AtTPR7 is a chaperone-docking protein of the Sec translocon in Arabidopsis

Most proteins localised to cellular compartments are translated in the cytosol and have to be targeted to the destined compartments where they are translocated post- or co-translationally across the organellar membranes (Schleiff and Becker, 2011). Post-translational import occurs predominantly into the endosymbiotic organelles chloroplasts and mitochondria, whereas proteins of the endoplasmic reticulum (ER) are imported either co- or post-translationally. Despite the co-translational pathway being the predominant one in mammals, yeast utilises both pathways equally (Rapoport, 2007; Wang et al., 2010; Zimmermann et al., 2011). In both cases the preproteins are transported across the membrane through the channel protein Sec61. During post-translational translocation in mammals the Sec61 channel associates with additional components, the Sec62/63 complex and the luminal HSP70 chaperone BiP, which assists in translocating the peptide across the membrane (Osborne et al., 2005; Zimmermann et al., 2006). For both pathways cytosolic components are indispensable. During co-translational transport the nascent polypeptide emerging from the ribosome in the cytosol is recognised at a signal or transmembrane sequence by the signal recognition particle. This complex associates first with the signal recognition particle receptor at the ER membrane and subsequently with the Sec translocation channel. The post-translational pathway is predominantly used by more hydrophobic proteins (Ng et al., 1996) and thus requires the assistance of molecular chaperones to prevent molecular crowding and to facilitate interaction with the translocon complex. Members of the HSP70 and HSP40 chaperone family have been described to aid the translocation of ER localised proteins (Ngosuwan et al., 2003; Zimmermann et al., 1988). Likewise, the insertion of tail-anchored ER membrane proteins is facilitated by HSP70 or by components of the so called ‘guided entry for tail anchored proteins’ (GET) pathway (Abell et al., 2007; Wang et al., 2010). HSP70 not only participates in ER protein translocation but also interacts with preproteins targeted to chloroplasts and mitochondria (Young et al., 2003; Zhang and Glaser, 2002). HSP70 can either function alone or in concert with HSP90. HSP90 and HSP70 together with cochaperones associate with a large population of chloroplast preproteins as well as with hydrophobic carrier proteins destined to the inner mitochondrial membrane in mammals (Fan et al., 2006; Fellerer et al., 2011; Qbadou et al., 2006; Young et al., 2003; Zara et al., 2009).

Preproteins can be recognised at the membrane surfaces either directly by receptor proteins or indirectly via their bound chaperones. Remarkably, all chaperone recognising docking proteins contain one or more tetratricopeptide repeat (TPR) motif facing the cytosol. These docking proteins are found throughout eukaryotic organisms and are distributed among all cellular compartments, suggesting a general mechanism for preprotein recognition through HSP90 and HSP70 (Kriechbaumer et al., 2012; Schlegel et al., 2007). TPR domains mediating interaction with chaperones, so called clamp-type TPR motifs, generally consist of at least three tandemly arranged TPR motifs which are highly degenerate 34 amino acid repeats without any strictly conserved residues. Proteins containing three TPR repeats are organised in a right-handed super helical structure capped by a so called solvation helix at the C-terminal end. This helix turn helix motif is packed into a regular series of antiparallel alpha helices. They form a dicarboxylate clamp coordinating the conserved aspartate residue in the C-terminus of HSP90 and HSP70 (D'Andrea and Regan, 2003; Scheufler et al., 2000). In addition to the chaperone-mediated pathway, TPR domains of some receptor proteins facilitate direct interaction with preproteins. Tom20 of the yeast and mammalian mitochondrial translocon binds to N-terminal presequences with its clamp-type TPR domain (Abe et al., 2000; Saitoh et al., 2007). Pex5, a component found in the peroxisomal import apparatus of plants, animals and fungi contains seven non-clamp-type TPR domains allowing recognition of the peroxisomal C-terminal targeting signal (Erdmann and Schliebs, 2005; Gatto et al., 2000; Lee et al., 2006). Tom70 is another well described TPR docking protein in the outer membrane of yeast and mammalian mitochondria containing seven TPR motifs and interacts with both, chaperones and preproteins. The C-terminal TPR motifs of Tom70 directly bind to internal targeting signals of mitochondrial preproteins (Young et al., 2003), whereas the N-terminal motifs interact with chaperone–preprotein complexes. In yeast, preproteins associated with HSP70 are recognised by Tom70, whereas mammalian Tom70 interacts additionally with HSP90–preprotein complexes (Young et al., 2003). Tom34 is an additional component in mammalian mitochondria and has recently been shown to associate with HSP70– and HSP90–preprotein complexes (Faou and Hoogenraad, 2012). In plant mitochondria Tom70 may functionally be replaced by OM64 (Chew et al., 2004; Lister et al., 2007) which is closely related to the chloroplast outer envelope TPR domain containing protein Toc64. Toc64 likewise mediates interaction with HSP90-bound chloroplast preproteins in the cytosol and is associated with the Toc translocon (Qbadou et al., 2006). A TPR containing component associated to the ER membrane, Sec72, which is involved in post-translational import of some ER proteins, has so far only been identified in yeast (Fang and Green, 1994; Feldheim and Schekman, 1994; Harada et al., 2011). Sec72 does not contain a transmembrane domain, but is anchored to the ER membrane by the integral membrane protein Sec71 and both associate with Sec62/63 and the channel protein Sec61 (Harada et al., 2011) to form a translocon complex for post-translational protein import. However, no functional homologues of Sec72 or Sec71 have been identified in higher eukaryotes so far.

The mechanisms of ER protein import in plants have not been analysed in great detail up to date, although the major components of the Sec translocon show a high homology to their mammalian and yeast counterparts. The Arabidopsis thaliana orthologue of the channel protein Sec61α is present as three isoforms and orthologues of the two associated membrane compounds Sec62 and Sec63 are also found in the Arabidopsis genome. AtSec62 is a single copy gene, whereas two isoforms of the J-domain containing AtSec63 exist, which are referred to as AtERdj2A and AtERdj2B (Yamamoto et al., 2008). However, neither co- nor post-translational import into plant ER has been investigated so far.

