Different effects of Sec61α, Sec62 and Sec63 depletion on transport of polypeptides into the endoplasmic reticulum of mammalian cells

Protein transport into the endoplasmic reticulum (ER) is the first step in the biogenesis of most secreted proteins and many organellar proteins in mammalian cells (Zimmermann et al., 2011). Furthermore, integration into the ER membrane is the first step in the biogenesis of many eukaryotic membrane proteins (Borgese and Fasana, 2011). Typically, protein transport into the ER involves N-terminal signal peptides or C-terminal tail anchors in the precursor proteins and occurs co- or post-translationally. In the case of signal-peptide-containing precursors, transport and membrane integration can be divided into three stages: targeting, initial membrane insertion (at the Sec61 complex, see below) and completion of membrane integration or translocation. Co-translational precursor protein targeting involves cleavable or non-cleavable signal peptides at or near the N-terminus of the precursor polypeptides plus the ribosome, the signal recognition particle (SRP), the SRP receptor (SR) (Walter and Blobel, 1981; Meyer and Dobberstein, 1980) and the ribosome receptor that is identical to the Sec61 complex (Kalies et al., 1994). Alternatively, synthesis of the respective precursor proteins can be initiated at ribosomes that are permanently associated with the ER (Potter et al., 2001), thus eliminating the involvement of the SRP and SR. The SRP, SR and ribosomes but not the N-terminal signal peptides are dispensable for post-translational precursor protein targeting that has been observed for small presecretory proteins (Schlenstedt and Zimmermann, 1987; Müller and Zimmermann, 1987; Müller and Zimmermann, 1988; Schlenstedt et al., 1990). This post-translational transport is typically facilitated by cytosolic molecular chaperones, such as Hsc70 and Hsp40 (Dierks et al., 1993). Initial co- and post-translational membrane insertion and completion of translocation involve the Sec61 complex, comprising α-, β- and γ-subunits (Görlich et al., 1992; Hartmann et al., 1994; Wirth et al., 2003; Van den Berg et al., 2004; Becker et al., 2009). However, these mechanisms have never been directly demonstrated at the cellular level. Following initial membrane insertion, both types of precursor protein are typically processed by signal peptidase and/or oligosaccharyl transferase.

In addition to the Sec61 complex, the ER-membrane-integrated Hsp40 chaperone Sec63p and its lumenal Hsp70 partner Kar2p (also termed BiP) were found to be essential and to be involved in co- and post-translational transport in Saccharomyces cerevisiae (Brodsky et al., 1995; Young et al., 2001). However, the function of mammalian Sec63 protein remains elusive, even though Sec63 is known to occur in stoichiometric amounts in canine pancreatic microsomes relative to the α-subunit of the Sec61 complex, and in association with this complex (Tyedmers et al., 2000). Mammalian BiP (also known as 78 kDa glucose-regulated protein; Grp78) appears to have two functions in protein translocation, it is involved in the insertion of precursor polypeptides into the Sec61 complex (assayed as processing by signal peptidase) or opening of the Sec61 channel (Klappa et al., 1991; Dierks et al., 1996), and it binds to the incoming precursor polypeptide and acts as a molecular ratchet, thereby facilitating completion of translocation (assayed as sequestration) (Nicchitta and Blobel, 1993; Tyedmers et al., 2003; Shaffer et al., 2005). Analogous to the situation in Saccharomyces cerevisiae (Brodsky et al., 1995; Young et al., 2001), Sec63 could recruit BiP to the Sec61 complex and to translating ribosomes; Sec63 also activates BiP for interaction with its substrates (Tyedmers et al., 2000). However, Görlich and Rapoport reported that Sec63 and BiP are not required for reconstitution of protein translocation using purified membrane proteins from the mammalian ER and model precursor polypeptides such as preprolactin, preimmunoglobulin κ light chain and VSV G protein (Görlich and Rapoport, 1993).

