Chemically monoubiquitinated PEX5 binds to the components of the peroxisomal docking and export machinery

Peroxisomal matrix proteins contain either a peroxisomal targeting sequence 1 (PTS1) or a PTS2 that are recognized by the import receptors PEX5 and PEX7, respectively. PEX5 transports the PTS1 proteins and the PEX7/PTS2 complex to the docking translocation module (DTM) at the peroxisomal membrane. After cargo release PEX5 is monoubiquitinated and extracted from the peroxisomal membrane by the receptor export machinery (REM) comprising PEX26 and the AAA ATPases PEX1 and PEX6. Here, we investigated the protein interactions of monoubiquitinated PEX5 with the docking proteins PEX13, PEX14 and the REM. “Click” chemistry was used to synthesise monoubiquitinated recombinant PEX5. We found that monoubiquitinated PEX5 binds the PEX7/PTS2 complex and restores PTS2 protein import in vivo in ΔPEX5 fibroblasts. In vitro pull-down assays revealed an interaction of recombinant PEX5 and monoubiquitinated PEX5 with PEX13, PEX14 and with the REM components PEX1, PEX6 and PEX26. The interactions with the docking proteins were independent of the PEX5 ubiquitination status whereas the interactions with the REM components were increased when PEX5 is ubiquitinated.

Mammalian peroxisomes are single membrane-bound organelles that do not contain DNA or RNA. All matrix proteins are nuclear encoded and translated on free ribosomes and hence all these proteins must be posttranslationally imported into peroxisomes 1 . Their correct sorting to the organelle is ensured by peroxisomal targeting signals (PTS), small peptide sequences present in their primary structure that are recognized by shuttling receptors. There are two types of PTSs, the so-called PTS1 and PTS2. The PTS1 2,3 is a short C-terminal signal peptide that is used by most peroxisomal matrix proteins. PTS1 proteins are recognized by the shuttling receptor PEX5 while still in the cytosol. This interaction involves the PTS1 itself, on one side, and the C-terminal tetratricopeptide repeat (TPR) domain of PEX5 4,5 , on the other side. The PTS2 is a degenerated nonapeptide present at the N-terminus of only a few mammalian enzymes 6,7 . PTS2 proteins are also transported to the peroxisome by PEX5. However, in this case the cargo protein-PEX5 interaction requires an extra factor, the co-receptor PEX7 [8][9][10] . We note that mammalian PEX5 is expressed in at least two forms, PEX5L and PEX5S, which are generated by alternative splicing of the PEX5 transcript 11,12 . PEX5L contains an insert of 37 amino acids that is positioned after amino acids 214 of PEX5S. This insert contains part of the binding-site for the PEX7/PTS2 cargo complex and thus only PEX5L (hereafter referred to as simply PEX5) can transport PTS2 proteins to the peroxisome 13,14 .
After binding their cargos in the cytosol, PEX5 or PEX5/PEX7 interact with the peroxisomal membrane docking translocation module (DTM), which in mammals comprises the peroxins 14 and 13 15 and the three RING (really interesting new gene) finger proteins PEX2, PEX10, and PEX12 16 . The N-terminal disordered region of PEX5 17 harbours several binding motifs for PEX14 and PEX13 that have been extensively studied by several groups 13,[18][19][20][21] . The interaction process may involve partly cooperative and sequential steps 22 . Although the composition of the DTM has been analysed in detail, the stoichiometry of each of its components is still unclear and

Results
Synthesis of monoubiquitinated PEX5 using the copper(I)-catalysed alkyne-azide cycloaddition (CuAAC). After delivering its cargos to the peroxisome matrix PEX5 gets monoubiquitinated at cysteine 11, a mandatory modification for its subsequent extraction from the DTM by the receptor export machinery REM. To gain more insights in the interaction of monoubiquitinated PEX5 with the components of the export machinery, it was necessary to have suitable amounts of the pure compound. Considering that naturally occurring monoubiquitinated PEX5 is a labile and a very low abundance protein we decided to chemically synthesise large amounts of it by linking ubiquitin to recombinant human PEX5 11 using a CuAAC reaction. A brief description of the procedure used is provided below.