In this study we describe a TPR domain-containing plant-specific protein, termed AtTPR7 (At5g21990), which was previously identified by Prasad et al. (Prasad et al., 2010). The protein was previously described as a protein of the outer envelope in chloroplast [OEP61 (von Loeffelholz et al., 2011)] and proposed to function as a chaperone receptor for preproteins. However, we could show that the protein resides most likely exclusively in the ER membrane. AtTPR7 associates with AtERdj2 in membrane complexes of 140 and 200 kDa and it specifically interacts with the HSP70 and HSP90 chaperones via its TPR domain. Moreover, we could functionally complement a yeast Δsec71 mutant which is deficient in post-translational protein translocation. We thus propose AtTPR7 to function as chaperone docking protein with a possible role during post-translational protein translocation into the ER in Arabidopsis.

A search for tail-anchored TPR domain containing proteins identified AtTPR7 as a plant specific TPR protein. The TPR motif of AtTPR7 is a carboxylate clamp-type domain and consists of a triple repeat of 34 amino acid long TPR degenerate consensus motif, which is located near the N-terminus (amino acids 104–212; Fig. 1A). AtTPR7 further contains a predicted hydrophobic transmembrane region at its C-terminus reaching from amino acids 531–551 (Fig. 1A; supplementary material Fig. S1) and probably spans the membrane once, which classifies it as a so called tail-anchored protein, found in many cellular compartments. Blast searches against the genomes of various organisms revealed that AtTPR7 homologues are represented in the unicellular green algae Chlamydomonas as well as in mosses, ferns, mono- and dicotyledonous plants, but are absent from bacteria, yeast and mammals. A complete sequence alignment of AtTPR7 homologues in land plants and algae showing the conserved TPR repeat domain as well as the conserved C-terminus containing the transmembrane domain is presented in supplementary material Fig. S1. AtTPR7 was also recently identified in an in silico search for potential HSP90/HSP70 interactors in plants (Prasad et al., 2010). An alignment comparing the TPR domain structure of AtTPR7 in plants with the three triple TPR repeats of the yeast HSP90 cochaperone Sti1/HOP is shown in Fig. 1B. Sti1 is known to bind selectively to HSP90 as well as HSP70 via its individual TPR domains (Schmid et al., 2012). TPR1 of Sti1 is mostly responsible for HSP70 binding, whereas TPR2a associates preferably with HSP90. The third domain TPR2b can bind to both, HSP90 and HSP70. Since the amino acid sequence of the TPR domain of AtTPR7 is 25% identical to TPR1 as well as to TPR2a, AtTPR7 likely binds both chaperones.

We have investigated the localisation of AtTPR7 using expression of a GFP fusion protein as well as cell fractionation. Since AtTPR7 does not contain a predicted N-terminal signal sequence its C-terminal transmembrane domain most likely functions as a targeting signal, as is common for tail-anchored proteins (Borgese et al., 2001; Hegde et al., 2007). In order to prevent masking of the transmembrane anchor by GFP we generated a N-terminal GFP-AtTPR7 construct using full-length AtTPR7.

This construct as well as constructs encoding marker proteins for ER and Golgi fused to mCherry (see Materials and Methods and Nelson et al. for detailed information) were transformed into agrobacteria (Nelson et al., 2007). Agrobacteria carrying the GFP-AtTPR7 construct and either ER or Golgi marker constructs were co-infiltrated into tobacco leaves. We analysed intact tobacco leaves (Fig. 2A) as well as protoplasts (Fig. 2B,C), which were isolated from tobacco leaves two days after infiltration. Fluorescent signals were observed by confocal laser scanning microscopy. Chlorophyll autofluorescence is false coloured in blue. As can be seen in merged pictures of GFP–AtTPR7 and an ER marker, as well as merged pictures of GFP–AtTPR7 and chloroplasts, signals clearly overlap with the ER marker fluorescence pattern in both, intact leaves and protoplasts. Magnifications (lower panels Fig. 2B,C) show that although ER structures and chloroplasts overlap at some points (see also merged pictures of chloroplasts and ER marker in supplementary material Fig. S2A), no clear ring around the chloroplasts of the GFP signal is visible as would be expected for a chloroplast envelope localisation. Also, no colocalisation of GFP–AtTPR7 with the Golgi marker is visible (Fig. 2C). Immunoblot analysis with membrane proteins extracted from infiltrated tobacco leaves detects a clear signal at 88 kDa, which is the expected size of the GFP–AtTPR7 fusion protein (supplementary material Fig. S2B). No signal is detected at 27 kDa, which would correspond to GFP alone. Therefore the fluorescent signal is derived from the fusion protein. Moreover, we generated a YFP-AtTPR7 construct (as was used by von Loeffelholz et al.) and co-expressed it in tobacco leaves together with the mCherry Marker. Again, the AtTPR7 signal is visible in the ER membrane (supplementary material Fig. S2C) (von Loeffelholz et al., 2011).

As a next step we aimed to further asses the localisation of the endogenous AtTPR7. To separate the individual membranes of different cellular compartments a membrane preparation of Arabidopsis leaves was loaded onto a sucrose density gradient (density: 1.05–1.25 g/ml). Gradients were fractionated and proteins were detected by immunodecoration with specific polyclonal AtTPR7 and AtToc64 antisera (Fig. 3A, upper panels). Toc64 was mainly found in fractions with a density of 1.08 g/ml, as it is expected for chloroplast outer envelope membranes (Cline et al., 1981). AtTPR7 in contrast was found in fractions with a sucrose density of 1.17 g/ml, corresponding to ER membranes (Ceriotti et al., 1995). The AtTPR7 antisera showed cross reactivity with the pea homologue and we therefore additionally performed the same experiment with a membrane preparation from pea leaves. Again, AtTPR7 was found in fractions corresponding to the ER membrane, clearly separated from PsToc64 (Fig. 3A, lower panels). Fractions 16–19 of the gradients containing Arabidopsis proteins, corresponding to ER membranes of wild-type Arabidopsis, pea and a attpr7 T-DNA mutant (supplementary material Fig. S3A–C) were subjected to SDS-PAGE and immunodecorated with AtTPR7 antisera (Fig. 3B). Intact and pure Arabidopsis and pea chloroplasts, isolated with a Percoll step gradient, were used additionally as a control. A signal for AtTPR7 was only obtained in wild-type Arabidopsis and pea ER membranes and no signal was detected in the chloroplast fractions or the attpr7 mutant. To test the purity of the ER membranes and chloroplasts and to provide a loading control we probed the samples with antisera against ER localised AtERdj2 as well as antisera against AtToc64/PsToc64.

