considering that alternative splicing occurs as a normal phenomenon in eukaryotes, the diversity of IMM proteins encoded by a genome may also increase [
25]. In rice, 12 genes of OsFKBPs and 14 genes of OsCYPs may encode alternatively spliced isoforms ranging from 2 to 4 as (Additional file
1), but it is not to purpose of this paper to provide the complete number of IMM proteins functioning in rice. More to the point,
Escherichia coli possesses 6 IMMs, including 3 FKBPs and 2 cyclophilins. The yeast
Saccharomyces cerevisiae contains 4 FKBPs and 8 CYPs. The soil nematode
Caenorhabditis elegans genome contains 8 genes for FKBPs and 24 genes for CYPs. The human genome contains 18 genes for FKBPs and 24 genes for CYPs. A large number of IMMs in eukaryotes and, in particular, in plants remains evolutionarily and functionally obscure. However, many IMMs in plants may serve a more diverse array of functions, especially in relation to photosynthesis, but may also involve a significant amount of redundancy [
4].
For the nomenclature of rice IMMs, we followed the previously reported rule for
Arabidopsis: the proteins were named FKBP for the FKBP506-binding protein and CYP for CsA, with prefix letters to indicate the species of origin (e.g., Os for
Oryza sativa) and a suffix number to indicate
Mr. For genes where > 2 protein isoforms can arise from the same gene by alternative splicing, only the longest form was used in this nomenclature [
4]. Where possible in considering characteristics such as > 50% identity in homology, the same subcellular location and the same domain architecture as with
Arabidopsis and
Chlamydomonas IMM proteins, we used the same suffix numbers for genes whose orthologs could be deduced. These include OsFKBP12, -13, -15-1, -15-2, -16-1, -16-2, -16-3, -16-4, -17-1, -17-2, -18, -19, -20-2, -53, and -72 for FKBPs, and OsCYP18-1, -18-2, -18-4, -19-2, -19-3, -19-4, -20-1, -20-2, -20-3, -21-1, -21-4, -22, -23, -26-2, -28, -37, -38, -57, -63, -65, -71, and -95 for CYPs. For some genes where > 2 genes show an orthologous sequence with the same protein as
Arabidopsis and Chlamydomonas, lowercase extension letters were added to the name according to the order of similarity. These were OsFKBP20-1a, -20-1b, -42a, -42b, -62a, -62b, and -62c for FKBPs and OsCYP40a, -40b, -59a, and -59b for CYPs. For other genes whose orthologs in
Arabidopsis or
Chlamydomonas were not obvious, these were named after the calculated
Mr of mature proteins predicted from cDNA-deduced sequences. Here, OsFKBP44, -46, -57, -58, -59, and -73 for FKBPs, and OsCYP17 for CYPs were included. One gene encoding trigger factor-like protein, distantly related to the FKBP family, was found in the rice genome, as in
Arabidopsis, and was named OsTIG. As for the nomenclature of the rice FKBP family, there has been a preceding report in which almost all the nomenclature agreed with our nomenclature; however, we substituted OsFKBP53b and -53a by Os FKBP53 and -58, respectively, and also OsFKBP64, -65 and -75 by OsFKBP62a, -62b and -62c, re-spectively. The reason for this is discussed below [
4,
8,
26].
Rice FKBP-506 Binding Proteins (OsFKBPs)
To compare the conservation patterns of the amino acid residues for the binding of immunosuppressive drug (ISD) FK506/PPIase activity and secondary structural details, the amino acids of 29 OsFKBPs were co-aligned with human FKBP12 (hFKBP12) as an external reference. hFKBP12 identifies the 14 amino acid residues for the FK506 binding and PPIase activity [
27-
29]. For FKBPs with a multiple (two or more) FKBP domain, the most conserved domain obtained from the analysis using the SMART program was used (Figure ). In addition, to analyze conservation in amino acids for FKBP domain and their orthologous relationship, the conservation patterns and ratio of the amino acid residues for FKBP domain between each ortholog of rice and
Arabidopsis FKBPs was compared (Table ). The 29 OsFKBPs identified could be classified into single-domain (SD) members with a FKBP catalytic domain, and multiple-domain (MD) members with other functional domains in addition to a single or multiple FKBP domains (Figure ). The functional domains include tetra-peptide repeats (TPR), the coiled-coil domain (CCD), and the internal repeats domain (RPT), each of which is involved in protein-protein interactions or recognitions, or in the assembly of multi-protein complexes [
30,
31]. The Arg/Lys amino acid-rich domain may function as a motif for non-specific RNA binding or mediate protein-protein interactions, and is also a frequent target for molecular interactions [
32-
34]. The calmodulin-binding motif (CaM) has been recognized as a major Ca
2+ sensor and as a modulator of regulatory events through its interaction with a protein [
35]. Finally, the full length of the amino acid sequences of all FKBPs was used to generate a phylogenetic tree in order to assay the evolutionary relationships among the OsFKBPs (Figure ).
