Evolution of Chloroplast J Proteins

Hsp70 chaperones are involved in multiple biological processes and are recruited to specific processes by designated J domain-containing cochaperones, or J proteins. To understand the evolution and functions of chloroplast Hsp70s and J proteins, we identified the Arabidopsis chloroplast J protein constituency using a combination of genomic and proteomic database searches and individual protein import assays. We show that Arabidopsis chloroplasts have at least 19 J proteins, the highest number of confirmed J proteins for any organelle. These 19 J proteins are classified into 11 clades, for which cyanobacteria and glaucophytes only have homologs for one clade, green algae have an additional three clades, and all the other 7 clades are specific to land plants. Each clade also possesses a clade-specific novel motif that is likely used to interact with different client proteins. Gene expression analyses indicate that most land plant-specific J proteins show highly variable expression in different tissues and are down regulated by low temperatures. These results show that duplication of chloroplast Hsp70 in land plants is accompanied by more than doubling of the number of its J protein cochaperones through adding new J proteins with novel motifs, not through duplications within existing families. These new J proteins likely recruit chloroplast Hsp70 to perform tissue specific functions related to biosynthesis rather than to stress resistance.


Introduction
The heat shock protein 70 kD (Hsp70) family chaperones have been shown to function in many biological processes including protein folding, protein translocation, protein complex assembly and degradation of misfolded proteins.Hsp70s located inside chloroplasts also perform many functions.In green algae analyzed, chloroplast Hsp70 is encoded by a single gene and has been shown to be involved in photoprotection and repair of photodamaged photosystem II [1].In all the land plants analyzed, chloroplasts contain more than one Hsp70 [2][3][4].Different functions have been suggested for land plant chloroplast Hsp70s, including folding and assembly of individual proteins or protein complexes [5][6][7][8].One of the chloroplast Hsp70s in Physcomitrella patens is important for protein translocation into chloroplasts [3].Arabidopsis has two chloroplast Hsp70s, cpHsc70-IV and cpHsc70-V (originally named cpHsc70-1 and cpHsc70-2, respectively; cp stands for chloroplast and the Roman numerals stand for the chromosome where the gene is located.We changed the names in order to avoid confusion with different mutant alleles of each locus).Both are important for protein import into chloroplasts [4] and cpHsc70-IV is also important for conferring thermotolerance to germinating seeds [9].It is not known whether the functions of algal chloroplast Hsp70 are still preserved in land plant chloroplasts or whether the identified functions of land plant chloroplast Hsp70s are unique to land plants.
One way to investigate the functions of Hsp70 and to reveal new involvement of Hsp70 is to characterize its J domaincontaining cochaperones, otherwise known as the J proteins.Hsp70 is recruited to specific functions by association with designated J proteins [10].One can follow the evolution of J proteins to identify the addition or loss of Hsp70 participation in certain processes.J domain is a four-helix structure of approximately 70 amino acids with an invariant histidine-proline-aspartic acid (HPD) tripeptide motif located in the loop between helix II and III.The HPD motif is necessary for stimulating ATP hydrolysis by Hsp70.Some J proteins also deliver substrate proteins to Hsp70.J proteins are ubiquitous in prokaryotes and eukaryotes.The prototype of J proteins is E. coli DnaJ, which contains four structural domains: an N-terminal J domain, followed by a Gly/Phe-rich domain, a Zn 2+ -finger domain and a less conserved C-terminal domain.Consequently, J proteins are classified into three types.Type-I J proteins contain all four domains described for E. coli DnaJ.Type-II J proteins contain the J domain and the Gly/Phe-rich domain.Type-III J proteins have only the J domain in common with E. coli DnaJ [11].Some J proteins contain additional structural domains not present in E. coli DnaJ, such as transmembrane domains, tetratricopeptide repeat (TPR) domains and ferredoxin (Fd) domains [12,13].Some proteins contain a J domain-like structure but lack the conserved HPD tripeptide motif.These proteins are referred to as J-like proteins.
In Chlamydomonas reinhardtii, five chloroplast J proteins, named CDJ1 to CDJ5, have been analyzed.They are suggested to function in processes ranging from biogenesis of thylakoid membranes, translation, to mRNA stability [12,[14][15][16].Several J proteins have also been identified in pea and Arabidopsis chloroplasts, including one type-I J protein (PCJ1, [17]) and four type-III J proteins [18][19][20][21].Arabidopsis J8, J11 and J20 (called DJC22, DJC23 and DJC26, respectively, in [22] and the current work) have been suggested to be involved in optimization of photosynthesis [18].CRRJ (called NdhT in [23] and DJC75 in [22] and the current work) is a thylakoid membrane protein and is essential for the activity of the chloroplast NAD(P)H dehydrogenase (NDH) complex functioning in cyclic electron transport [21].It is not known which higher plant chloroplast J proteins have homologs in green algae and cyanobacteria or which J proteins are newly evolved and thus their functions may be unique to land plants.It is also not known how multiplication of chloroplast HSP70 genes in land plants has affected the number of chloroplast J proteins.For example, J proteins may have also duplicated within families, or new families of J protein may have evolved to direct chloroplast Hsp70 to new functions.
As one of the first steps toward characterizing the evolution and functions of chloroplast Hsp70s and J proteins, we investigated the constituency, domain structure and evolutionary origins of Arabidopsis chloroplast J proteins.Because predictions of chloroplast-targeting transit peptides tend to have a higher false positive rate and proteomic analyses tend to only identify proteins of higher abundance, we combined all available transit peptide prediction and proteomic databases and then verified all candidate proteins by individual chloroplast protein import assays.Although laborious, this approach will provide the best-possible complete list of chloroplast J proteins.Only with a near complete list of members can analyses on the evolution of chloroplast J proteins be performed.We found that Arabidopsis chloroplasts contain at least 19 J proteins.Phylogenetic analyses showed that these 19 J proteins could be classified into 11 clades.All 11 clades are conserved in land plants, and more than half of them are present only in land plants, suggesting many new functions have evolved for land plant chloroplast Hsp70s.Analyses of expression patterns indicate that land plant-specific J proteins in general show highly variable expression levels in different tissues and are down regulated by low temperatures.