To generate a ubiquitin-alkyne we generated a fusion protein (1) comprising (from the N-to the C-terminus) a Strep-tag, yeast ubiquitin, an intein (GyrA from Mycobacterium xenopi) and a chitin binding domain (CBD). Note that the ubiquitin moiety in this fusion protein lacks the last two glycines in order to approximate the native spacing between PEX5 and ubiquitin after the formation of a 1,2,3-triazole linkage by the CuAAC reaction (5) 42 (see Supplementary Fig. S1 for the comparison of the native and chemical linkage). The fusion protein Strep-UbΔGG-GyrA-CBD, referred to as Strep-Ub-GyrA-CBD (1), was subjected to thiolysis by treatment with 2-mercaptoethanesulfonic acid (MESNA). Subsequent addition of propargylamine (PA) resulted in Strep-Ub-alkyne (3) that was then purified by size exclusion chromatography (SEC) and verified by ESI-MS (Fig. 1c, lane 2 and Supplementary Fig. S2).
Next, a CuAAC reaction was performed using purified Strep-Ub-alkyne (3) and H 6 -PEX5AzF (4) at a ratio of 3:1 (Fig. 1b). The reaction was monitored by SDS-PAGE analysis which revealed the appearance of a new species migrating 10-15 kDa above H 6 -PEX5AzF (4), indicating formation of the desired product H 6 -PEX5AzF-Ub-Strep, referred to as H 6 -PEX5-Ub-Strep (5) (Fig. 1c, compare lanes 3 and 4). Residual starting materials (i.e., non-clicked Strep-Ub-alkyne and H 6 -PEX5AzF) were removed by subjecting the reaction mixture to two purification steps (Fig. 1c, lanes 5 and 6; see materials and methods for details). The chemically ubiquitinated structure differs slightly in the linkage between the PEX5 and ubiquitin motif from the natural one but was expected to retain all the key interactions of ubiquitinated PEX5 (see Supplementary Fig. S1 for the comparison of the native and chemical linkage).
H 6 -PEX5-Ub-Strep binds a PEX7/PTS2 protein complex in vitro and is functional in importing PTS2 proteins in vivo. To test the functionality of the chemically synthesised H 6 -PEX5-Ub-Strep protein several experiments were performed. In one of these we asked whether H 6 -PEX5-Ub-Strep can interact with the PEX7/PTS2 protein complex in vitro. Figure 2a shows the results of a pull-down assay performed with magnetic Ni-beads coupled with purified H 6 -PEX5 or H 6 -PEX5-Ub-Strep or no recombinant protein (negative control). The prepared Ni-beads were incubated with 35 S-pre-thiolase-myc (a PTS2 protein, referred to as 35 S-thiolase) and 35 S-myc-PEX7 (referred to as 35  S-methionine. The results show that both H 6 -PEX5 and H 6 -PEX5-Ub-Strep, can bind the two radiolabelled proteins (Fig. 2a). Note that in contrast to 35 S-thiolase (which contains 12 methionines) 35 S-PEX7 contains only two methionines in its primary structure, hence the weak signal in the autoradiograph. We obtained similar results with other PTS2-reporter proteins such as PTS2-CAT (a bacterial chloramphenicol transferase) and PTS2-GFP (see Supplementary Fig. S3). Unexpectedly, similar pull-down assays using radiolabelled PTS1 proteins, such as PTS1-GFP and pre-SCP2 (pre-sterol carrier protein 2), revealed that the clicked H 6 -PEX5-Ub-Strep is not able to bind these proteins (see Supplementary Fig. S4a   (a) Pull-down assay with magnetic Ni-beads that were coupled with purified H 6 -PEX5 and H 6 -PEX5-Ub-Strep, respectively or without protein (control). After incubating the beads with in vitro synthesised 35 S-PEX7, 35 S-thiolase, buffer A and ATP, the samples were eluted with imidazole. The bound (50%) and the unbound (10%) fractions as well as the input (TNT) were