The ER membrane protein Sec62 is associated with the Sec61 complex and Sec63 (Meyer et al., 2000; Tyedmers et al., 2000) as well as with ribosomes on the ER surface in mammalian cells (Müller et al., 2010). The yeast ortholog Sec62p is only involved in post-translational protein transport into the yeast ER and supposedly forms a signal peptide receptor together with the proteins Sec71p and Sec72p (Lyman and Schekman, 1997; Plath et al., 1998). Mammalian cells appear to lack orthologs of the latter two yeast proteins, whereas trypanosomes lack Sec62p and Sec72p orthologs but contain an ortholog of Sec71p (Zimmermann and Blatch, 2009). The function of mammalian Sec62 remained elusive, too.

Tail-anchored (TA) membrane proteins use C-terminal tail anchors for post-translational integration into the mammalian ER membrane plus one of a number of different cytosolic factors, such as the SRP, cytosolic molecular chaperones (Hsc70 and Hsp40) or the TA-dedicated machinery, the so-called transmembrane recognition complex (TRC) (Abell et al., 2004; Abell et al., 2007; Stefanovic and Hegde, 2007; Rabu et al., 2008). At the level of the ER membrane, certain TA proteins, such as cytochrome b₅ (cytb5), can enter the ER membrane unassisted (Brambillasca et al., 2005; Brambillasca et al., 2006). Other TA proteins, however, require the ER-membrane-resident SR or TRC receptors (Abell et al., 2004; Vilardi et al., 2011). An additional involvement of the Sec61 complex seemed to be a possibility for TA proteins, such as synaptobrevin2 (Syb2) and Sec61β, that can involve the SRP and SR (Abell et al., 2004). However, other in vitro experiments have suggested that the Sec61 complex is not involved in membrane integration of TA proteins (Kutay et al., 1995; Stefanovic and Hegde, 2007). Physiologically, TA proteins are not modified by signal peptidase or oligosaccharyl transferase. However, the established model TA proteins typically contain glycosylation sites at the C-terminus in order to allow quantitative analysis of membrane integration.

We addressed the central role of the human Sec61 complex in protein transport into the ER. The strategy was to silence the SEC61A1 gene in human cells by employing different SEC61A1-targeting siRNAs and to study the transport of various precursor proteins into the ER of the respective semi-permeabilized cells. Although silencing of the SEC61A1 gene affected signal-peptide-dependent protein transport in general, it did not affect transport of various TA proteins. In addition, we addressed the putative roles of mammalian Sec62 and Sec63 in protein transport into the ER with the same experimental strategy. The silencing of the SEC63 gene only affected a subset of precursor proteins. This phenotype was reproduced in murine Sec63^−/− cells. By contrast, silencing of the SEC62 gene inhibited post-translational transport of a signal peptide-containing precursor protein.

In order to demonstrate that the Sec61 complex is involved in protein transport into the ER in mammalian cells, the effect of silencing of the SEC61A1 gene had to be analyzed. HeLa cells were treated twice with candidate siRNAs for various incubation times. Western blot analysis showed that two different siRNAs that were directed against the coding (SEC61A1) and untranslated (SEC61A1-UTR) regions of the SEC61A1 gene were the most efficient at silencing the gene when used individually at a concentration of 20 nM or used in combination at 10 nM each (supplementary material Fig. S1A). Whether used alone or in combination, there was maximum silencing (~90%) after 96 hours of incubation. At that time, the growth and viability of the silenced cells only began to be affected compared with cells that had been transfected with a negative control siRNA (supplementary material Fig. S1B,C; Fig. S2). Between 96 and 120 hours, however, the SEC61A1-silenced cells started to die. As expected, the SEC61A1 gene is essential in human cells. However, these results defined an experimental window between 90 and 100 hours after first transfection with SEC61A1-targeting siRNAs that allows the functional analysis of the SEC61A1 gene product.