Two isoforms of the yeast Sec63 orthologue are found in the Arabidopsis genome, AtERdj2A and AtERdj2B, sharing 71% identity. The antibody used in this study was raised against AtERdj2B although it has been shown to recognise both, AtERdj2A and AtERdj2B (Yamamoto et al., 2008). However, we could detect only a single band either representing AtERdj2B or both proteins were not separated sufficiently. Nevertheless, AtERdj2 was only detected in the ER fractions, whereas Toc64 was mainly detected in the chloroplasts, showing a high degree of purity of the fractions, apart from a slight contamination of the ER with chloroplast membranes.

To test whether AtTPR7 is an integral membrane protein, a total microsomal membrane preparation was treated with high salt, urea, Na₂CO₃ (pH 11), SDS and buffer as a control. AtTPR7 was only extracted from the membrane by SDS, clearly demonstrating that it is a membrane protein (Fig. 3C). AtERdj2 was monitored as a control integral membrane protein and showed similar behaviour.

Next, we applied a magnesium-induced-shift assay, which is frequently used as an indicator of ER membrane localisation (Ceriotti et al., 1995; Levitan et al., 2005). Ribosomes are released from the ER membrane upon treatment with EDTA. This in turn leads to a shift in migration of ER membranes in sucrose density gradients to a lower density in contrast to other membranes. In the presence of Mg²⁺, however, the small and the large subunit of the ribosome remain associated. Therefore, ER proteins are found in different fractions depending on the addition of EDTA in sucrose gradients. Total leaf extract was loaded on sucrose density gradients (1.05–1.25 g/ml) with 1 mM EDTA or with 2 mM MgCl₂, fractions were separated by SDS-PAGE and probed with AtTPR7 antisera as well as antisera against the ER specific protein SMT1 [sterol methyltransferase 1 (Boutté et al., 2010)] and the chloroplast outer envelope protein AtToc64 as a control. A clear shift is visible for AtTPR7 as well as SMT1 (Fig. 3E) confirming the localisation of AtTPR7 in the ER membrane, whereas Toc64 as a non-ER protein does not shift to lighter density fractions.

von Loeffelholz et al. showed that AtTPR7 is imported in an in vitro assay into isolated chloroplasts, also in competition with mitochondria (von Loeffelholz et al., 2011). We have therefore performed a competitive import assay with chloroplasts and ER and found that in this case AtTPR7 inserts preferentially into the ER membrane, although a minor amount of the protein is inserted into the chloroplast as was previously observed (von Loeffelholz. et al., 2011). AtToc64, which was used as a control, is predominantly targeted to chloroplasts (supplementary material Fig. S4). However, our cell fractionation assays analysing the endogenous AtTPR7 clearly point to an exclusive localisation in the ER.

Tail-anchored proteins are known to be inserted preferably post-translationally into the ER, where the transmembrane domain usually functions as a signal sequence (Borgese et al., 2001). We therefore investigated whether AtTPR7 is integrated post- or co-translationally into the ER. The Arabidopsis homologue of the co-translationally inserted luminal ER protein BiP (Denecke et al., 1991; Maruyama et al., 2010), AtBiP2, was used as a control. AtTPR7 and AtBiP2 were translated in vitro in a reticulocyte lysate and imported into dog pancreatic microsomes. To monitor post-translational protein import the proteins were added to the microsomes 20 min after the translation reaction. RNase A as well as cycloheximide were added to remove RNA, inhibit translation and prevent co-translational import before addition of microsomes. To observe co-translational translocation, translation was performed directly in the presence of microsomes (Fig. 4A). Indeed, AtTPR7 seems to be imported post-translationally in comparison to AtBiP2, which shows a strong import when allowing co-translational translocation and no import during post-translational import conditions. AtBiP2 is protected from digestion by proteinase K after co-translational import, demonstrating efficient import of AtBiP2, since AtBiP2 is not predicted to be glycosylated the processed signal peptide is only 3 kDa in size. Solubilisation of the membrane with 1% Triton (v/v) results in proteolysis of AtBiP2 by proteinase K. AtTPR7 in contrast is degraded by proteinase K in the absence of detergent, thus suggesting that the major part of the protein faces the cytosol. To ensure insertion of AtTPR7 microsomal membranes were treated after AtTPR7 import with buffer, 1 M NaCl or 4 M urea (Fig. 4B).

To further test the topology of the endogenous AtTPR7, we analysed Arabidopsis microsomal membranes. The membranes were likewise treated with proteinase K and probed with antisera against AtTPR7 and SMT1 as a control. AtTPR7 is again degraded in contrast to SMT1, which is a transmembrane protein facing the lumenal side (Boutté et al., 2010) and only proteolysed upon prior solubilisation of the membranes with Triton (Fig. 4C). Moreover, we performed the same experiment with membranes isolated from tobacco leaves transiently transformed with GFP–AtTPR7. Like the endogenous AtTPR7 GFP–AtTPR7 is rapidly degraded, thus confirming the predicted topology (supplementary material Fig. S5).

We therefore conclude that the N-terminal TPR domain of AtTPR7 is facing the cytosol, thus fulfilling an important prerequisite for a possible function as a chaperone receptor.