| Table 1Conservation of key residues for FK506 binding/PPIase activity in O. sativa FKBPs (OsFKBPs) and comparison with those of A. thaliana FKBPs. |
The conservation patterns of the amino acid residues for the binding of ISD showed great differences between OsFKBPs (Figure , Table ). For example, some OsFKBPs (OsFKBP13, -15-1, -15-2, -20-1a, -20-1b, -58, -62a, -62b, and -62c) share identities of > 70%, with 14 key residues of hFKBP, whereas other OsFKBPs (OsFKBP16-1, -17-2, -18, -20-2, -42a, -42b, and -57) share identities to < 30%. In the case of OsFKBP17-2, all key residues are substituted by other residues. The conservation patterns of OsFKBPs with the 14 amino acid residues of hFKBP12 were very similar to those of the
Arabidopsis (Table ) and
Chlamydomonas orthologs [
6]. However, the repertoire and amino acid sequence of rice FKBPs are more similar to those of
Arabidopsis than
Chlamydomonas, from which they have a distant genetic distance. Moreover, the diversity of IMM proteins was somewhat different for the monocot rice than for the dicot
Arabidopsis. This information added broader evidence in favor of the theory that the diversity of IMMs was already established in their last common ancestor, probably close to the root of the "green" lineage of plants, but they may have continued to change and become more diverse since the completion of the "green" lineage. In particular, the conservation between FKBPs, as in the FKBP repertoire, among
O. sativa,
Arabidopsis and
Chlamydomonas was lower than in cyclophilins (see below), as reported in a comparative analysis of various organisms containing
H. sapiens, D. melaogaster, C. elegans, S. cereviseiae and Sz. Pombe [
5,
6].
Of the 29 FKBPs containing OsTIG, 15 have been characterized as SD members. All these proteins contain a single FKBP domain, and some harbor a targeting sequence that determines the subcellular localization of the mature protein (Figure ). OsFKBP12 is the only cytosolic FKBP having SD members, as in
Arabidopsis. It features a possible role for the regulation of the cell cycle through the interaction with AtFIP, a phosphatidyl-inositol kinase [
36]. Moreover, it was also expressed strongly (Additional file
1) as in
Arabdopsis AtFKBP12 and
Chlamydomonas FKB12, with the largest number of ESTs [
4,
6].
Ten OsFKBPs (OsFKBP13, -16-1, -16-2, -16-3, -16-4, -17-1, -17-2, -18, -19, and -20-2) contain putative chloroplast-targeting sequences in the N-terminal region. The signal sequences of all chloroplast FKBPs have double arginine residues and follow the hydrophobic region needed for chloroplast lumen translocation via the Tat (twin arginine translocase) pathway [
37]. The majority of chloroplast FKBPs have a peptidase cleavage site, Ala-Xaa-Ala, at the terminus of the hydrophobic region (Figure ), suggesting that all the chloroplast FKBPs target the thylakoid lumen like the chloroplast FKBPs of
Arabidopsis and Chlamydomonas. However, like orthologs of
Arabidopsis, AtFKBP17-2, -18, and -20-2 also lack a cleavage site at the terminus of the hydrophobic extension region, suggesting that they act as insoluble proteins within the thylakoid lumen, but proteins anchored in the thylakoid membrane [
4]. Compared with
Arabidopsis, in which the thylakoid lumen FKBPs have two FKBPs (AtFKBP17-2 and 17-3) with high sequence similarity, all thylakoid lumen FKBPs are well conserved in both rice and
Arabidopsis, except in the absence of OsFKBP17-3 in rice, considering the domain architecture (Figure ), the alignment of each amino acid sequence between those with two (data not shown), and the high similarity of conserved key amino acid residues for ISD binding (Table ). Thus, AtFKBP17-2 and 17-3, seen as gene duplicates based on their high sequence similarity, have been duplicate since the divergence of the evolutionary lineage between rice and
Arabidopsis. The common ancestor of the eukaryotic green lineage,
Chlamydomonas, also has 11 FKBPs (FKB), but orthology among them can only be claimed between FKB16-3, -16-4, -18, -19, and 20-2 and the
Arabidopsis genes with the same numbering. For the others, it has been difficult to match the ortholog relations [
6]. All in all, the diversity of chloroplast lumen FKBPs has been established in
Chlamydomonas based on the number of IMMs.
Two OsFKBPs (OsFKBP15-1 and OsFKBP15-2) contain putative ER targeting sequences at the N-terminus and have possible C-terminal endoplasmic reticulum (ER) retention signals (DSEL and NSEL for OsFKBP15-1 and OsFKBP15-2, respectively) (Figure and ), although it will need to be confirmed whether they function as real anchors in the ER membrane. They are closely related in sequence, and they are 81% similar to AtFKBP15-1 and 76% similar to AtFKBP15-2 in their amino acid sequences (Additional file
1).