The cells were also characterized by quantitative RT-PCR (qPCR) at various time points to determine the levels of particular mRNAs (supplementary material Fig. S1D). The qPCR data demonstrated efficient depletion of SEC61A1 mRNA at 48 hours. According to the western blot analysis, however, most efficient depletion of the Sec61α1 protein took an additional 48 hours (supplementary material Fig. S1A). Silencing the SEC61A1 gene also affected the levels of the tail-anchored β- and γ-subunits of the Sec61 complex (supplementary material Table S1). At the mRNA level, however, there was no decrease in the SEC61B gene transcription (supplementary material Fig. S1D). These results were consistent with a scenario in which there is degradation of the β- and γ-subunits in the absence of the crucial α-subunit.

The levels of all other tested ER membrane and lumenal proteins were not reduced in SEC61A1-silenced cells compared with control-siRNA-treated cells after 96 hours of SEC61A1 gene silencing (supplementary material Table S1), suggesting that this silencing time did not result in global secondary effects on the ER proteome. Furthermore, BiP and Grp170 mRNA levels as well as protein levels did not indicate protein misfolding in the ER, i.e. did not show that the unfolded protein response had been activated (supplementary material Fig. S1D; Table S1). However, we detected an upregulation of the SRA and SRB gene expression as a result of silencing the SEC61A1 gene (supplementary material Table S1). Furthermore, the cellular ribosome content was not affected by SEC61A1 silencing, suggesting that no adaptive processes in global protein synthesis had been initiated (supplementary material Table S1).

Next, the morphology of the ER in the silenced cells was characterized by three-dimensional (3D)-structured illumination microscopy (3D-SIM) and electron microscopy (EM). HeLa cells were treated with control, SEC61A1 or SEC61A1-UTR siRNA up to 96 hours and then, subjected to 3D-SIM and EM. Cells treated with control or one of the SEC61A1 siRNAs did not show gross alterations in the typical ER morphology regardless of how much Sec61 complex was present (supplementary material Fig. S3). However, qualitative and quantitative EM analysis demonstrated the central role of the Sec61 complex as an ER membrane resident ribosome receptor (Kalies et al., 1994), in that ~60% less ER-bound polysomes were observed after 96 hours of SEC61A silencing, i.e. the average number of ER-bound ribosomes was reduced from 17 per μm in control-siRNA-treated cells to 7 per μm in cells depleted of Sec61 complex (Fig. 1).

The protein transport activity of the Sec61 complex in the silenced cells was monitored using an established transport assay that employs in vitro synthesis of different model precursor polypeptides in the presence of semi-permeabilized cells (Wilson et al., 1995). In addition to the presecretory protein preprolactin (ppl; also known as pre-Prl), several types of precursor were employed in the transport assay to gain a more general picture of ER protein translocation. These precursor proteins included the presecretory yeast protein prepro-α-factor (ppαf), ERj3 (ER lumen; also known as DnaJ homolog subfamily B member 11 in human), the invariant chain of the human class II major histocompatibility complex (ivc; type II plasma membrane protein), aquaporin-2 (AQP2; polytopic plasma membrane protein) and presecretory protein preprocecropin A (ppcecA; post-translationally transported). In the case of precursors of soluble proteins, initial insertion of precursors into the Sec61 complex was assayed as modification by signal peptidase and/or oligosaccharyl transferase because these two enzymes act on incoming polypeptides at the lumenal face of the ER membrane. By contrast, completion of translocation was assayed as protease resistance of the mature protein in the absence of detergent (in sequestration analysis). HeLa cells were incubated with the two SEC61A1-targeting siRNAs or the negative control siRNA for 96 hours and then semi-permeabilized with digitonin. First, ppl was synthesized in reticulocyte lysate in the presence of [³⁵S]methionine. The translation reactions were supplemented with equivalent amounts of the different semi-permeabilized siRNA-treated cells; after incubation, all samples were subjected to protease treatment and SDS-PAGE and phosphorimaging analysis. In the presence of cells incubated with the negative control siRNA, ppl was cleaved to the mature prolactin (pl) that was protected against externally added protease by the ER membrane. This demonstrated that pl had entered the ER lumen (Fig. 2A). In the SEC61A1-silenced cells, ppl was processed to pl less efficiently (between 4±2% and 7±3% of the control-siRNA-treated cells; Table 1), and, therefore, there was very little of the protease-protected pl. Similar results were obtained for the precursors of pαf and ERj3 (Table 1). The model membrane- and glycoprotein invariant chain was also analyzed in the transport assay (Fig. 2A; Table 1). In the presence of cells incubated with the negative control siRNA, the invariant chain was glycosylated and partially protected against the externally added protease by the ER membrane, demonstrating that it had entered the ER membrane. In the SEC61A-silenced cells, the invariant chain was glycosylated less efficiently (between 18±5% and 28±4% of the control siRNA-treated cells), and very little of the invariant chain fragment was protected from the protease. Similar results were obtained for the precursor of AQP2 (Table 1). Thus, silencing of the SEC61A1 gene led to a reduced ability to co-translationally import presecretory and membrane protein precursors.