In the following experiment we aimed to elucidate whether AtTPR7 is able to interact with the cytosolic chaperones HSP90 and/or HSP70. We showed previously that the chloroplast TPR domain containing protein Toc64 interacts with HSP90–preprotein complexes (Qbadou et al., 2007). Therefore we tested whether the TPR domain of AtTPR7 could interact with HSP90 and/or HSP70. Since there are four isoforms of HSP90 present in the Arabidopsis cytosol (Krishna and Gloor, 2001), we tested whether any of them can interact with AtTPR7 in an in vitro pull-down assay.

The coding sequence of AtTPR7 lacking the C-terminal transmembrane domain was fused to a His tag replacing the transmembrane domain and recombinant AtTPR7–His was overexpressed in E. coli and purified via Ni-NTA. HSP90 isoforms were fused to an N-terminal Strep tag, which does not interfere with binding of the C-terminal EEVD motif to the TPR domain and purified accordingly. AtTPR7–His was incubated with either Strep–AtHSP90.1, Strep–AtHSP90.2, Strep–AtHSP90.3 or Strep–AtHSP90.4 and recovered by Ni-NTA. A sample without AtTPR7–His served as a control. Associated HSP90 was detected by immunoblotting with HSP90 antisera (Fig. 5A,B upper panel). All isoforms are able to interact with AtTPR7–His, although HSP90.2 and HSP90.3 showed a slightly increased binding.

Moreover, we tested the association of AtTPR7–His with Strep–AtHSP70.1, which is an abundant and constitutively expressed HSP70 isoform of the five HSP70 isoforms found in the Arabidopsis cytosol (Lin et al., 2001) (Fig. 5B, lower panel). Interaction with this chaperone was even stronger compared to HSP90 and is in line with previous results (von Loeffelholz et al., 2011). In addition to the ability of AtTPR7–His to interact with recombinant chaperones we tested whether AtTPR7–His could co-purify the chaperones from wheat germ lysate. AtTPR7–His was incubated with wheat germ lysate, re-purified by Ni-NTA and associated chaperones were again detected by immunoblotting. The results showed that AtTPR7–His is indeed also able to interact with wheat germ lysate HSP90 and HSP70 (Fig. 5B).

Since association of HSP70 among other factors with post-translationally inserted tail-anchored proteins is often observed during their membrane insertion (Abell et al., 2007; Rabu et al., 2008), we aimed to ensure that chaperone interaction is mediated specifically via the TPR domain. Therefore, an AtTPR7 mutant with a point mutation in the TPR region was generated. Comparison of the TPR domain of AtTPR7 with the well analysed TPR domains of the yeast HSP70 HSP90 organising protein (HOP), of which the TPR domain TPR1 is known to interact with HSP70 and the TPR domain TPR2a with HSP90, revealed a conserved lysine in position 181, which has been shown to be essential for HSP70 binding in HOP (Smith, 2004). The lysine was mutated by site-directed mutagenesis to a glutamic acid, thus introducing a negative charge in the protein binding pocket, which interferes with the chaperone EEVD peptide interaction. Recombinant AtTPR7–His wild-type and AtTPR7–His K181E proteins were incubated with (i) Strep–AtHSP70.1, (ii) Strep AtHSP90.2 and (iii) wheat germ lysate. AtTPR7–His was recovered with Ni-NTA (Fig. 5B) and the interacting chaperones were detected with antisera against HSP70 and HSP90. Association of Strep–AtHSP70.1 to AtTPR7–His was strongly reduced in the AtTPR7–His K181E mutant, indicating a specific interaction via the TPR region. Likewise, interaction with Strep–AtHSP90.2 was affected in the mutant. Nevertheless, residual binding was observed for both, HSP70 and HSP90, indicating that a single inserted negatively charged amino acid does not change the structure of the TPR domain drastically and thus still allows a weak chaperone association.

To ensure the specific interaction with chaperones via the TPR domain of AtTPR7 we generated a deletion mutant of AtTPR7 lacking the TPR domain (amino acids 104–212, AtTPR7ΔTPR–His). Furthermore, we tested chaperone association solely with the TPR domain (TPR-Domain–His). Wild-type AtTPR7–His as well as the two constructs were incubated with Strep–AtHSP70.1 and Strep–AtHSP90.2 and His-tagged AtTPR7 constructs were recovered and immunodecorated as described earlier. Binding of Strep–AtHSP70.1 as well as Strep–AtHSP90.2 to AtTPR7ΔTPR–His was completely abolished. In contrast to that association of Strep–AtHSP70.1 and Strep–AtHSP90.2 to TPR-Domain–His was comparable to wild-type AtTPR7–His (Fig. 5C).

Next, we wanted to elucidate whether AtTPR7 is part of a higher molecular weight complex possibly containing components of the Sec translocon. Therefore we analysed the composition of ER membrane complexes by blue native gel electrophoresis (BN-PAGE). Microsomal membranes were solubilized with 0.4% n-decyl-β-maltoside (DeMa) and complexes were separated on a native 5–12% gel in the first dimension. The composition of the separated protein complexes was further analysed by SDS-PAGE and immunoblotting in the second dimension. Fig. 6A (upper panel) shows that AtTPR7 is found in at least two higher molecular weight complexes of ∼140 and ∼200 kDa. Intriguingly, AtTPR7 co-migrates with AtERdj2, which is found in complexes of almost identical running behaviour (Fig. 6A, lower panel). Moreover, we subjected microsomal membranes of wild type and the attpr7 mutant to BN-PAGE. Interestingly, the higher molecular weight complexes at ∼200 kDa disappear in the mutant, which again points to an interaction between the two proteins, which cannot be formed in the mutant any longer (Fig. 4B). These data suggest that AtTPR7 and AtERdj2 form complexes, possibly containing further subunits.

To follow this line of evidence we tested whether a direct interaction between AtTPR7 and AtERdj2 could be observed. We chose AtERdj2A for these experiments, since AtERdj2A mutants are lethal in comparison to AtERdj2B and therefore seems to represent the physiologically more relevant isoform (Yamamoto et al., 2008). Purified AtTPR7–His was bound to Ni-NTA and incubated with radiolabelled AtERdj2A translated in reticulocyte lysate. Ni-NTA without bound protein was used as a control (Fig. 6C, upper panel). Indeed, AtERdj2A could be co-purified with AtTPR7–His. We could moreover show that the association seems to involve the C-terminal 161 amino acids of AtERdj2A, since AtTPR7–His was also able to interact with this peptide of AtERdj2A (Fig. 6C, lower panel).