Two OsFKBPs (20-1a and 20-1b) are characterized as putative nuclear targeting paralogous proteins by the presence of nuclear localization signals (NLS) in the C-terminal region (Figure ). They are also closely related in their amino acid sequence (Figure ), and are orthologous to AtFKBP20-1, with 77 and 76% similarity in their amino acid sequences, respectively (Additional file
1). Interestingly,
Arabidopsis has only single AtFKBP20-1 and AtFKBP15-3, and another nuclear targeting SD FKBP has its signal at the N-terminal region, whereas rice lacks an ortholog for AtFKBP15-3 [
4].
14 OsFKBPs containing OsTIG belong to MD members. Among them, two isoforms (a/b) of OsFKBP42 are characterized by a single FKBP domain, TPR domain, and CaM domain (Figure ). However, unlike with orthologs
Arabidopsis AtFKBP42 and
Chlamydomonas FKBP42 that have a membrane-anchoring domain at the C-terminus, and also contrary to a previous report on OsFKBP42a/b, we could not find a C-terminal membrane-anchoring domain at either OsFKBP42a or -42b using various prediction programs found on the ExPASY web site
http://www.expasy.ch/tools[
4,
6,
8]. In reality, the amino acids at the C-terminus in which the membrane-anchoring domain is positioned showed significant changes between AtFKBP42 and OsFKBP42a/b (data not shown). It is therefore supposed that the anchoring of this protein to the tonoplast or plasma membrane may be not essential for the functioning of the protein, unlike with AtFKBP42 that was predicted to be membrane-localized [
38]. OsFKBP42a (LOC_Os12g05090) and OsFKBP42b (LOC_Os11g05090) are positioned in terminal regions of chromosome 11 and 12 in rice that are known to result from a duplication 7.7 million years ago, the most recent large-scale duplication in the rice genome [
39].
Other MD FKBPs with a TPR domain are OsFKBP62a, -62b, and -62c. These three genes were previously named
OsFKBP64, -65, and
-75 and
rFKBP64, -65, and
-75, respectively, based on their molecular weight and similarity to two MD AtFKBP62 and -65 with triple FKBP domains, a TPR domain, and a CaM domain in common [
4,
8,
40]. However, OsFKBP62a, -62b, and -62c retain 77, 75 and 73% similarity with AtFKBP62 in amino acid sequence, respectively, and 41, 41 and 42% similarity with FKB62, respectively (Figure , Additional file
1), and all belong to one stem branch of clade II in the phylogenetic tree (Figure ). In addition, these three proteins have almost the same 14 key residues for PPIase activity and are conserved in 12 out of 14 key residues of hFKBP12 for PPIase activity, suggesting that they are catalytically active isoforms (Table ). AtFKBP62 and -65 were thought to have originated from regional duplicates between chromosomes I and V [
4]. Taken together, all this suggests that rice MD OsFKBP62a, -62b and -62C are paralogs duplicated from an ancestral gene, probably
Chlamydomonas FKB62, whereas
Arabidopsis has only two duplicated paralogs, AtFKBP62 and -65. In the meantime, unlike OsFKBP62a and -62b, OsFKBP62c (OsFKBP75, rFKBP75) contains ~50 residues of a glycine repeat domain in the N-terminal area and an extension of ~40 residues at the C-terminus, identified to the putative transmembrane domain (Figure ). The glycine repeat domain at N-terminus is associated with consensus targeting for the endoplasmic reticulum, and rFKBP75 was predicted to be an ER membrane protein [
40]. In
Arabidopsis, AtFKBP62 (ROF1) and AtFKBP65 (ROF2) are regulated by age and biotic stresses [
19,
41,
42]. Recently, ROF1 was thought to have played a role in the prolongation of thermotolerance by sustaining the levels of small HSPs essential for survival at high temperatures. Additionally, the ROF1-HSP90.1 complex was localized in the nucleus during exposure to heat stress, whereas it was localized in the cytoplasm under normal conditions [
19].
OsFKBP72 - with a triple FKBP domain, TPR domain, and NLS at the N-terminal area - shows 67% similarity and absolute similarity in terms of amino acid residues for PPIase with AtFKBP72 (Figure , Table ). Unlike other MD FKBPs, which are composed of triple FKBP domains, the second FKBP domain of OsFKBP72 is the most highly conserved domain and shows remarkable differences in terms of key residues of hFKBP12 for ISD binding/PPIase activity. It also forms a separate branch within clade II in the phylogenic tree (Figure ), suggesting that it originated from another ancestral gene and is likely to lose its function as PPIase. AtFKBP72 (PAS1) is involved in the control of cell proliferation and differentiation during plant development [
43,
44]. The C-terminus of PAS1 is required for binding with the NAC-like transcription factor and nuclear translocation [
44]. OsFKBP72 also shows considerable similarity with AtFKBP72 in the C-terminal area. However, we could not find the CaM domain C-terminal area of OsFKBP72 using a CaM prediction program, unlike with AtFKBP72 (PAS1), a putative CaM domain in the C-terminal area due to the binding of calmodulin
in vitro [
43].