Nomenclature of Arabidopsis J Proteins
Four systems have been used for naming Arabidopsis J proteins in the literature.The first and most commonly used is the prefix ''AtJ'' followed by an Arabic numeral that stands for the order of appearance of the proteins in publications.For example, the first and second Arabidopsis J proteins published were named AtJ1 and AtJ2 [24,25].Later, a report searching the then newly finished Arabidopsis genome identified 89 J proteins in Arabidopsis [26].This report used a different nomenclature: ''atDj(A, B, C)x'', in which the ''at'' stands for Arabidopsis thaliana, ''Dj'' stands for DnaJ, A, B, and C represent type I, II and III J proteins, respectively, and ''x'' is a number from 1 to 89 which mirrors the original system of ordering by the appearance of the J proteins in published works, regardless of the type of protein.However, most subsequent papers continue to use the ''AtJx'' system for Arabidopsis J proteins.In 2009, Rajan & D'Silva searched the Arabidopsis genome again and found that there were 116 J proteins in Arabidopsis [27].Although these investigators also used the ''atDj(A, B, C)x'' acronym, in their work the ''x'' stood for a serial number within each type of J proteins.For example, atDjA3 and atDjB9 in Miernyk (2001) were renamed atDjA1 and atDjB1, respectively, by Rajan & D'Silva (2009).Recently, Finka et al. (2011) revised the total number of J proteins from 116 to 105, and renamed J proteins as ''DJ(A, B, C)x'' with the numeric number ''x'' re-sorted [22].This correction was based on the finding that 8 of the J proteins listed by Rajan & D'Silva contain transposable elements and another 3 were given two independent numbers.In this article, we have decided to follow Finka et al. (2011); however, for clarity Table 1 lists all the J proteins tested in this study and their corresponding names according to all the nomenclature systems.