To further test whether this interaction also persists in vivo, BiFC (bimolecular fluorescence complementation) analysis was applied. Full-length AtERdj2A and AtTPR7 were fused to complementary portions of fluorescent tags and expressed simultaneously in tobacco leaves resulting in a reconstituted fluorescence signal upon physical interaction (Gehl et al., 2009). The C-terminal part of SCFP was fused to the N-terminus of TPR7 to avoid interference with targeting. The N-terminal part of Venus was fused to the C-terminus of AtERdj2A, according to its likely topology as described for the homologue yeast Sec63 (Feldheim et al., 1992) (Fig. 7A). The vectors additionally contain either a HA or a c-myc tag, respectively, to monitor expression on protein level by immunoblotting with specific antisera. Both constructs were co-transformed with the ER mCherry marker protein described earlier in agrobacteria infiltrated tobacco leaves under the expression of the CaMV 35S promoter. Reconstituted fluorescence at 515 nm was clearly observed upon simultaneous expression of AtTPR7 and AtERdj2A (Fig. 7B, upper and middle left panels), indicating that the two proteins indeed interact physically in vivo. Comparison with the fluorescence of the ER marker protein shows clearly overlapping signals, as confirmed by colocalisation analysis, representing colocalised ER and BiFC signals in white (Fig. 7B, upper and middle right panels). This confirms an interaction within the ER membrane. As a negative control AtTPR7 was expressed together with the luminal protein AtBiP2, which should not be able to interact and prevent reconstitution of a fluorescent signal. The negative control shows slight background fluorescence not comparable to the AtERdj2A/AtTPR7 sample, even though pictures were taken with identical microscope settings. Fluorescent pixels were quantified and relative fluorescent values after background subtraction from 10 individual pictures are shown (Fig. 7C). Coexpression of AtTPR7 and AtERdj2A results in a relative fluorescence of 50% in contrast to coexpression of AtTPR7 and AtBiP2, which only reaches 10% relative fluorescence. To ensure proper expression of proteins in the negative control immunoblotting was performed. Signals were obtained with HA antisera, recognising SCFP– HA–AtTPR7 and with c-myc antisera recognising the AtERdj2A–c-myc–Venus and AtBiP2–c-myc–Venus protein (Fig. 7D, left and middle panel). No proteins were detected with the same antisera in untransformed tobacco leaves (Fig. 7D, right panel). Coomassie staining is shown as a loading control for all samples.

To investigate a possible in vivo function of AtTPR7 in post-translational ER import we asked whether AtTPR7 could complement a Δsec71 phenotype in yeast. Sec71 recruits the TPR domain containing component Sec72 to the ER membrane and Δsec71 mutants have been shown to lack accumulation of Sec72 due to rapid degradation (Feldheim and Schekman, 1994). Both proteins are required for post-translational import and we therefore tested the functional complementation of Δsec71 with AtTPR7 expressing the unprocessed precursor of a viral A/B toxin in mutant and complemented yeast cells. In S. cerevisiae, K28 killer strains are naturally infected by a double-stranded (ds)RNA killer virus which stably persists in the cytosol and encodes an unprocessed precursor (preprotoxin; pptox) of a secreted α/β heterodimeric protein toxin (Schmitt and Breinig, 2006; Schmitt and Tipper, 1990). After post-translational pptox import into the lumen of the ER, the toxin preprotein is processed during passage through the yeast secretory pathway, resulting in a disulfide-bonded α/β toxin which arrests sensitive cells at the G1/S boundary of the cell cycle and finally kills by causing cell death (Reiter et al., 2005). Post-translational pptox import into the ER is assisted by cytosolic chaperones and mediated through the Sec61 complex, the major import channel in the ER membrane. Since we previously demonstrated that pptox import into the ER and subsequent K28 toxin secretion is severely blocked in the genetic background of a Δsec71 knockout (Breinig et al., 2006), we asked whether AtTPR7 expression can complement this phenotypic defect by restoring post-translational pptox import into the ER and subsequent secretion of biologically active K28 toxin. The experimental setup for such complementation analysis is schematically illustrated in Fig. 8A; wild-type strain BY4742 and its isogenic Δsec71 knockout mutant were transformed with a vector constitutively expressing a V5 epitope-tagged version of the K28 toxin precursor (pptox–V5) from the phosphoglyceratkinase (PGK1) promoter. Thereafter, cells were co-transformed with a second plasmid expressing either (i) full-length AtTPR7, (ii) a C-terminal truncated variant (AtTPR7ΔTM) lacking its transmembrane domain, or (iii) wild-type SEC71. In each case, aliquots of cell-free culture supernatants adjusted to 2×10⁸ cells/ml were subjected to SDS-PAGE and probed with monoclonal V5 antisera. As shown in Fig. 8B, toxin secretion was blocked in a yeast Δsec71 mutant, confirming our previous data on post-translational pptox import into the ER (Breinig et al., 2006). In contrast, K28 toxin secretion was fully restored to wild-type level after co-expression of either wild-type SEC71 from S. cerevisiae (positive control) or full-length AtTPR7 from A. thaliana. Interestingly and as expected, C-terminal truncated AtTPR7ΔTM was not capable to complement the pronounced defect in post-translational pptox import into the ER of a Δsec71 mutant, as illustrated by the unaffected block in toxin secretion after AtTPR7ΔTM expression. Additionally, expression of both TPR constructs is shown on protein level (Fig. 8C) in the soluble and membrane fraction of transformed Δsec71 yeast cells by immunoblotting with AtTPR7 antisera. Full-length AtTPR7 is found in the membrane fraction, which is in accordance with the successful complementation, whereas AtTPR7ΔTM is found in the cytosol as expected.