analysed by SDS-PAGE/autoradiographic detection. (b) The bound (2.9%) and unbound (1.1%) fractions of the pull-down assay in (a) were also analysed by immunoblot detection with α-PEX5. The lower migrating band of PEX5 is a common degradation product in E. coli that also gets ubiquitinated. The two cropped blots originate from one gel. The pull-down is representative for four experiments. (c) Human ΔPEX5 fibroblasts were electroporated with purified H 6 -PEX5, H 6 -PEX5-Ub-Strep and Strep-Ub-alkyne (control), respectively. On the second day, an immunofluorescent staining was performed against thiolase (Alexa 594) and the peroxisomal membrane protein PMP70 (Alexa 488). Both proteins, H 6 -PEX5 and H 6 -PEX5-Ub-Strep could import thiolase into peroxisomes which is co-localised with PMP70, while no import was detected in the control. The experiment was performed twice with H 6 -PEX5-Ub-Strep and four times with H 6 -PEX5. Complementation rates were calculated by counting punctate thiolase positive cells. At least 500 cells were counted for each electroporation. The scale bar represents 20 µm. (d) The stability of purified H 6 -PEX5 and H 6 -PEX5-Ub-Strep after electroporation in ΔPEX5 fibroblasts was analysed after the first and the second day by immunoblot detection using α-PEX5. MAP-Kinase (MAPK) was used as a loading control and detected with α-MAPK. Both cropped blots are from one blot that has been cut horizontally, developed with the different antibodies and exposed on a single film together. The numbers to the right indicate the molecular mass (M) of proteins in kDa (a,b and d). Finally, we examined the capacity of H 6 -PEX5-Ub-Strep to complement the peroxisomal protein import defect of ΔPEX5 fibroblasts. In these cells, the import of PTS1 and PTS2 proteins is impaired and therefore many of these proteins display a cytosolic localization 45,46 . For this purpose, we used electroporation to introduce purified H 6 -PEX5-Ub-Strep, H 6 -PEX5 or Strep-Ub-alkyne (control) into ΔPEX5 fibroblasts. Two days after electroporation, immunofluorescent staining was performed using antibodies against endogenous thiolase and against the peroxisomal membrane protein PMP70. As shown in Fig. 2c electroporation of both proteins, H 6 -PEX5 and H 6 -PEX5-Ub-Strep, resulted in the import of thiolase into peroxisomes as determined by its colocalization with PMP70 ( Fig. 2c, "merged" panels). We note that larger complementation rates were obtained with H 6 -PEX5 (32.4%) than with H 6 -PEX5-Ub-Strep (11.3%). No complementation could be detected when Strep-Ub-alkyne was electroporated. Immunoblot analysis of lysates from cells harvested one and two days after electroporation revealed that already on the first day the amount of H 6 -PEX5-Ub-Strep was reduced compared to H 6 -PEX5 (Fig. 2d, lanes 2 and 1, respectively). Furthermore, two days after electroporation we could still detect H 6 -PEX5 in cell lysates whereas H 6 -PEX5-Ub-Strep was barely detectable (Fig. 2d, lanes 4 and 5, respectively). Apparently, although H 6 -PEX5-Ub-Strep can complement the PTS2-import defect in ΔPEX5 fibroblasts it seems to be more unstable than H 6 -PEX5 (see discussion). We also asked whether H 6 -PEX5-Ub-Strep could restore protein import of EGFP-PTS1 in ΔPEX5 fibroblasts. However, in agreement with the PTS1 protein pull-down assays described above, H 6 -PEX5-Ub-Strep did not complement the import defect of PTS1 proteins in ΔPEX5 fibroblasts while the complementation rate for H 6 -PEX5 was 14% (see Supplementary Fig. 4b and discussion).