Nineteen J Proteins were Imported into Chloroplasts
Using various prediction algorithms, Miernyk (2001) and Rajan & D'Silva (2009) suggested that 18 J proteins might have chloroplast-targeting transit peptides.In addition, we searched various published chloroplast proteomes and found that DJC42, although predicted as a nuclear protein [27], was detected in chloroplasts [28].DJC62, a TPR domain-containing J protein, was also predicted to localize in chloroplasts by Pradas et al. (2010).In addition, At3g05345, named DJC82 in our current study, is a newly annotated J protein that is predicted to be localized in chloroplasts (TAIR, http://www.arabidopsis.org/).To determine how many of these 21 putative chloroplast J proteins are localized to chloroplasts, protein import assays were performed.The 21 J proteins were synthesized and labeled with [ 35 S]Met by in vitro translation, and then incubated with isolated pea chloroplasts under import conditions.After import, a portion of the chloroplasts was further treated with thermolysin to remove surface-associated precursor proteins.Our initial results showed that lower-molecular-weight mature proteins were produced after the import of 15 of the J proteins analyzed (Figure 1A) and their mature proteins were also thermolysin-resistant after import.For all 15 proteins, no protein with the same molecular weight as the mature protein was detected from thermolysin-treated precursor proteins without import, indicating that the thermolysin-resistant mature protein was produced as a result of import into chloroplasts, not as a result of the intrinsic protease resistance of the proteins.
Four of the J proteins analyzed (DJC42, DJC72, DJC73 and DJC78) produced no thermolysin-resistant proteins after import.Among them, DJC72 only has the initial methionine, and DJC73 and DJC78 only contain methionines within the N-terminal half of their polypeptides.If these three J proteins were imported into chloroplasts, their mature protein might not be seen after the removal of the N-terminal transit peptide.Constructs with two methionines added to the C terminus of these three J proteins were therefore generated (DJC72MM, DJC73MM and DJC78MM).Lower-molecular-weight mature protein could indeed be detected after the import of DJC72MM and DJC73MM (Figure 1B), but not DJC78MM (Figure 1C).DJC78 was previously annotated as a protein of 347 residues [27] and contained a predicted N-terminal transit peptide.Recently, based on EST information, a new annotation from TAIR shows that DJC78 has only 230 residues, lacking the N-terminal 117 residues of the previous annotation.We also found that the coding sequence of the newly annotated DJC78 could be amplified by RT-PCR, but the 347-residue DJC78 from the previous annotation could not (data not shown).These data suggest that the annotation of previous 347-residue DJC78 with a predicted transit peptide was incorrect.No attempt was made to retest DJC42 (Figure 1C) by methionine addition because it contains 1,104 residues and the last methionine is at residue 1,033.DJC42 was identified in a chloroplast proteome experiment by only one peptide [28].Therefore association of DJC42 with chloroplasts might be a result of contamination.
DJC31 is a large TPR domain-containing J protein of approximately 129 kD.Its import efficiency into chloroplasts was low and three proteins were seen after its import (Figure 1B, DJC31, arrow and asterisks).Of these, the protein with the highest molecular weight (Figure 1B   proteins with sizes between 14 to 18 kD were produced (Figure 1D, lanes 3 and 4, m1 and m2).This result confirmed that DJC31 has a short transit peptide of about 5 kD.The relationship between the two mature proteins m1 and m2 is not known but both were only produced after import and were localized within chloroplasts as shown by their resistant to thermolysin treatment (Figure 1D, lane 4).We also do not know the origins of the two smaller proteins produced after the import of the full-length DJC31 (Figure 1B, DJC31, asterisks), but they were most likely degraded forms of mature DJC31.Interestingly, similar results were seen after the import of the other TPR domain-containing J protein DJC62, which has even poorer efficiency of in vitro import.Two proteins were seen after the import of DJC62, with one very close in size to the precursor and the other about 50 kD smaller (Figure 1B, DJC62, arrow and asterisk).We thus generated a Cterminally truncated clone, DJC62-R180x, by mutating the arginine of residue 180 to a stop codon, and performed import experiments.After thermolysin treatment of the chloroplasts after import, one protein the same size as the precursor, and two lowermolecular-weight mature proteins were produced (Figure 1E, lanes 3 and 4, pr, m1 and m2).To further confirm the location of these three proteins, chloroplasts after import were treated with trypsin (Figure 1E, lanes 5 to 10).Trypsin is more effective in removing outer membrane proteins but still will not penetrate the inner membrane.The result showed that all three imported proteins were resistant to trypsin (Figure 1E, lane 9) and suggested that they were located inside the inner membrane.They were degraded if the trypsin treatments were performed in the presence of Triton X-100, indicating that their resistance was due to membrane protection, not protein aggregation (Figure 1E, lane 10).The effectiveness of the trypsin treatment was shown by the degradation of the outer membrane protein Toc75 and the resistance of the inner membrane protein Tic110.The presence of the precursor form as one of the imported products indicated that the transit peptide of DJC62 is not always removed after import or the efficiency of processing was low in the in vitro import experiments.In summary, our data show that Arabidopsis has at least 19 J proteins localized in chloroplasts (Figure 1 and Table 1).All 19 have predicted transit peptides.In addition, consistent with the stroma location of the two chloroplast Hsp70s [9,29], none of the J proteins has the bipartite transit-peptide structure typical of thylakoid luminal proteins.

The Nineteen Chloroplast J Proteins in Arabidopsis Can Be Classified into 11 Clades
Among the 19 chloroplast J proteins, four are type-I J proteins.The other 15 are all type-III J proteins.Some of the chloroplast J proteins contain additional known structural domains (Figure 2A).Both DJC31 and DJC62 have two TPR domains.DJC76, DJC77 and DJC82 have an Fd domain.DJA4, DJC69, DJC75 and DJC76 have one or more predicted transmembrane domains.Phylogenetic analysis revealed that the 19 J proteins could be classified into 11 clades, consisting of 4 groups and 7 singletons (Figure 2B).All of the type-I J proteins, DJA4, DJA5, DJA6 and DJA7, fall into one group.The other three groups include the two TPR domain-containing J proteins (DJC31 and DJC62), the three smallest chloroplast J proteins (DJC23, DJC24 and DJC66) and the three Fd domain-containing J proteins (DJC76, DJC77 and DJC82).Other than the type-I J proteins, no homologues are found for the rest of the 10 clades in Arabidopsis.