Our study characterises AtTPR7 as a novel ER membrane protein in Arabidopsis that interacts with the Sec translocon and has a potential function in post-translational protein transport into the ER. A working hypothesis, summarising the data obtained on AtTPR7 in this work is presented in Fig. 9. The TPR domain of AtTPR7 is exposed in the cytosol thus allowing a specific interaction with HSP90 and/or HSP70. Moreover, interaction with AtERdj2A, as observed in this study, strongly suggests an association with a hypothetical plant Sec translocon for post-translational protein import composed of AtSec61 as a channel, AtERdj2 and possibly AtSec62 as it is found in yeast (Harada et al., 2011).

Several experiments in this study clearly demonstrate that AtTPR7 is localised to the ER, although previous data presented by von Loeffelholz et al., 2011 suggested AtTPR7 to reside in the outer membrane of chloroplasts. We have performed additional experiments to those presented by von Loeffelholz et al., 2011 showing the specificity of our antibody with a T-DNA insertion mutant and characteristic behaviour of AtTPR7 in a magnesium-induced-shift assay, which is widely used as an indicator for ER localised proteins (Ceriotti et al., 1995; Levitan et al., 2005). Although our data strongly imply that AtTPR7 predominantly – if not exclusively – resides in the ER, we cannot exclude that a minor amount of AtTPR7 is associated with the chloroplast outer membrane. Von Loeffelholz et al. show in their in vitro experiments that AtTPR7 interacts with a number of chloroplast preproteins and that soluble AtTPR7 can inhibit the import of these preproteins (von Loeffelholz et al., 2011). Since interaction with the preproteins is mostly mediated by HSP70, this might result in unspecific binding of AtTPR7 to chloroplast preproteins, although the authors could not observe a similar effect when using HOP TPR1 as a control TPR protein. Nevertheless, further careful analysis is necessary to explore whether AtTPR7 indeed plays a role in chloroplast import in vivo.

Intriguingly, several lines of evidence imply a function in post-translational protein translocation for AtTPR7. TPR domain containing proteins, which are found in all organisms and cellular compartments are known to contribute to post-translational protein import by the interaction with chaperones (Feldheim and Schekman, 1994; Qbadou et al., 2006; Schlegel et al., 2007; Young et al., 2003). HSP90 and HSP70 interact with preproteins in the cytosol and subsequently mediate a first contact between preproteins and membrane surfaces. In plants, the TPR domain containing protein Toc64 is associated with the chloroplast Toc translocon and has been shown to interact with HSP90 bound to preproteins (Qbadou et al., 2007). OM64 is a TPR protein in the outer mitochondrial membrane in Arabidopsis, which might functionally replace Tom70 and is involved in the import of some mitochondrial preproteins (Chew et al., 2004; Lister et al., 2007). Mammalian Tom70 and Tom34 also play a role in the delivery of chaperone-bound preproteins to mitochondria (Faou and Hoogenraad, 2012; Young et al., 2003). Although post-translationally imported proteins in the ER remain to be identified in plants, a mechanism involving chaperones seems likely. Many ER proteins are known to interact with cytosolic HSP70 (Ngosuwan et al., 2003) in yeast, and cytosolic and membrane associated chaperones have been suggested to assist post-translational import of tail-anchored ER proteins (Abell et al., 2007; Mariappan et al., 2011). So far no TPR domain containing protein has been identified in the ER besides Sec72 in fungi. Whether Sec72 recognises chaperones has not been investigated, however, in some fungi the TPR domain of Sec72 is considered a clamp-type TPR domain and phylogenetic clustering with the TPR1 motif of HOP suggest a potential for HSP70 binding (Schlegel et al., 2007). Involvement of HSP90 next to HSP70 in ER protein translocation remains to be elucidated in the future.

Our studies revealed that AtTPR7 exposes its clamp-type TPR domain to the cytosol, a prerequisite for its interaction with cytosolic chaperones. Indeed, AtTPR7 interacts with all cytosolic Arabidopsis HSP90 isoforms, most strongly with AtHSP90.2 and AtHSP90.3, which are constitutively expressed in Arabidopsis (Krishna and Gloor, 2001). Possibly different HSP90 isoforms might participate in chloroplast and ER protein sorting, however, this issue has not been addressed so far. A clear interaction was likewise observed with AtHSP70.1, which is also a constitutively expressed and highly abundant HSP70 isoform of the five cytosolic isoforms found in Arabidopsis (Lin et al., 2001). Interaction of recombinant AtTPR7 with both chaperones was additionally observed in wheat germ lysate. Preferential interaction with either of the chaperones remains to be analysed by biophysical measurements. HSP70 interaction with AtTPR7 was described previously (Kriechbaumer et al., 2011; von Loeffelholz et al., 2011); however, both these studies reported that no HSP90 interaction could be detected. Possibly variations in the reaction setup and/or experimental conditions favoured HSP90 binding in our experiments. For example in the previous work an N-terminal His-tagged AtTPR7 was used and HSP90 was treated with protease to cleave the His tag. In contrast to this, we used an N-terminal His tag, replacing the transmembrane domain and Strep-tagged HSP90. Nevertheless, interaction with HSP90 and HSP70 via the cytosolic TPR domain strongly suggests a possible interaction with cytosolic preprotein–chaperone complexes. However, to identify such target preproteins in plants, which are imported post-translationally into the ER, is challenging and remains to be elucidated. Moreover, a possible direct interaction of potential post-translationally imported ER preproteins with AtTPR7, as it is known from Tom20 and Pex5, remains to be analysed. The notion of AtTPR7 being involved in post-translational protein import was further strengthened by the successful complementation of a yeast Δsec71 mutant by full-length AtTPR7, which fully restored the defect in toxin secretion in Δsec71, which is due to a less efficient post-translational import process. Importantly, Sec71 recruits the TPR domain containing protein Sec72 to the membrane, which also plays a role in post-translational import, and is rapidly degraded in Δsec71. We therefore suggest that AtTPR7 functionally replaces both yeast proteins, since it comprises not only a cytosolic TPR domain but also a transmembrane anchor. AtTPR7 may therefore represent the first ER TPR docking protein in higher eukaryotes.

