Phylogenetic analyses based on an Hsp70 amino acid alignment reliably assigned the detected P. caudatum hsp70 genes to the respective Hsp70-subfamilies cytosol (CY), endoplasmic reticulum (ER) and mitochondria (MT) (Figure 2). These analyses further showed not only that these homologs can be divided into the three subfamilies, but they also revealed the existence of five putative Hsp70-groups in P. caudatum: one MT group consisting of only one homolog, but two distinct groups within the CY-, as well as two groups within the ER-type subfamily. Here, each CY- and ER-type Hsp70-group shows a closer relationship to putative orthologous sequences of P. tetraurelia than to the other P. caudatum Hsp70-group of the same subfamily (Figure 2). These findings suggest a gene duplication event before the speciation of P. caudatum and P. tetraurelia, but obviously after the divergence of Paramecium and Tetrahymena (another closely related oligohymenophorean ciliate) since both Paramecium CY- and ER-type homologs form monophyletic clades (see Figure 2). Therefore, the last common ancestor of Paramecium and Tetrahymena should have had one functional CY- and one ER-type hsp70 homolog, while the common ancestor of P. caudatum and P. tetraurelia has possessed one functional gene of each of the four non-organellar Hsp70-groups, meaning one CY-A, CY-B, ER-A and ER-B homolog.
Using an RT-qPCR approach, we could show considerable differences among the five assigned Hsp70-groups of Paramecium caudatum. These analyses revealed the group CY-A as the major constitutively expressed Hsp70-group in P. caudatum at optimum physiological temperatures. Even though this study showed a significant up-regulation in mRNA levels of the CY-A group members after heat shock (~3.0-fold), this does not necessarily cause the rejection of their Hsc70 (heat-shock cognate protein 70) affiliation, since many constitutively expressed hsp70s can be induced under specific conditions (e.g. [41–44]). We have also seen that the Hsp70-group CY-B holds a very low mRNA abundance at optimum growth temperatures, but was highly induced after heat shock (~84-fold). This induction resulted in the second most abundant Hsp70s during heat stress and suggests that the two genes PcHsp70Cy2a and PcHsp70Cy2b of group CY-B represent the major heat-inducible homologs in P. caudatum.
As previously mentioned, the P. caudatum Hsp70-groups CY-A and CY-B showed a closer relationship to orthologous Hsp70 sequences of P. tetraurelia than to each other. The comparative analysis of two P. tetraurelia EST libraries (constructed from RNA of cells grown at 27°C or 35°C) revealed comparable expression patterns between the P. caudatum and the P. tetraurelia Hsp70-groups (cf. Figure 3A and 3B). Here, the CY-A group homologs are also highly expressed under normal physiological conditions, indicating constitutively expressed hsp70 genes, while the CY-B related homologs appear to represent the major heat-inducible hsp70s in P. tetraurelia as well. This result suggests conserved functions and a general expression pattern of the Paramecium Hsp70 gene family, at least between P. caudatum and P. tetraurelia. In this context it is interesting to note that Tetrahymena features also constitutively expressed and heat-inducible CY-hsp70s [32, 45], which form together with the single Ichthyophthirius multifiliis CY-hsp70 homolog a sister clade to all Paramecium CY-homologs (Figure 2). In this cytosolic Ichthyophthirius–Tetrahymena clade, the constitutively expressed T. thermophila hsp70-4 gene clusters together with orthologs of T. malaccensis, T. borealis and I. multifiliis by forming a sister clade to all other Tetrahymena CY-homologs including the heat-inducible T. thermophila hsp70-2 gene. Furthermore, while heat-inducible and constitutively expressed CY-hsp70s are common in higher eukaryotes too, they obviously do not form clear separated clades (cf. Additional file 4: Figure S2). In conjunction with our findings for ciliate hsp70s, this strongly indicates that heat-inducibility in cytosolic Hsp70s evolved several times independently during eukaryote evolution; at least twice within closely related unicellular eukaryotes and within metazoans. This result provides a striking example of convergent evolution during subfunctionalization among eukaryotes.
Our relative expression analyses also demonstrated that the P. caudatum Hsp70-groups ER-A and ER-B showed comparatively similar transcription levels at optimum temperatures, but the relative amount was considerably lower compared to the major constitutively expressed Hsp70-group CY-A. Heat treating the P. caudatum cells significantly induced the mRNA expression of both ER-type groups, but only to a comparatively small amount of 2.0-fold for ER-A and 2.7-fold for ER-B. Our study, therefore, partially supports the findings of Hori and Fujishima , who showed only trace amounts of ER-type hsp70 mRNA when P. caudatum cells were cultured at 25°C, but also an up-regulation of approx. 4-fold when cells were heat shocked at 35°C. On the other hand, they suggested a predominant expression of ER-type hsp70s because they could not detect cytosolic homologs . This is in contrast to our results showing that the cytosolic Hsp70-group CY-A encompassed the major constitutively expressed hsp70s, while group CY-B covered the major inducible homologs. Further, the relative mRNA levels of the ER-groups were nearly 23% less of the relative cytosolic mRNA transcript amount at optimum temperatures. And after heat stress, the cytosolic hsp70 transcription level was about 2.5-fold higher than that of all ER-type homologs, mainly because of the striking induction of the CY-B group, but also due to the significant up-regulation of the CY-A group hsp70s. Nevertheless, studies on other eukaryotic organisms revealed that ER-type Hsp70 proteins, mostly designated as GRP78 or BiP, are abundant proteins in animal cells as well . They are primarily induced by glucose depletion , and also due to cadmium exposure , during apoptosis  or due to the presence of unfolded or misfolded proteins in the ER. However, they are not significantly heat-inducible , while P. caudatum ER-type hsp70s seem to be. Therefore, experiments on the gene expression of these ER-type hsp70 genes investigating the effect of endosymbionts [46, 52] or other stressors would be valuable in unveiling their role in the stress response of Paramecium.