Recombinant H 6 -PEX5-Ub-Strep enables import of PTS2 proteins into peroxisomes in vitro.
Next, we addressed the question whether H 6 -PEX5-Ub-Strep can be used in an in vitro import system using post-nuclear supernatant (PNS) from mouse liver 47,48 . To investigate the import of PTS2 proteins into peroxisomes PNS containing import-competent peroxisomes was incubated with radiolabelled 35 S-PEX5, 35 S-pre-thiolase and 35 S-PEX7 and with different amounts of recombinant PEX5 proteins for 30 min, to allow import into the organelles. The PNS was subsequently treated with proteinase K to eliminate all not imported proteins.
No protease protection of 35 S-PEX7/ 35 S-pre-thiolase could be observed when these proteins were incubated alone with PNS (Fig. 3, lane 1). This indicates that endogenous PEX5 is a limiting factor for the import of matrix proteins and that no processing of the 35 S-pre-thiolase into the mature (mat) form occurred. A small amount of 35 S-PEX7/ 35 S-pre-thiolase is protease protected and processed when 35 S-PEX5 was added (Fig. 3, lane 2). 35 S-PEX5 itself also seems to be protease protected and embedded into the DTM. The addition of recombinant H 6 -PEX5 and H 6 -PEX5-Ub-Strep leads to a competition with the 35 S-PEX5 and results in a higher amount of protease-protected and processed thiolase (Fig. 3, lanes 3 to 5). This indicates that both recombinant proteins are functional for PTS2 protein import. Increasing the amount of recombinant H 6 -PEX5-Ub-Strep does not enhance the system but leads to less import (Fig. 3, lane 5). At 0 °C neither 35 S-PEX5 nor 35 S-PEX7 or 35 S-thiolase are protease-protected, indicating no import of these components (Fig. 3, lane 6).
H 6 -PEX5-Ub-Strep interacts with the receptor export module components PEX1, PEX6 and PEX26 and with PEX14 and PEX13. Export of DTM-embedded monoubiquitinated PEX5 is an ATP-dependent step catalysed by the receptor export module, which comprises PEX1, PEX6 and PEX26. Aiming to better understand the role of ubiquitin in this process, a series of in vitro pull-down assays were performed. We used magnetic Ni-beads containing purified H 6 -PEX5, H 6 -PEX5-Ub-Strep or no recombinant protein (negative control) as baits. In vitro synthesised 35 S-PEX1 and 35 S-PEX6 (individually or pre-mixed) and S-PEX26/ 35 S-PEX19 (note that PEX26 was co-synthesised with PEX19 in order to ensure that it remains soluble; see materials and methods for details) were investigated as preys. Some of these assays also contained 35 S-PEX14 and 35 S-PEX13, two components of the DTM. The aim was to cover the possibility that the interactions between PEX5 and the REM components might be dependent on DTM proteins. The results of these experiments are shown in Fig. 4 and Supplementary Fig. S5a. The samples used for Fig. 4a and Supplementary Fig. S5a were incubated with ATP and cytosol from ΔPEX5 fibroblasts to provide putative co-factors that might be required for some protein-protein interactions. Both, H 6 -PEX5 and H 6 -PEX5-Ub-Strep pulled-down 35 S-PEX1 and 35 S-PEX6 (Fig. 4a, lanes 7 and 19, respectively and Supplementary Fig. S5a). However, the amounts of radiolabelled PEX1 and PEX6 recovered with H 6 -PEX5-Ub-Strep were clearly larger than those obtained with H 6 -PEX5. A similar result was found with 35 S-PEX26. Note that although large amounts of 35 S-PEX19 were present in these assays (to maintain 35 S-PEX26 in solution) almost no PEX19 was recovered in the pull-down samples (Fig. 4a compared to 4c, see also Supplementary Fig. S5a, lanes 1 and 5). These results strongly suggest distinct interactions of H 6 -PEX5 and H 6 -PEX5-Ub-Strep with the peroxisomal export proteins PEX1, PEX6 and PEX26.