However, AtTPR7 Arabidopsis loss-of-function mutants do not show a visible phenotype under standard growth conditions, indicating that the chaperone recognition process can be bypassed, as it is also the case for the chloroplast receptor Toc64 and the mitochondrial TPR protein OM64 in Arabidopsis (Aronsson et al., 2007; Lister et al., 2007). Likewise, neither the loss of Sec71/Sec72 nor deletion of the yeast/mammalian mitochondrial TPR receptors Tom70 and Tom34 results in a lethal phenotype (Fang and Green, 1994; Faou and Hoogenraad, 2012; Young et al., 2003).

Since AtTPR7 might functionally replace Sec72 we would expect it to be associated with components of the Sec translocon of Arabidopsis to transfer potential substrate proteins to the channel Sec61. In yeast, Sec63 and Sec62 participate in post-translation protein import next to the Sec61 channel and Sec71/Sec72 (Harada et al., 2011). Indeed, our data present the interaction of AtTPR7 with AtERdj2, the Arabidopsis homologue of yeast Sec63. This interaction was observed not only by in vitro experiments, but also in vivo with the help of BiFC analysis and co-migration of the two proteins in native gels. Again this supports our hypothesis concerning the function of AtTPR7 as part of the Sec translocon. In addition, it provides a first insight into the architecture of the Sec complex in plants, which has not been analysed so far, although several essential components showing homology to yeast/mammalian components have been identified (Yamamoto et al., 2008).

In summary, AtTPR7 presents a third chaperone docking protein, next to Toc64 and OM64, which implies a general mechanism in recognition of post-translationally imported proteins in plant cells and presents an intriguing and exciting system to explore ER, mitochondrial and chloroplast protein translocation in a single cellular context.

Wild-type Arabidopsis thaliana (ecotype Columbia) and the attpr7 mutant line were grown on soil under 16 h light/8 h dark in the climate chamber at 22°C at 120 µE/m²/s. Nicotiana benthamiana was grown in soil under greenhouse conditions. Pisum sativum (var. Arvica) were grown under a 14 h light/10 h dark regime at 20°C/15°C. The attpr7 T-DNA insertion line with an insertion in exon 9 (SALK_057977) was obtained from the SALK collection (http://signal.salk.edu) and homozygous lines were isolated by PCR using the oligonucleotides AtTPR7-Ex9-f, AtTPR7-Ex11-rev and LBa1 (supplementary material Table S1).

Leaves of 4- to 6-week-old Nicotiana benthamiana were infiltrated with agrobacteria for transient expression of gene constructs. The Agrobacterium tumefaciens strain Agl1 was transformed with the respective gene constructs and cell cultures were resuspended in infiltration medium [10 mM MgCl₂, 10 mM MES/KOH (pH 5.7), 150 µM acetosyringone] to a final OD₆₀₀ of 1.0 and incubated for 2 h in the dark. Protein expression was monitored after 2 days. Protoplasts were prepared as described previously (Koop et al., 1996), however, cell walls were digested for 90 min at 40 rpm in 1% cellulase R10 and 0.3% macerase R10 after vacuum infiltration. Fluorescence was observed with a confocal laser scanning microscope at 20°C (Leica, Type: TCS SP5; objective lens: HCX PL APO CS, magnification: 63×, numerical aperture: 1.3; imaging medium: glycerol, software: Leica Application Suite/Advanced Fluorescence).

For localisation experiments the AtTPR7 coding sequence was cloned into the binary gateway vector pK7WGF2 (Plant Systems Biology, Gent). The Golgi marker is composed of the cytoplasmic tail and transmembrane domain of soybean α-1,2-mannosidase I and the ER marker consists of the signal peptide of Arabidopsis thaliana wall-associated kinase 2 at the N-terminus of the fluorescent protein and the ER retention signal His-Asp-Glu-Leu at its C-terminus (Nelson et al., 2007). Construction of the plant binary gateway vector pAM-PAT-35S-YFP-GW used for the YFP fusion construct is described elsewhere (Lefebvre et al., 2010). Constructs for fluorescence and BiFC analysis were cloned into appropriate entry and destination vectors with the gateway system (Invitrogen; see supplementary material Table S1) (Gehl et al., 2009). Quantification of fluorescent pixels and colocalisation analysis was performed with ImageJ (http://imagej.nih.gov/ij/).

Three week old Arabidopsis thaliana leaves were homogenised in a buffer containing 0.05 M Tris-HCl (pH 7.5), 0.5 M sucrose, 1 mM EDTA using an ultra turrax. The filtrated lysate was centrifuged twice to remove thylakoids, intact chloroplasts and mitochondria (first centrifugation 4200 g, 10 min; second centrifugation 10,000 g, 10 min, 4°C). To pellet the microsomal membranes the remaining lysate was centrifuged at 100 000 g for 1 h. Microsomes were resuspended in appropriate buffers and either loaded on sucrose density gradients or directly subjected to further experiments.

Proteins were separated on 10 or 12% polyacrylamide gels and immunodetection was performed as described previously (Lamberti et al., 2011). Arabidopsis chloroplasts were isolated on percoll gradients as described previously (Benz et al., 2009). AtERdj2A Antisera were kindly provided by Shuh-ichi Nishikawa. SMT1 antiserum was obtained from Agrisera (Vännäs, Sweden), GFP and HA antisera from Roche (Grenzach-Wyhlen, Germany) and c-myc antisera from Santa Cruz Biotechnology (Santa Cruz, USA). HSP90 and HSP70 antisera were generated against wheat chaperones and are described elsewhere (Fellerer et al., 2011). Polyclonal PsToc64 antisera were raised against pea PsToc64.