of the 11 Clades of Chloroplast J Proteins are Conserved in Flowering Plants
To investigate whether the 11 clades of chloroplast J proteins identified in Arabidopsis were conserved in other flowering plants, we performed BLASTP searches against genome databases of rice (Oryza sativa), soybean (Glycine max) and wine grape (Vitis vinifera) using polypeptide sequences of the 19 Arabidopsis chloroplast J proteins as queries.As shown in Table 2, ortholog(s) for all of the 11 clades could be identified except that DJC69 ortholog could not be found in rice.We further searched for DJC69 homolog in other available monocot databases but still did not find DJC69 homologs in any monocots, suggesting that DJC69 has been lost in monocots.
Classification of chloroplast J proteins into 11 clades is still evident when the phylogenetic analysis is expanded from Arabidopsis to the other flowering plants analyzed (Figure S1).The Fd domain-containing J-protein clade (DJC76 clade) can be further divided into three subclades, DJC76, DJC77 and DJC82, when more homologs were included.These results suggest that the 11 clades of chloroplast J proteins identified from Arabidopsis are conserved in flowering plants with the exception that DJC69 homolog does not exist in monocots.

Evolutionary Origins of the Chloroplast J Proteins
We next investigated where the chloroplast J proteins appeared in the evolutionary lineage.Polypeptide sequences of the 19 Arabidopsis chloroplast J proteins were used to perform TBLASTN searches against the genomes of cyanobacterium Synechocystis sp.PCC 6803, glaucophyte Cyanophora paradoxa, 7 green algae (three Ostreococcus, two Micromonas pusilla, Chlamydomonas reinhardtii and Volvox carteri), and two lower land plant species, the moss Physcomitrella patens and lycopod Selaginella moellendorffii.As shown in Table 2 and Table S1, only type-I J-protein homologs exist in cyanobacterium Synechocystis sp.PCC 6803 and glaucophyte Cyanophora paradoxa.Homologs for an additional three clades, the TPR domain-containing DJC31 clade, the Fd domaincontaining DJC76 clade, and the DJC73 clade, are present in all the green algae analyzed (Table S1).Six more clades were found in Physcomitrella and Selaginella, finally the DJC72 clade appears only in flowering plants and the DJC69 clade was then lost in monocots.
Homologs of all six Chlamydomonas chloroplast J proteins previously reported are present in all the land plants analyzed (Table 2), suggesting that functions identified for Chlamydomonas chloroplast Hsp70 and J proteins are most likely preserved in higher plant chloroplasts.These Chlamydomonas chloroplast J proteins comprise three of the four clades we identified: the type-I J proteins (Chlamydomonas CDJ1 and CDJ6), the Fd domaincontaining DJC76 clade (CDJ3 to 5) and the DJC73 clade (CDJ2).Our searches identified additionally locus Cre02.g108800 of Chlamydomonas as encoding a homolog of the TPR domaincontaining DJC31 clade.The protein encoded by Cre02.g108800contains a large, less conserved, N-terminal region, followed by two TPR domains and a C-terminal J domain, resembling Arabidopsis DJC31 and DJC62.All of the green algae analyzed contain the same four clades (Table S1).These results suggest that green algae only possess part of the higher plant chloroplast J protein constituency.However, we cannot exclude the possibility that green algae contain some yet unidentified chloroplast J proteins and these proteins have been lost in higher plants.
Only the Type-I J Protein from Synechocystis is Retained Synechocystis sp.PCC 6803 genome has 7 genes encoding J proteins (Table S2) [30].Our search results indicate that only the type-I J protein in Arabidopsis chloroplasts were derived from the cyanobacterium.To confirm that all other Synechocystis J proteins were not retained, we used the polypeptide sequences of these 7 J proteins to perform BLASTP searches against the Arabidopsis genome.Only the type-I J protein of Synechocystis sp.PCC 6803, Sll0897, has homologs in Arabidopsis (Table 1).Although sll1384 encodes a protein that is predicted to contain one TRP domain, the protein has no sequence similarity to Arabidopsis DJC31 or DJC62 and its J domain was located N terminal to the TPR domain, rather than C terminal as in Arabidopsis DJC31 and DJC62.Therefore it is unlikely that Arabidopsis DJC31 and DJC62 have evolved from Sll1384.To further confirm that the 6 Synechocystis J proteins have been lost, the 6 Synechocystis J proteins were searched against the glaucophyte Cyanophora paradoxa genome.Glaucophytes have chloroplasts with cyanobacterial appearance [31,32], and have been shown to be the earliest divergence in Plantae [33].Again our results showed that only the type-I J protein homologs were found in the Cyanophora paradoxa genome and no homologs were found for the other 6 Synechocystis J proteins.Thus, our result suggests that during the endosymbiotic process, only type-I J protein was retained, and the other six cyanobacterial J proteins were no longer retained, at least since glaucophytes.