Among the MD FKBPs belonging to clade II, OsFKBP73 with its triple FKBP domain and OsFKBP46 and OsFKBP57 with their double FKBP domains diverge within the same branch node and show partial similarities in terms of key residues for ISD binding/PPIase (Figure , Table ). OsFKBP73 has a long Arg/Lys amino acid-rich domain (from aa 402 to 653) at the C-terminus, and OsFKBP57 retains a coiled coil domain in the C-terminal area (Figure ). OsFKBP46 and OsFKBP57 that have no signal motifs were predicted to be cytosolic, whereas OsFKBP73, with its long Arg/Lys amino-acid-rich domain, was predicted to be nuclear by several prediction programs. Otherwise, as in the case of PAS1, we infer that it is a cytosol/nucleus shuttle protein. Orthologs of these three FKBPs are unlikely to be in
Arabidopsis, although they have weak identity with AtFKBP65 in their amino acid sequences (Additional file
1). Also, the putative orthologs of these genes that are highly conserved in amino acid sequence existed in other plants containing
Poplus and
Vitis (data not shown). The novelty of these genes and the suggestion that these three genes originated from the regional duplications of rice chromosome 1 has recently been reported [
8].
In clade III, four MD FKBPs (OsFKBP44, -53, -58, and -59) were grouped as 4 branches of the same node (Figure ) and were analyzed to the highest identity with AtFKBP53 and FKB53 (Additional file
1) via blastP analysis. Among them, OsFKBP53 and -58 contain highly charged domains, putative nuclear targeting signals in the N-terminal regions, and a single FKBP domain (Figure ). OsFKBP58 and -53 were previously named OsFKBP53a and -53b, respectively [
8]. However, as stated above, all four OsFKBPs of clade III have the closest similarity to AtFKBP53 and FKB53 as a probable ancestral protein. In addition, unlike OsFKBP53a that displays 53% similarity to AtFKBP53, the other three FKBPs display < 50% similarity to AtFKBP53 (Additional file
1). Thus, we named only OsFKBP53 as an ortholog of AtFKBP53, and the others were referred to by molecular weight because there were no adequate orthologs in
Arabidopsis. OsFKBP58 has 36% similarity to AtFKBP53 (Additional file
1) and retains 8 - 20 mer of amino acid insertion in 3 areas besides the FKBP domain, and thus the CCD domain was predicted to be between amino acids 353 and 381 (Figure ). OsFKBP44 has a long N-terminal area containing two internal repeat domains and a FKBP domain, showing some degree of similarity to AtFKBP15-3 (data not shown). MD AtFKBP53 and a SD AtFKBP15-3 showed high sequence similarity, but also that rice lacks an ortholog of AtFKBP15-3, suggesting that these proteins (OsFKBP44 and AtFKBP15-3) also originated from
Chlamydomonas FKB53, but that duplication of two proteins occurred after the lineage divergence of the two plants from a common ancestor.
Finally, the distantly related member of the FKBPs is the trigger factor OsTIG, which also contains an FKBP domain, an N-terminal and C-terminal domains, both of which help to bind ribosomes, probably being located in the chloroplast stroma. The precise function of the trigger factor in the chloroplast remains unknown, but it is a PPIase and chaperone associated with the ribosome that is involved in the early steps of protein folding [
45].
Rice Cyclosporin-A Binding Protein (OsCYPs)
The 27 identified OsCYPs can be classified as either SD members with a CYP catalytic domain or MD members with other functional domains in addition to a single CYP domain. This number compares well with the 29 AtCYPs for Arabidopsis and the 26 genes for Chlamydomonas, indicating that the diversity of CYPs has been maintained or even slightly increased since they were established in the photosynthetic green ancestor. To compare the conservation patterns of the amino acid residues for CsA binding/PPIase activity and secondary structural details, the full length of the single domain of CYPs and a truncated CYP domain for the MD CYPs were used for the amino acid alignment. Human cyclophilin A was co-aligned as an external reference (Figure ), and the full-length of the amino acid sequences of all CYPs was used to generate a phylogenetic tree.
Among the 27 CYPs, 18 are SD members and 9 have been characterized as MD proteins. As with
Arabidopsis, none of the cyclophilins contains multiple catalytic domains. They also contain more divergent functional domains, such as the TPR domain, the WD-40 repeat, the U-box domain and the Zinc finger, each of which is involved in protein-protein or protein-DNA interactions. An RNA recognition motif (RRM) that may interact with RNA was also found. 6 of the 9 MD CYPs have Arg/Lys amino acid-rich domains that may function as a motif for non-specific RNA-binding or mediate protein-protein interactions, and are also frequent targets of molecular interactions (Figure ) [
32-
34].