In addition, both, H 6 -PEX5 (Fig. 4a, lanes 1 and 8, respectively) as well as H 6 -PEX5-Ub-Strep (Fig. 4a, lanes 13 and 20, respectively) have the capability to bind to the peroxisomal docking proteins 35 S-PEX13 and 35 S-PEX14. Interestingly, we consistently found that slightly more 35 S-PEX26 was recovered with the H 6 -PEX5 and H 6 -PEX5-Ub-Strep beads when these assays were made in the presence of 35 S-PEX14 and 35 S-PEX1/ 35 S-PEX6 (Fig. 4a, compare lanes 4 and 16 with lanes 6 and 18, respectively). Additional binding assays performed in binding buffer instead of cytosol from ΔPEX5 fibroblasts and in the presence of ATPγS instead of ATP yielded similar results (Fig. 4b). The eluates of the samples in Fig. 4a,b were additionally blotted for PEX5 to investigate the binding to the Ni-beads (see Supplementary Fig. 5b,c). We regularly noticed that less H 6 -PEX5-Ub-Strep compared to H 6 -PEX5 could be coupled to the Ni-beads (see Supplementary Fig. 5b). The results above suggest that 35 S-PEX1/ 35 S-PEX6, 35 S-PEX13, 35 S-PEX14 and 35 S-PEX26 interact independently with H 6 -PEX5 and H 6 -PEX5-Ub-Strep.
The interaction between PEX26 and the DTM proteins. To better characterize the interactions involving PEX26 and other components of the peroxisomal import machinery, additional pull-down assays were performed, but this time using as a bait in vitro synthesised myc-tagged 35 S-PEX26. As shown in Fig. 5, only low amounts of radiolabelled prey proteins (PEX1, PEX6, PEX13 and PEX14) were recovered in these assays. The fact that only nanograms of bait are used in these experiments (in contrast with typical pull-down assays in which micrograms of recombinant bait proteins are generally used) may explain this result. For radiolabelled PEX1, and probably also for PEX6, the amounts recovered with PEX26-myc beads do not differ much from those appearing in the negative control with myc-tagged phythanoyl-CoA-hydroxylase (PHYH-myc) (Fig. 5, compare lanes 1 and 2 with lanes 6 and 7, respectively). However, a clear enrichment of radiolabelled PEX14 at PEX26-myc beads compared to beads loaded with the negative control was observed (Fig. 5, lanes 4 and 9). A possible PEX13 binding seems to be weak. Thus, these results suggest that PEX14 interacts with PEX26, as described previously 34,49 . S-PEX26/ 35 S-PEX19, or a mixture of these proteins. Both, radiolabelled PEX6 and PEX26 bound much better to the H 6 -PEX5-Ub-Strep-containing beads than to those containing Strep-Ub-alkyne (Fig. 6, compare lanes 6 and 7 with lanes 2 and 3, respectively), while almost no binding occurred to the control with Strep-GFP-coupled beads. The interaction of PEX1, PEX6, and PEX26 together with Strep-Ub-alkyne is weak, compared to the stronger binding to H 6 -PEX5-Ub-Strep (see Fig. 6 lanes 4 and 8).

Discussion
To get more insights into the extraction of monoubiquitinated PEX5 from the DTM by the REM components we generated PEX5 that was chemically ubiquitinated (H 6 -PEX5-Ub-Strep, see Fig. 1). In vitro analysis revealed that H 6 -PEX5-Ub-Strep binds PEX7/PTS2 proteins ( Fig. 2a and Supplementary Fig. 3) and interacts with the DTM components PEX14 and PEX13 (Fig. 4). In addition, this protein is functional in PTS2 protein import as shown by complementation studies in ΔPEX5 fibroblasts with endogenous thiolase (Fig. 2c) and by the in vitro import assay (Fig. 3). Thus, we have no evidence that the binding of H 6 -PEX5-Ub-Strep to PEX7/PTS2 proteins or to DTM proteins is affected by the ubiquitination. This also indicates that the deubiquitination is not a mandatory step neither for the binding to cargo nor for the binding to PEX13 and PEX14.