The Seven Land Plant-Specific J-Protein Clades May Serve Different Functions
All the 7 clades of land plant-specific chloroplast J proteins are relatively small with no additional known domains other than the J domain (Figure 2A, names labeled in blue).We were interested in knowing whether they still contain some clade-specific motifs that suggest they play different functions.Their sequences from six land plant species were aligned.If a species has multiple family members for a particular clade, a representative gene was selected.As shown in Figures 3 and 4, all 7 clades have at least one cladespecific highly conserved motif in addition to the J domain.All, except DJC72, have the clade-specific motifs located close to the C terminus (Figure 3).In agreement with the fact that it is the newest addition since flowering plants, sequences of DJC72 from different species are similar across the entire polypeptide even in the transit peptide regions.The high degree of sequence identity across different species in the clade-specific motif suggests that the motif has conserved function from moss to flowering plants, and is most likely used to interact with a specific client protein or a protein that recruits the J protein to a specific location within chloroplasts.These highly conserved motifs are clade-specific, suggesting that each clade interacts with a different protein.In the DJC22 clade, rice does not seem to share the conserved C-terminal motif found in Physcomitrella, Selaginella and the dicot plants (Figure 4).However, when we retrieved DJC22 homologs from other monocots, we found that the C-terminal domain of all monocot DJC22s is highly conserved (Figure 4).It is possible that the client protein for DJC22 is conserved from Physcomitrella and Selaginella to dicots.In monocots, the structure of the client protein, and thus the client-recognition motif in DJC22, may have further evolved to adapt to some monocot-specific physiology.

Gene Expression Patterns of Chloroplast J Proteins in Arabidopsis
As a first step toward investigating the functions of the chloroplast J proteins, gene expression levels of the Arabidopsis chloroplast J proteins in various tissues was retrieved from the public Affymetrix microarray database using Genevestigator (www.genevestigator.com)[34].DJC72 and DJC82 do not yet have probe sets in the Affymetrix gene chips and thus were not included in the analyses.As shown in Figure 5, in general, land plant-specific chloroplast J proteins show larger variations in expression levels in different tissues, suggesting that they have some tissue-specific functions or that their amount needs to be adjusted according to plastid types.The two Fd domaincontaining J proteins, DJC76 and DJC77, although not land plant-specific, also show variable expression in different tissues with the highest expression in leaves.It is likely that they have preserved their function from algal chloroplasts and thus mostly function in chloroplasts of leaves.DJC75 (also named CRRJ or NdhT) is land plant-specific and is expressed almost exclusively in green tissues, suggesting that it has a function unique to land-plant chloroplasts.Indeed DJC75 is essential for the activity of the NDH complex functioning in cyclic electron transport [21].NDH complex is not found in green algae [23].DJC23 has the highest expression level among all Arabidopsis chloroplast J proteins.Interestingly, the other two members of the same clade, DJC24 and DJC66, are expressed predominantly in flowers and roots respectively, two highly specialized organs of higher plants.These expression patterns suggest that this clade may have some functions that require tissue-specific adjustment of different isoforms.Other J proteins with homologs present in green algae, including the four type-I J proteins, the two TPR domaincontaining J proteins (DJC31 and DJC62) and DJC73, all show more uniform and lower expression levels in the major tissues we selected, suggesting that they serve some primordial constitutive functions in all plastid types.