Based on these data, we assigned the putative one-to-one orthology between O. sativa and Arabidopsis CYPs. Among 27 OsCYPs, 26 members could have matching orthology with a protein among Arabidopsis CYPs, whereas for only OsCYP17, we could find no obvious orthology with Arabidopsis CYPs.
6 SD CYPs (OsCYP17, -18-1, -18-2, -18-4, -19-2, -19-3) lack an N-terminal signal peptide and are probably cytosolic. Among the OsCYPs, OsCYP17 is the smallest cyclophilin due to the deletion of β1 and β2 sheets (Figure ); it retains 37% identity with AtCYP19-1 in amino acid sequences checked with BlastP (Additional file
1). However, OsCYP17 and AtCYP19-1 are significantly different in the conservation ratio of key residues for CsA binding, each of which showed 43 and 100% conservation, respectively (Table ). In addition, OsCYP17 forms a separate branch within clade I of the phylogenic tree (Figure ). It is noteworthy that except for
O. sativa ssp. Japonica and ssp. Indica, no proteins with orthology to OsCYP17 were found in any other plants using BlastP analysis (data not shown). These results suggest that OsCYP17 occurred most recently in the evolution of
O. sativa. OsCYP18-1 and -18-2 retain 89 and 82% identity with AtCYP18-1 and -18-2 in amino acid sequence, and 70 and 75% with CYN18-1 and -18-2, respectively (Additional file
1). Also, the two proteins exhibit phylogenetic divergence due to a tetra- (18-2) or penta- (18-1) peptide deletion in the β2 domain (Figure and Figure ). It is suggested that these two proteins may have a distinct and unchanged function for a long period of time, though with a low level of expression (Additional file
1). In addition, these two proteins belong to clade II, which is comprised of the two SD CYPs (OsCYP18-1 and -18-2), as well as 3 MD CYPs (OsCYP57, -65 and -71), and OsCYP18-1 and OsCYP65 share the same branch (Figure ). These patterns are almost the same in rice,
Arabidopsis, and
Chlamydomonas, suggesting that these genes occurred in a common ancestor before
Chlamydomonas [
4,
6]. Putative cytosolic OsCYP18-4, -19-2, and -19-3; secretional OsCYP19-4 and OsCYP21-1; mitochondria-targeting OsCYP20-1 and OsCYP22; and chloroplast-targeting OsCYP20-2 and OsCYP20-3 are closely related, with similar amino acid sequences, and mostly belong to 2 branches of clade I (Figure ). Among them, OsCYP18-4, OsCYP19-2, OsCYP19-3, OsCYP19-4, OsCYP20-1, OsCYP21-1 and AtCYP22 share a 5-to-8 amino acid insertion located between the α1 helix and β3 sheet (Figure ). However, all of these members are perfectly conserved in the 7 amino acids for CsA binding, except for Os-CYP19-4 in which 121-tryptophan is changed to arginine (Table ), suggesting that these are catalytically active isoforms. Based on these characteristics, there is a relatively clear one-to-one orthology between rice CYPs and
Arabidopsis CYPs, but no orthologs corresponding to AtCYP18-3 and AtCYP19-1 were found in rice. It has been suggested that two pairs of
Arabidopsis CYPs, AtCYP18-3 and AtCYP19-1, and AtCYP18-4 and AtCYP19-2, arose by gene duplication, implying that the gene duplication occurred only in the
Arabidopsis lineage because of divergence of the lineage between
O. sativa and
Arabidopsis, and these two pair of genes may entail redundancy [
15,
46]. Another cytosolic OsCYP19-3 that shows 79% identity with AtCYP19-3 also is very similar to OsCYP19-2 in amino acid sequence (Additional file
1), except for some divergence in the α3 domain (Figure ). The OsCYP19-3 may have a distinct function from OsCYP19-2, considering that OsCYP19-3 has been survived a long period of evolution in most plants, including
Arabidopsis and rice (data not shown). It is also widely accepted that gene duplication is a primary source of new functional genes [
47]. Putative ER targeting SD OsCYPs, OsCYP19-4 and -21-1 show 70 and 69% similarity to the ER proteins AtCYP19-4 and AtCYP21-1, respectively. They also display 68 and 69% similarity with putative ER localized CYN20-1, respectively, probably indicating an ancestral gene (Additional file
1). The ER location of AtCYP19-4 (CYP5) has been confirmed by N-terminal green fluorescent protein fusion, and may involve the modulation of a guanine nucleotide exchange factor essential for vesicle trafficking [
48,
49]. Interestingly, OsCYP20-1 has the greatest similarity to putative ER located AtCYP20-1 and CYN20-1; however, it showed mitochondrial localization by both TagetP and PSORT analysis. Unlike AtCYP20-1, OsCYP20-1 retains additional 21-mer (maybe a signal for mitochondrial targeting) at the N-terminus of the protein (Figure ), and this additional peptide proved to be common in several other plants (data not shown). Thus, the exact location of OsCYP20-1 needs to be determined by further experiments. In particular, AtCYP21-1, the ortholog of OsCYP21-1, has yet to have its location verified, being either cytosolic or ER [
4,
50]. In the meantime, rice lacks the ortholog for another ER CYP, AtCYP21-2, suggesting that AtCYP21-2 was duplicated from one of the other ER targeting isomeric genes: AtCYP19-4, -20-1, or -21-1, in the
Arabidopsis lineage alone. Among OsCYPs that have highly conserved CsA binding amino acids, OsCYP20-2 (chloroplast thylakoid lumen) and OsCYP20-3 (chloroplast stroma) show ortholog relationships with AtCYP20-2 and AtCYP20-3 and CYN20-3, respectively. However, OsCYP22, the ortholog of AtCYP22 and CYN22 that are putatively cytosolic, scored high for targeting to chloroplast stroma based on both PSORT II and Target P analysis. AtCYP22 also contains an N-terminal extension, but it did not score highly for targeting to any subcellular compartment, whereas the signal peptide of AtCYP22 shows the changes in some amino acids containing the insertion of hydrophobic 5 amino acids, which may affect its signaling properties (data not shown).