However, both the PTS1 protein pull-down assays and the complementation studies in ΔPEX5 fibroblasts indicate that H 6 -PEX5-Ub-Strep is not able to bind to PTS1 proteins and thus cannot complement the import defect of PTS1 proteins in ΔPEX5 fibroblasts ( Supplementary Fig. 4). We performed additional experiments that suggested that the oxidative conditions used in the CuAAC reaction affect the PTS1-binding domain of PEX5 (there are 5 cysteines in this domain) inducing small conformational changes. Indeed, H 6 -PEX5 subjected to this treatment (H 6 -PEX5 "click") as a control did not interact with PTS1 proteins in an in vitro pull-down experiment. This was shown using the PTS1 protein pre-SCP2 as an example (Supplementary Fig. 4c). We also recognized that the electroporation of purified H 6 -PEX5-Ub-Strep into ΔPEX5 fibroblasts resulted in a significant reduction of the amount of this protein after the first and the second day (Fig. 2d). Albeit H 6 -PEX5-Ub-Strep is functional in recovering the PTS2 protein import it is conceivable that the half-life of H 6 -PEX5-Ub-Strep might be shorter than that of H 6 -PEX5. The chemically added ubiquitin cannot be cleaved from PEX5, in contrast to the wild type modification. With one ubiquitin permanently attached, it is possible that PEX5 is poly-ubiquitinated leading to an increased proteasomal degradation. We do not have any data whether H 6 -PEX5-Ub-Strep is used for several rounds of peroxisomal transport like it is assumed for the endogenous PEX5 28,50-52 or just once. In the complementation studies in ΔPEX5 fibroblasts (Fig. 2) H 6 -PEX5-Ub-Strep was added in a vast excess to compensate for the short half-life of this protein.
Although the REM proteins were thought to be involved in the export of monoubiquitinated PEX5, until very recently neither the binding of Ub-PEX5 to the PEX1/PEX6 complex nor to PEX26 had been reported 32,34 . Our results show a binding of H 6 -PEX5 to the docking/REM components that is weak but clearly increased when PEX5 is ubiquitinated. The interactions were independent of each other ( Fig. 4 and Supplementary Fig. S5). While this manuscript was under review similar findings, i.e. the result that the AAA ATPases Pex1p and Pex6p interact with N-terminally linked ubiquitin-Pex5p fusion protein were reported for the yeast proteins 53 .
Our in vitro binding assays suggest that in particular PEX6 and PEX26 proteins bind only slightly to H 6 -PEX5 ( Fig. 4a and Supplementary Fig. S5a) and do not, or bind only slightly to Strep-Ub-alkyne (Fig. 6). However, both PEX6 and PEX26 independently display a substantial binding to ubiquitinated PEX5 (Fig. 4 and Supplementary  Fig. S5a). A similar binding behaviour of PEX1 to H 6 -PEX5-Ub-Strep was obtained, if the latter was pulled down using the His-tag (Ni-beads) (Fig. 4). However, if H 6 -PEX5-Ub-Strep was coupled engaging the Strep-tag (Strep-beads), less PEX1 could be pulled down (Fig. 6, lane 5). This might indicate a steric hindrance for the PEX1 binding to H 6 -PEX5-Ub-Strep when the protein is coupled to the beads using the Strep-tag of the ubiquitin moiety. Thus, this would support the idea that the PEX1 binding engages the ubiquitin part of H 6 -PEX5-Ub-Strep. This has also been shown recently using a photo-crosslinking approach in an in vitro export assay 31 or in an assay with purified yeast proteins 53 . Thus, it seems likely that H 6 -PEX5-Ub-Strep provides binding-sites for the docking and REM proteins. It has been shown that PEX5 comprising the first 10-125 residues can be correctly monoubiquitinated and exported from peroxisomes 17,23,31 , indicating that the binding sites are located in this first N-terminal part of PEX5. Taken together, our data may present a stalled