Regulation of cpHsc70 and J-Protein Gene Expression by Abiotic Stresses
As sessile organisms, plants have to adapt to the environmental changes.To determine if some of the chloroplast J proteins are involved in the adaptation, expression patterns of chloroplast Jprotein and the two cpHsc70 genes under abiotic stresses, such as heat, cold, drought, osmotic and salt stresses, were analyzed using Genevestigator.As show in Figure 6, when the filters for fold change and p-value were set to |2| and ,0.05, respectively, heat becomes the only stress that can up-regulate the expression of the cpHsc70 genes in multiple experiments.DJA6 and DJC66 were also up-regulated by heat stress in multiple experiments, suggesting that DJA6 and DJC66 may function together with cpHsc70s in thermotolerance.DJA4, DJA5, DJC23 and DJC66 were upregulated by cold stress.They may recruit cpHsc70 to assist the folding of some cold-labile proteins [35].DJC77, and many land Names of land plant-specific J proteins are written in blue.J, J domain; Gly/Phe-rich, glycine/phenylalanine-rich domain; Zn 2+ finger, zinc-finger domain; C-terminal, C-terminal domain; Fd, ferredoxin domain; TRP, tetratricopeptide repeat domain.(B) Phylogenetic relationship of the 19 Arabidopsis chloroplast J proteins.A neighbor-joining tree was constructed using multiple sequence alignments from peptide sequences of the 19 chloroplast J proteins.Bootstrap analyses were computed with 1,000 replicates, and the values of percentage larger than 85 are shown on the branches.Grouping of four clades (shown with gray background) had at least 85% bootstrap support.doi:10.1371/journal.pone.0070384.g002Table 2. Homologs of Arabidopsis chloroplast J protein in other plant species.rice DJC82 homolog, OsDJC82 (accession: BK008487), was re-annotated according to the alignment with DJC82 homologs from other plant species (Figure S3).c Selaginella DJC22 homolog, SmDJC22 (accession: BK008488), was re-annotated according to the alignment with DJC22 homologs from s other plant species (Figure 4).d Selaginella DJC23 homolog, SmDJC23 (accession: BK008489), was re-annotated according to the alignment with DJC23 homologs from other plant species (Figure 3).e Selaginella DJC73 homolog, SmDJC73 (accession: BK008490), was re-annotated according to the alignment with DJC73 homologs from other plant species (Figure S4).
Note: Homolog shown in boldface is the representative from each species for the 7 clades of land plant-specific chloroplast J protein aligned in Figures 3 and 4. doi:10.1371/journal.pone.0070384.t002 plant-specific J-protein genes, such as DJC24, DJC22, DJC26, DJC65, DJC69 and DJC75, were down-regulated by cold stress.DJC76 is the only gene regulated by salt stress.Drought and osmotic stresses do not seem to have a significant effect on the expression of chloroplast J-protein genes since no gene was affected significantly in multiple experiments.

Discussion
We show here that at least 19 Arabidopsis J proteins are localized in chloroplasts.Since our starting pool was J proteins with a predicted chloroplast-targeting transit peptide and J proteins identified in various plastid proteomes, J proteins with non-canonical chloroplast targeting signals and present in very low amounts would be missed in this analysis.Thus the exact number of chloroplast J proteins can be expected to be even higher.However 19 J proteins already make chloroplasts the organelle with the highest number of confirmed J proteins.
Our analyses indicate that three new clades of J proteins were added since green algae.New J proteins may have evolved to recruit cpHsc70 to new functions, or to processes already present in lower organisms but not yet involving Hsp70.For example, CDJ2 of Chlamydomonas has been suggested to assist the assembly and disassembly of the vesicle-inducing protein 1 (VIPP1) oligomers [15], which is also present in cyanobacteria [36].Since there is no CDJ2 homolog in cyanobacteria (Table 2), VIPP1 may work with proteins other than Hsp70 and J proteins in cyanobacteria.
Our comparison of the numbers of J proteins in Chlamydomonas and Arabidopsis chloroplasts (Table 3) suggest that when the number of chloroplast Hsp70-encoding genes increased from one to at least two in the transition from green algae to land plants, the number of genes encoding chloroplast J proteins also have more than doubled by adding seven new clades with novel clade-specific motifs.It will be very interesting to identify the new client proteins interacting with these new J proteins.Surprisingly, we had expected the land plant-specific J proteins to function in stressrelated conditions like drought, which would be expected to affect land plants more severely than algae.However, gene expression analyses suggest that only DJC66 may be involved in heat and cold stress tolerance.Most land plant-specific J proteins are down regulated by cold treatments.No chloroplast J proteins could be suggested to be involved in drought tolerance, suggesting that drought tolerance may not involve cpHsc70.Down-regulations of many land plant-specific J proteins by cold stress were observed in the late response (after cold treatment $24 hours), not in the early stage (1-hour cold treatment) (Figure 6).It has been shown in coldresponsive transcriptome analyses, the majority of genes downregulated by cold stress in the late response are genes related to metabolism [37].Hence it is possible that chloroplast J proteins down regulated by cold treatments may function in folding and assembly of proteins and protein complexes involved in metabolism.
In prokaryotes, yeast and animals, the ratio between the number of J protein-encoding genes and the number of Hsp70encoding genes is about 2, and this ratio has increased to around 9 since Chlamydomonas and has remained similar in all higher plants (Table 3, [10,16,30,[38][39][40][41]).The incorporation of a cyanobacterium as a new organelle may have incurred many new levels of coordination, for example, regulation of the expression of nuclear genes encoding chloroplast proteins, photorespiration, and other crosstalks between chloroplasts and other organelles.These additional needs may have been accommodated by fusing the J domain to a motif that can bind new client proteins and then tapping into the abundant Hsp70 system for protein homeostasis maintenance and protection.
All the J proteins, except the type-I J protein, from cyanobacteria have been lost during the endosymbiotic process.This Figure 3. Land plant specific chloroplast J proteins contain highly conserved clade-specific motifs.Sequence alignments of 6 of the land plant-specific chloroplast J-protein clades.Representative genes (bold typed in Table 2) from each species were aligned.The J domain is underlined with a blue line.The position of the HPD tripeptide is indicated.doi:10.1371/journal.pone.0070384.g0032) from each species, and homologs from maize (GRMZM2G086841) and sorghum (Sb04g025270) were aligned.The J domain is underlined with a blue line.The position of the HPD tripeptide is indicated.doi:10.1371/journal.pone.0070384.g004suggests that type-I J proteins probably play some primordial role, like assisting in thermotolerance.Functions played by cyanobacterial type-II and type-III J proteins may no longer be required by chloroplasts as intracellular organelles.Type-II J proteins may also have been lost because the functions of type-I and type-II J proteins are partially redundant [30,42].The identification of 19 Arabidopsis chloroplast J proteins and the analyses of their evolutionary origins will provide useful leads for finding J proteins assisting chloroplast Hsp70 in specific functions.