| Table 2Conservation of key residues for Cyclosporin A binding/PPIase activity in O. sativa cyclophilins (OsCYPs) and comparison with those of A. thaliana cyclophilins. |
Among SD CYPs showing significant phylogenic diversity in amino acids for CsA binding mitochondria-located, OsCYP21-4 is not conserved in 7 amino acids for PPIase and shows 54% identity in the amino acid sequence with AtCYP21-4. It also does not have an ortholog in
Chlamydomonas, suggesting it appeared after
Chlamydomonas (Additional file
1 and Table ). Rice lacks the ortholog for another mitochondria CYP, AtCYP21-3, which shows very similar signal peptides for 47 amino acids and highly homologous C-terminal CYP domains with AtCYP21-4, suggesting that it originated by duplication of AtCYP21-4 only in the
Arabidopsis lineage [
4]. ER-located OsCYP23 is also significantly divergent in key residues for CsA binding, but it has orthologs in other plants (data not shown) containing both
Arabidopsis and
Chlamydomonas, which suggests that it has a distinct function with ER-located other CYPs in which PPIase activity is well conserved (Table ). Chloroplast-located SD CYPs OsCYP26-2, -28, -37 and -38 show a very low conservation ratio and have significant modifications in key residues (Figure ). They also belong to clade III, based on phylogenetic analysis (Figure ), unlike AtCYP20-2, which is present in clade I and has highly conserved key residues for CsA binding [
51-
53]. Although they show high divergence in their amino acid sequences, including the CYP area, they have orthologs with significant homology in the CYP area, including key residues in both
Arabidopsis and
Chlamydo-monas. The diversity may be inherited from cyanobacteria, which contains two CYPs similar to CYN37 and -38 [
6]. The putative leucine-zipper domain of AtCYP38 (TLP40) is not conserved in OsCYP38, CYN38 and orthologs of other plants (data not shown), and the role of the domain has not been clarified after several studies, indicating that this is not a conclusive point of evolution for the proper function of this protein [
54-
56]. All chloroplast lumen-located OsCYPs also contain double arginine residues, and follow the hydrophobic region needed for chloroplast lumen targeting (the signal sequences of OsCYP37 and OsCYP38 are not shown here for adequate CYP domain alignment) in the extension signal of the N-terminus (Figure ).
Nine OsCYPs belong to MD members. Among them, two putative cytosolics, OsCYP40 (a/b), contain the C-terminal TPR domain and show 73 and 70% similarity to AtCYP40, respectively (Additional file
1). The existence of two isomeric genes has even been confirmed in some plants (
Poplus,
Sorghum and
Vitis) through BlastP analysis (data not shown). Duplication probably occurred immediately after the rice and
Arabidopsis lineages separated. The putative nuclear-located OsCYP59a and -59b, seemingly representing another duplicated gene among MD OsCYPs, are organized in tandem on the same chromosome (Additional file
1), and have RRM, zinc-finger and Arg/Lys-rich domains (Figure ). They are orthologous to AtCYP59, which was identified as a dynamic interplayer between transcription and pre-mRNA processing through interaction with SR proteins and the C-terminal domain of RNA polymerase II [
57]. Interestingly, only OsCYP59a includes the long insertion between the α1/β3 and β4/β5 domains, and they may have distinct functions. However, only one form of two isomeric genes was found in the other plants (data not shown), indicating that the duplication occurred most recently within the rice lineage. Nuclear-located OsCYP57 and OsCYP65 and cytosolic OsCYP71 belong to clade II (Figure ) and have well conserved CYP domains (Table ). These are orthology relationships with CYPs with the same suffix number for AtCYP and CYN. OsCYP65 contains the U-box and Arg/Lys-rich domain, and was classified as putatively nuclear targeting, unlike cytosolic AtCYP65, for which no Arg/Lys-rich domain was located (Figure ). OsCYP71 contains the N-terminal WD40 repeats and shows high orthology (81% identity) with AtCYP71 (Additional file
1). Finally, putative nuclear-located MD OsCYP95 is annotated to 436 amino acids in the GRAMENE database and 670 amino acids in the NCBI database, respectively, although they are proteins from the same loci of chromosome II. From the analysis of other plant CYPs containing
O. sativa ssp. Indica using the GRAMENE and NCBI database, we conclude that the 670 amino acid from the NCBI database is probably correct. The ortholog of OsCYP95 is absent in
Chlamydomonas, and belongs to the same branch as OsCYP63 within clade II (Figure ), indicating that OsCYP95 may be a duplicate of OsCYP63 from before the divergence of rice and
Arabidopsis, but after
Chlamydomonas.