100 µg microsomes were resuspended in 50 µl ACA buffer (750 mM aminocaproic acid, 50 mM Bis Tris (pH 7), 0.5 mM EDTA) and solubilized with 0.4% DeMa for 10 min on ice. After centrifugation (10 min, 16,100 g, 4°C) the supernatant was loaded on a 5–12% native gel. The first dimension gel stripes were incubated for 15 min in 62.5 mM Tris-HCl (pH 6.8), 1% SDS, 1% β-mercaptoethanol, and 15 min in the same buffer without β-mercaptoethanol. The gel stripes were applied on a second dimension 12% SDS-PAGE and analysed by immunoblotting.

Microsomal fractions or total plant lysate ground in homogenisation buffer (100 mM Tris-HCl pH 7.8, 10 mM KCl, 12% sucrose) was applied on a linear sucrose gradient (1.05–1.25 g/ml sucrose) and centrifuged for 2 h at 166,900 g and 4°C in a swing-out rotor.

Templates for in vitro transcription were cloned into pF3A (Promega, Madison, USA; see supplementary material Table S1 for oligonucleotides) and in vitro transcription was performed with SP6 polymerase (Fermentas, St-Leon-Rot, Germany). In vitro translation in reticulocyte lysate (Promega) was performed according to manufacturer’s instructions. Co-translational import of radiolabelled proteins into dog pancreatic microsomes was achieved by incubation of the microsomes with the translation mix at 30°C for 60 min. Post-translational import was achieved by incubation of the microsomes with freshly translated protein after RNase A digest for 5 min at 30°C (80 µg/ml) and cycloheximide treatment (100 µg/ml) as described elsewhere (Lang et al., 2012), except that ER membranes were washed with 100 mM Na₂CO₃ and pelleted prior to SDS-PAGE to remove residual uninserted protein. Triton X-100 was added in a concentration of 1% (v/v). Proteinase K (50 µg/ml final concentration) digest was performed for 30 min on ice and the reaction was stopped with 200 mM PMSF and addition of SDS loading buffer.

AtTPR7 lacking the transmembrane domain of AtTPR7 (amino acids 1–500) used for pull down experiments was cloned into pET21a⁺ (Novagen, Darmstadt, Germany), expressed in M9ZB medium at 25°C for 5 h and purified via Ni-NTA-affinity chromatography (GE Healthcare). For production of polyclonal antisera full length AtTPR7 and TOC64 were cloned with a C-terminal StrepII tag into pET21a⁺ and expressed in M9ZB medium over night at 18°C and purified accordingly. Antibodies against the purified proteins were generated by Biogenes (Berlin, Germany). AtHSP90 isoforms were initially amplified from Arabidopsis cDNA using oligonucleotides recognising the 3′ and 5′ UTR to ensure amplification of the correct isoform. In a second step the genes were cloned into pET51b with an N-terminal StrepII tag. Site-directed mutagenesis of AtTPR7 was performed as described previously (Kunkel et al., 1987). Sequences of all clones were controlled by DNA sequencing. Oligonucleotides and constructs are summarised in supplementary material Table S1.

AtTPR7–His was incubated with Strep-tagged chaperones, wheat germ lysate or proteins translated in reticulocyte lysate as indicated for 1 h at RT. AtTPR7–His was subsequently re-purified by incubation with Ni-NTA for 1 h at RT and proteins were eluted with 300 mM imidazol and either probed with HSP70 or HSP90 antisera or detected by autoradiography.

The S. cerevisiae wild-type strain BY4742 and its isogenic Δsec71 knockout mutant were derived from the Saccharomyces Genome Deletion Consortium and obtained from Open Biosystems. Yeast transformations and immunoblot analysis of secreted K28 toxin in cell-free culture supernatants of wild-type and Δsec71 cells before and after co-transformation with the expression vectors pAtTPR7, pAtTPR7ΔTM, pSEC71 or pLEU2 were performed as previously described (see supplementary material Table S1 for constructs and oligonucleotides) (Breinig et al., 2006; Breinig et al., 2002; Heiligenstein et al., 2006; Schmitt and Tipper, 1990).

Chloroplasts from P. sativum were isolated as described previously (Waegemann and Soll, 1995). Chloroplasts (corresponding to 10 µg chlorophyll) and 1 µl dog pancreatic microsomes were mixed in a 100 µl reaction volume in a buffer containing (50 mM HEPES, pH 8.0, 330 mM sorbitol, 8.4 mM methionine, 13 mM ATP, 13 mM MgCl₂) and incubated at 30°C for 20 min with radiolabelled translation products synthesised in reticulocyte lysate. After the import reaction chloroplasts were separated from ER membranes by centrifugation through a 40% percoll cushion and pelleted at 3000 g. ER membranes were pelleted by centrifugation at 200,000 g. Both membranes were washed with 100 mM Na₂CO₃ to remove uninserted protein.

Sequence data from this article can be found in the NCBI data libraries under accession numbers: At5g21990 (AtTPR7), At3g17970 (TOC64), At1g79940 (AtERdj2A), At5g42020 (AtBiP2), At5g02500 (AtHSP70.1). At5g52640 (AtHSP90.1), At5g56030 (AtHSP90.2), At5g56010 (AtHSP90.3), At5g56000 (AtHSP90.4), SPAC4D7.01c (SEC71), Bd1g56980 (Brachypodium distachyon TPR7), NP_001058942 (Oryza sativa TPR7), XP_001765981 (Physcomitrella patens TPR7), XP_002977395 (Selaginella moellendorfii TPR7), XP_001701985 (Chlamydomonas reinhardtii TPR7), YOR027W (STI1). TPR and transmembrane domains were predicted by NCBI conserved domain search and Aramemnon, respectively (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi and http://aramemnon.botanik.uni-koeln.de). Alignments were generated using ClustalW.

Richard Zimmermann is acknowledged for the generous gift of dog pancreatic microsomes and advice on ER import experiments. AtERdj2 antisera were a kind gift from Shuh-ichi Nishikawa. Vectors and strains for tobacco infiltration were kindly provided by Norbert Mehlmer and Thomas Ott. Katharina Schöngruber and Stefanie Rapp are acknowledged for excellent technical assistance. We further thank Sandra Ring for the construction of AtTPR7 plasmids and WaiLing Chang for preparation of chloroplasts.