Plant Materials, Chloroplast Isolation and Protein Import into Chloroplasts
Pea (Pisum sativum cv.Little Marvel) seedlings were grown as described [19], and 6-to 9-day-old seedlings were harvested for chloroplast isolation.Chloroplast isolation, protein import into isolated chloroplasts and thermolysin treatment of chloroplasts after import were performed as described [43].Import reactions were performed at room temperature for 25 min with 3 mM ATP. Trypsin treatment of chloroplasts after import was performed as described

Sequence Analysis and Construction of Phylogenetic Tree
To search for homologs of Arabidopsis chloroplast J proteins in rice (Oryza sativa; GenBank v. 171), soybean (Glycine max; GenBank v. 181) and grape (Vitis vinifera; GenBank v. 179), reciprocal BLASTP searches were performed.In the first query, protein sequences of the 19 chloroplast J proteins were used as queries to search the databases in PlantGDB (http://www.plantgdb.org/)and Phytozome (v.7.0; http://www.phytozome.net/),which collect and reorganize the most updated sequence information for most of the sequenced plant species.The E-value for BLAST was set to 1e-10.Sequences of putative homologs were retrieved and used as the second query to perform BLASTP searches against the Arabidopsis database.If the most similar homolog in Arabidopsis was exactly the Arabidopsis J-protein sequence used as the first query, the sequence of the second query was selected as a homolog.BLASTP results obtained from PlantGDB and Phytozome are the same.Locus names of homologs obtained from PlantGDB are shown in Table 2. Homologs for DJC31 and DJC82 were originally not found in rice, but were found in sorghum and maize and their exon-intron junctions are conserved with the Arabidopsis homologs.We therefore compared the genomic sequence of rice to other higher plants and re-annotated the sequences of the rice homologs for these two J proteins.The peptide sequences of the re-annotated rice DJC31 homolog (original annotation: LOC_Os05g31056/LOC_Os05g31062) and rice DJC82 homolog (original annotation: LO-C_Os05g33010) are shown in Figures S2 and S3.The reannotated rice DJC31 and DJC82 homologs are highly similar to homologs from other plants.To search for chloroplast J-protein homologs in lower plant species, TBLASTN was performed using databases in PlantGDB and Phytozome for Volvox carteri, Chlamydomonas reinhardtii, Physcomitrella patens and Selaginella moellendorffii, Cyanophora genome project (http://cyanophora.rutgers.edu/cyanophora/home.php) for glaucophyte Cyanophora paradoxa, and NCBI/BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) for the others.Homologs were identified using the method described above.In Selaginella, protein sequences annotated for 3 J-protein homologs located at loci g8923, g100874 and g8874 were short.Through comparison of their genomic sequences with other species, their sequences were re-annotated.The peptide sequences of re-annotated Selaginella DJC22 (original annotation: g8923) and DJC23 (original annotation: g100874) homologs are shown in Figures 3 and 4. The peptide sequence of re-annotated Selaginella DJC73 homolog (original annotation: g8874) is shown in Figure S4.Multiple sequence alignments were performed using the BLOSUM protein weight matrix and visualized by GeneDoc (v.2.5; http://www.nrbsc.org/gfx/genedoc/).Phylogenetic trees were constructed using the neighbor-joining method employed by the ClustalX program [49] and visualized by NJplot (v.2.3; http://pbil.univ-lyon1.fr/software/njplot.html) or TreeView (v.1.6.6;http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).Domain structures are predicted using InterPro (http://www.ebi.ac.uk/interpro/).