Expression Patterns of Immunophilins under Salt and Desiccation Stress
The expression of most of rice's IMMs was identified by the presence of ESTs in the NCBI
O. sativa EST database. OsFKBP44 and OsCYP17 did not show ESTs, but their expression was confirmed by RT-PCR analysis (Additional file
1). The expression level using the latter method corresponds roughly to the number of ESTs with all tissues of young plantlets (data not shown). To identify stress-induced Os IMMs under salt and desiccation conditions, RT-PCR was carried out at least 3 times on each IMM. Under the salt stress of 200 mM NaCl, OsFKBP20-1b, -58, -16-1, and -62a mRNA levels were increased. The level of gene expression began to increase 24 h after salt treatment, but in the case of OsFKBP62a it rose 1 h after salt treatment. The strongest salt induction pattern can be seen on OsFKBP20-1b, which is rarely expressed in the absence of stress (Figure ). In the OsCYPs, OsCYP19-3, -20-2, -21-4, -23, and -28, gene expression was strongly induced 3 h after salt treatment and continued until 72 h. OsCYP17 -19-4, -20-1, -57, and 59b expression was also induced under salt stress. In particular, OsCYP59b expression increased only 24 h after treatment (Figure ).
Regarding desiccation stress, 8 OsFKBPs and 5 OsCYPs were upregulated in our water stress conditions (see the methods). Expression in OsFKBP19, -20-1a, and -20-1b was strongly induced by desiccation stress, with OsFKBP16-3, -42a, -42b, -58, and -62a also showing increased expression levels. Expression in OsFKBP20-1a, -58, and -42b was induced at 1 h after desiccation stress, but expression in OsFKBP16-3, -19, -20-1b, -42a, and OsFKBP62a was induced 24 h after desiccation (Figure ). In OsCYPs, the expression of the OsCYP21-4 and -59b genes were induced significantly under desiccation stress, whereas expression for OsCYP19-2, -20-2, and -57 was only slightly increased. Among the desiccation affected OsCYPs, OsCYP19-2, -20-2, -21-4, and -57 showed increased expression 6 h after stress, but only OsCYP59b showed a change in the induced expression at 24 h after desiccation, being notably similar to salt stress response (Figure ). As with OsCYPs induced under salt stress, desiccation inducing OsCYPs also are composed of CYPs targeted to the various cellular organs. These results suggest that various subcellular Os IMMs respond to abiotic water-stress (salt and desiccation) to defend or regulate cellular effects under changed circumstances. Some of the Os IMM genes were upregulated by both salt and desiccation stress, but others showed gene expression against only one stress. This suggests that although both salt and desiccation cause water stress to cells in common, some IMMs may be involved in distinct signal transduction or defense pathways against each individual stress before the cell activates a common stress response.
Three OsFKBPs, OsFKBP20-1b, -58 and 62a, were upregulated under both salt and desiccation stress.
Arabidopsis ROF1 (AtFKBP62) expression was induced by heat stress and developmentally regulated. ROF1 binds heat-shock proteins HSP90.1 via the TPR domain and is located in the cytoplasm under normal conditions, but in the nucleus during heat stress [
19]. The expression of rice IMM FKBP62a (rFKBP64) was also elevated by heat stress in various organs, indicating that OsFKBP62a plays specific physiological functions in other stress responses as well as in the heat stress response in plants [
40]. As stated above, the putative nucleus OsFKBP58 (OsFKBP53a) lacks an ortholog in
Arabidopsis, and only the OsFKBP58 gene responded to both salt and desiccation stress, suggesting that it plays a different role from OsFKBP53 and AtFKBP53.