situation where the maximum amount of interactions is possible. Especially, as we encounter almost no differences when we compare the pulldowns in the presence of ATP and cytosol (Fig. 4a) with the results with binding buffer and ATPγS (Fig. 4b). With ATPγS the action of the AAA ATPases PEX1 and PEX6 should be blocked. We assume that PEX5 regardless of the ubiquitination status is embedded in the pore that is formed by the DTM. Given that PEX26 interacts with PEX14 and monoubiquitinated PEX5, we hypothesise that PEX26 is also located in a close distance to the DTM or is even a part of it. This assumption is additionally encouraged by quantitative proteomic studies with PEX14 indicating that yeast PEX15 is a component of the transient pore 16 . Thus, PEX26 might act as a bridge between embedded (monoubiquitinated) PEX5, the DTM and PEX1/PEX6 complex. In recent studies, Miyata et al. identified AWP1/ZFAND6 as a PEX5 export factor that recognizes PEX6 and monoubiquitinated PEX5 and enhances the export of PEX5 from the peroxisomal membrane 54 , but did not bind the PEX1/PEX6 complex. Our interaction studies demonstrate that the binding of ubiquitinated PEX5 and PEX6 does not require additional cytosol. However small amounts of rabbit reticulocyte lysate are present in our pull-down assays and thus we cannot exclude that traces of rabbit AWP1 or other factors are present and needed for the interactions. According to very recent data the PEX1/PEX6 complex extracts Ub-PEX5 from the peroxisomal DTM using a processive threading mechanism 31,40 . However, exactly how the PEX1/PEX6 complex engages the DTM-embedded Ub-PEX5 protein and what is the mechanistic role of ubiquitin in all this process remains completely unknown. Clearly, an answer to these questions will require reconstituting the extraction reaction in vitro using pure recombinant proteins. The chemical synthesis of the bona fide substrate of the PEX1/PEX6 unfoldase described here is thus the first step towards this goal.

Expression in E. coli. H 6 -PEX5
was expressed in E. coli BL21 (DE3) in 2 L of YEP (10 g peptone, 10 g NaCl and 5 g yeast extract per 1 mL) containing appropriate antibiotics. Cultures were grown to an OD 600 = 0.4-0.6 at 37 °C, induced with IPTG (1 mM) and further grown for 18 h at 18 °C.
H 6 -PEX5AzF was co-expressed with pEVOL-pAzF, a vector encoding the orthogonal pAzFRS/tRNA CUA pair 56 (kindly provided by P.G. Schultz, Addgene plasmid 31186) in E. coli BL21 (DE3) in 2 L of M9-minimal media containing appropriate antibiotics as described before 43 . Cultures were grown to an OD 600 = 0.3-0.4 at 37 °C. The unnatural amino acid p-azidophenylalanine (AzF; Bachem) was added to a final concentration of 1 mM. Protein expression was induced with IPTG (1 mM) and 0.02% arabinose for 12 h at 28 °C.
Strep-Ub-GyrA-CBD was expressed in E. coli BL21 (DE3) in 8 L of YEP containing appropriate antibiotics. Cells were grown to an OD 600 = 0.4-0.6 at 37 °C, induced with IPTG (0.4 mM) and further incubated for 4 h at 28 °C.
In vitro synthesis of proteins and autoradiography. The in vitro transcription and translation (TNT) experiments were performed with the TNT ® Quick Coupled Transcription/Translation System (Promega) according to manufacturer's instruction. The translation products from the rabbit reticulocyte lysate (RRL) were labelled with 35 S-methionine (35 mCi/mL; Hartmann Analytic). The amounts of protein synthesised in our reactions were not determined but yields of 2-6 ng of protein/microliter of lysate are common according to the manufacturers information. The reactions were carried out for two hours at 30 °C. Generation of PEX26-myc (referred to PEX26) was performed together with the chaperone PEX19 62 (pCS3). For this, both