Expression Pattern Analyses
Tissue-specific expression pattern and stress-regulated gene expression ratios were retrieved from the public Affymetrix microarray database using Genevestigator (www.genevestigator.com).
Nucleotide sequence data of OsDJC31, OsDJC82, SmDJC22, SmDJC23 and SmDJC73 reported in this work are available in the Third Party Annotation Section of the DDBJ/EMBL/ GenBank databases under the accession numbers TPA: BK008486-BK008490.

Figure 1 .
Figure 1.Nineteen J proteins were imported into chloroplasts.(A) Fifteen J proteins that were imported into chloroplasts.(B) Chloroplast import of DJC72MM, DJC73MM, DJC31 and DJC62.(C) DJC42 and DJC78MM were not imported into chloroplasts.(D) Chloroplast import of DJC31-G184x.(E) Chloroplast import of DJC62-R180x.In vitro-translated [ 35S]Met-labeled precursor proteins were incubated with isolated pea chloroplasts under import conditions.Part of the precursor proteins without import (Ivt) or re-isolated intact chloroplasts after import (Chpt) were further treated with thermolysin (Th) or trypsin.Sample for lanes 7 and 10 of (E) were trypsin treated in the presence of 0.1% of Triton X-100.Samples were analyzed by SDS-PAGE followed by fluorography.For each precursor panel, the Ivt lanes (with and without protease treatment) contain the same amount of precursor, and the Chpt lanes contain the same amount of proteins.The Ivt lanes contained 0.9% of the in vitro-translated proteins added to the import reactions shown in the Chpt lanes, except for DJC62, in which the Ivt lanes represent 0.4% of in vitro-translated proteins added to the import reactions.pr, precursor form; m, mature form.m1 and m2, different-sized mature proteins produced after the import of DJC31-G184x and DJC62-R180x.Samples from lane 8 to 10 in (E) were also analyzed by immunoblotting for Toc75 and Tic110 as controls.doi:10.1371/journal.pone.0070384.g001

Figure 2 .
Figure 2. Domain structure and phylogenetic tree of the 19 Arabidopsis chloroplast J proteins.(A) Schematic representations of domain structures of the 19 Arabidopsis chloroplast J proteins.Each bar is drawn to scale according to the number of amino acids, shown at the bottom.Names of land plant-specific J proteins are written in blue.J, J domain; Gly/Phe-rich, glycine/phenylalanine-rich domain; Zn 2+ finger, zinc-finger domain; C-terminal, C-terminal domain; Fd, ferredoxin domain; TRP, tetratricopeptide repeat domain.(B) Phylogenetic relationship of the 19 Arabidopsis chloroplast J proteins.A neighbor-joining tree was constructed using multiple sequence alignments from peptide sequences of the 19 chloroplast J proteins.Bootstrap analyses were computed with 1,000 replicates, and the values of percentage larger than 85 are shown on the branches.Grouping of four clades (shown with gray background) had at least 85% bootstrap support.doi:10.1371/journal.pone.0070384.g002 homolog, OsDJC31 (accession: BK008486), covering previously annotated LOC_Os05g31056 and LOC_Os05g31062, was re-annotated according to the alignment with DJC31 homologs from Arabidopsis and other monocots (FigureS2). b

Figure 4 .
Figure 4. DJC22 proteins from monocots have distinct clade-specific motif from dicots and lower land plants.Sequence alignments of DJC22 homologs from dicot plants, Selaginella and Physcomitrella, and sequence alignments of DJC22 homologs from monocot plants.Representative genes (bold typed in Table2) from each species, and homologs from maize (GRMZM2G086841) and sorghum (Sb04g025270) were aligned.The J domain is underlined with a blue line.The position of the HPD tripeptide is indicated.doi:10.1371/journal.pone.0070384.g004

Figure 5 .Figure 6 .
Figure 5. Tissue-specific expression patterns of chloroplast J-protein genes in Arabidopsis.Expression of 17 chloroplast J-protein genes and two cpHsc70s genes in various tissues was retrieved using Genevestigator.Mean values of expression level after normalization were plotted.Land plant-specific J proteins are indicated.doi:10.1371/journal.pone.0070384.g005

Table 1 .
, DJC31, arrow) was extremely close in size to the precursor protein.If this protein is the mature Putative chloroplast J proteins of Arabidopsis analyzed in this work.

Table 3 .
Number of Hsp70 and J-protein homologs in various organisms, and in chloroplasts of Arabidopsis and Chlamydomonas.