Expression in OsFKBP16-3, 19, -20-1a, -42a, and -42b was upregulated under desiccation, but not salt, stress. OsFKBP16-3 contains two cysteine residues separated by 3 intervening non-cysteine residues at the N-terminus of the putative mature protein, predicting an intra-disulfide bond and suggesting a role for the redox response related to stress. Actually, the activity of a stromal AtCYP20-3 (ROC4) and the thylakoidal AtFKBP13, are regulated by redox relay of the thiol group [
58,
59]. Another thylakoid IMM, OsFKBP19, which has two cysteine residues in the loop regions between the β2 and β3 domains, is also upregulated by desiccation stress, whereas thylakoid lumen OsFKBP13 and OsFKBP16-2 that have two coupled cysteine residues forming a disulfide bond and upregulated by light irradiation, did not respond to salt and desiccation stress [
4]. OsFKBP20-1a proved to be nuclear-targeting and heat-inducible; its over-expression in yeast indicates a capacity for high-temperature tolerance [
60]. They also show that the OsFKBP20-1a and SUMO-conjugating enzyme interacted physically to mediate the stress response of rice plants. The paralog OsFKBP42a and 42b are ~90% identical, but they have slightly different EST numbers (Additional file
1) and distinct spatial expression profiles, with OsFKBP42a being expressed at the highest levels in crown tissue, whereas OsFKBP42b is expressed more highly in the roots [
8]. Expression of both OsFKBP42a and -42b was upregulated by desiccation, but the timing of the increase for both genes was different. Taken together, the two paralog genes OsFKBP42a and -42b may retain the same function, but have different spatial/temporal regulation.
Among the 9 OsCYP genes upregulated under salt and desiccation stress, 4 genes (OsCYP20-2, -21-4, -57, and -59b) responded to both forms of stress, while others responded only to salt stress. AtCYP20-2, the ortholog of OsCYP20-2, the only CYP to show PPIase activity in the thylakoid lumen, was increased when plants were exposed to strong light or low temperature, and it has already been suggested that this PPIase plays a functional role in acclimation responses [
52,
61,
62]. However, the phenotype and protein composition of the thylakoid lumen in
AtCYP20-2 mutants seems indistinguishable from that of the wild type plants, and it is concluded therefore that AtCYP20-2 is unrelated to its PPIase capacity [
52,
53]. The OsCYP20-2 gene, the ortholog of AtCYP20-2, is also expressed at the highest level in the OsCYP family (Additional file
1), showing distinct upregulation under both salt and desiccation stress. From these results, we suggest that At/OsCYP20-2 maintains the homeostatic state of thylakoid lumen under stress, although it may be unrelated to its PPIase activity. Another two CYPs, the putative mitochondrial OsCYP21-4 and nuclear OsCYP57, were positively regulated in response to both salt and desiccation stress; none of the functions of the two proteins and their other plant orthologs are known. Nuclear-located AtCYP59b with the
Arabidopsis ortholog OsCYP59b and the RNA-recognition motif (RRM), interacted with the C-terminal domain of the largest subunit of RNA polymerase II and SR proteins, and this may function in activities connecting transcription and pre-mRNA processing [
57]. The OsCYP59b gene was transiently upregulated only 24 h after both stress treatments; however, OsCYP59a (maybe a duplicate of OsCYP59b) expression was unchanged by salt/desiccation stress.
The expression of 5 OsCYPs (OsCYP19-3, -19-4, 20-1, -23, and -28) was upregulated by salt, but not by desiccation, stress. Cytosolic AtCYP18-1, -18-2, -18-3, -18-4, -19-1, -19-2, and -19-3 are highly conserved and probably have similar functions [
4]. Similarly, cytosolic OsCYP19-2, -19-3, and -19-4 are highly conserved; OsCYP19-2 expression was upregulated by desiccation stress, and OsCYP19-3 and OsCYP19-4 were upregulated by salt stress, suggesting that these genes also have a similar function. The ortholog of OsCYP19-4, AtCYP19-4 (CYP5), interacts with GNOM, which is involved in the coordination of cell polarity along the apical-basal embryo axis in
Arabidopsis [
48]. OsCYP19-4 was abundant only in pistils at anthesis and 1 h after anthesis and was undetectable in leaves, roots, flowers, and embryos. CYP5 may regulate the function of the GNOM protein during embryogenesis [
63]. However, considering our results indicating that the OsCYP19-4 gene was expressed in young rice plants without pistils or in the embryogenesis stage under salt stress, the full role of this protein remains to be clarified. AtCYP20-1, the ortholog of OsCYP20-1, interacted with the regulatory subunit RCN1 of a Ser/Thr-specific protein phosphatase, which suggests that AtCYP20-1 and the RCN1 complex is required for the control of root cell proliferation [
64]. AtCYP20-1 was identified as an unfolded protein response (UPR) gene following endoplasmic reticulum stress [
65]. Unlike with AtCYP20-1, nothing about the function of the putative mitochondria locating OsCYP20-1 has been reported. Putative ER-locating OsCYP23 and thylakoid lumen-locating OsCYP28 are not conserved in amino acids for CsA binding (Table ), and given their upregulated expressions only under salt stress, but not under desiccation stress, we must infer that they are associated with specific substrates for metabolic regulation related to sodium or chloride ions.
2 and Hye Sun Cho
This article has been 