In this study, we further identified one hsp70 gene encoding for a mitochondrial Hsp70 protein (mtHsp70). These proteins are required for the translocation of cytosolic preproteins across the mitochondrial membrane as well as the subsequent folding reactions in the mitochondrial matrix [53, 54]. In most organisms, organelle-specific Hsp70s are generally encoded by a single gene , but in Saccharomyces cerevisiae three distinct mtHsp70 genes are described, and P. tetraurelia seems to have at least two functional mthsp70s. Humans seemingly hold only one chaperone-active mtHsp70 protein (HSPA9B), while another highly similar protein (HSPA9A) plays a major role in the import of preproteins into the mitochondria. The single mtHsp70 gene PcHsp70MT1a that we have observed in P. caudatum showed a constitutive expression pattern (Figure 3A), suggesting its essential role for mitochondria to function properly under normal physiological conditions. The 1.7-fold up-regulation in mRNA levels after heat shock, which is comparable to S. cerevisiae or Blastocladiella emersonii mtHsp70s [27, 53], implies also its key role for proper mitochondrial function under heat stress. Though elevated temperatures can lead to enhanced oxidative stress, this might be overcome by an increased protein import into the mitochondria or an enhanced chaperone activity. Whether the mthsp70 gene in P. caudatum covers both chaperone- and preprotein translocation activity and whether the P. tetraurelia mthsp70s have undergone subfunctionalization after gene duplication should be determined by future studies addressing Hsp70 protein interactions.
The phylogenetic and evolutionary patterns detected within the Paramecium Hsp70 gene family support the model of divergent evolution, while the different Hsp70-groups seem to have evolved under functional and structural constraints. On the other hand, in relation to the further investigated ciliate species and metazoans, these patterns indicate distinct evolutionary pathways and convergent evolution (cf. Figure 2). Therefore, this multigene family seems to be differentially affected also in ciliates by gene duplication events and/or whole genome duplications, by gene loss and retention, subfunctionalization and/or pseudogenization. The finding of a similar basal gene duplication pattern of the CY- and the ER-type hsp70s in Paramecium, for example, is indicative of a whole-genome duplication (WGD) event within the last common ancestor of P. caudatum and P. tetraurelia. It also may point to an intermediary WGD that occurred within the genus Paramecium, because the most recent WGD is supposed to coincide with rapid speciation events that gave rise to the P. ‘aurelia’ complex (cf. Figure five in ). Further, an old WGD is thought to have occurred before the divergence of Paramecium and Tetrahymena, but is still under question due to recent studies that failed to detect evidence for WGD in T. thermophila and on the Ichthyophthirius/Tetrahymena branch [56, 57].
The age of these WGD events would be of major significance to clarify speciation and radiation events in ciliates as well as to understand the dynamics of gene loss and retention or the metabolic adaptation after a WGD . For example, the dates of these duplications could be interesting in consideration of the phylogenetic relationships among the genus Paramecium, whether these events occurred within the last common Paramecium ancestor or before/after certain speciation events. Insights into the different scenarios would be of importance to potentially support the proposed subdivision of the genus Paramecium into the four subgenera Chloroparamecium, Helianter, Cypriostomum and Paramecium. Here, further analyses of the Hsp70 multigene family of at least three additional Paramecium species (e.g., Paramecium [Chloroparamecium] bursaria, Paramecium [Helianter] putrinum and Paramecium [Cypriostomum] calkinsi) would be required. However, due to a missing reliable molecular clock for ciliate divergence times (e.g. approx. 1500 to 600 Ma for Paramecium/Tetrahymena divergence [60, 61]) and recent findings on P. tetraurelia that indicate extraordinarily reduced DNA mutation rates , more genome data would be favourable to estimate the dates of Paramecium species divergence or the timing of WGD events in the ciliate phylum.
Assuming a WGD event within the last common ancestor of P. caudatum and P. tetraurelia, the lack of MT paralogs in P. caudatum has to be the result of pseudogenization and a rapid gene loss. Even though many of the duplicates of the recent WGD in P. tetraurelia are functionally redundant and are supposed to be progressively lost, most of the gene duplicates did not go through such a rapid elimination . On the other hand, the four putative mthsp70s in P. tetraurelia, which seem to comprise two functional genes and two pseudogenes, are suggestive of an intermediary WGD event causing the Hsp70-group differentiation in Paramecium. Since our study on P. caudatum was based on cDNA and all the detected homologs show typical Hsp70 family signatures and motifs as well as no premature stop codons or frameshifts, they can be considered as functional hsp70 genes. Therefore, we have not detected either potentially expressed or unexpressed hsp70 pseudogenes, which would represent a hallmark of the birth-and-death process after duplication events. Hence, to disentangle in more detail the different patterns of concerted and non-concerted evolution that shaped the P. caudatum Hsp70 gene family, further gDNA/genome analyses would be necessary to unveil potential hsp70 pseudogenes in P. caudatum. Moreover, comparative genome analyses can uncover the gene order and orientation of recent hsp70s to infer pre-duplication states and to predict patterns of duplicate loss and retention. Such analyses can further improve our understanding of the evolution of such gene families in Paramecium and other microbial eukaryotes.