In order to further substantiate the conclusion from SEC61A1 silencing we attempted expression of the SEC61A1 cDNA, lacking the UTR of the SEC61A1 gene, in the presence of the SEC61A1–UTR siRNA and observed that it indeed rescued the phenotype of SEC61A1 silencing, i.e. in contrast to the vector control it partially restored the transport activity (Fig. 2B). Averaging three signal-peptide-containing precursor proteins, processing activity was increased by the SEC61A1 cDNA from 16±6% to 62±19% of the control-siRNA-treated cells (data not shown). Based on western blot analysis, the complementation efficiency of the SEC61A1 expression plasmid was ~64% (n=17), which could explain the incomplete rescue. Thus the observed effects of SEC61A1 silencing were indeed due to the SEC61A1 gene. Also, the levels of Sec61β (60±5%, n=12) and SRβ (181±11%, n=15) partially returned to control levels under these complementation conditions, and complementation restored cell proliferation and viability (supplementary material Fig. S4).

Next, ppcecA, a post-translationally transported model presecretory protein with a cleavable signal peptide, was examined in the transport assay (Fig. 2C; Table 1). The precursor protein was synthesized in the absence of membranes and, subsequently incubated with siRNA-treated semi-permeabilized cells as indicated. The samples were then subjected to protease treatment, SDS-PAGE and phosphorimaging. As expected, in the presence of cells with intact Sec61 complexes, ppcecA was inserted into the Sec61 complex (i.e. processed to mature procecropin A; pcecA) and imported into the ER, where it was protected against externally added protease. By contrast, in the SEC61A1-silenced cells, ppcecA was barely processed (between 21±3% and 27±5% of the control siRNA-treated cells) and imported. Thus, silencing of the SEC61A1 gene also led to a reduced ability of the ER to post-translationally import small presecretory proteins.

Finally, we studied membrane integration of various TA proteins, such as cytb5-ops28 (Colombo et al., 2009), Sec61β-ops13 and Syb2-ops13 (Rabu et al., 2008) that employ different cytosolic factors. SEC61A1 silencing had no effect on membrane integration of these TA membrane proteins, as can be deduced from equally efficient glycosylation of the C-terminus under all conditions (Fig. 2D; Table 1). Thus, silencing of the SEC61A1 gene did not lead to a reduced ability to import TA proteins and had no overall effects on the integrity of the ER membrane, as can be concluded from the activity of oligosaccharyl transferase that depends on a lipid-linked oligosaccharide.

Having established a powerful experimental system to analyze the role of membrane proteins in protein transport into the mammalian ER, we next employed this system for Sec62. An efficient protocol for silencing of the human SEC62 gene had already been developed (Greiner et al., 2011). Western blot analysis characterized two different siRNAs that were directed against the coding (SEC62) and untranslated (SEC62-UTR) regions of the SEC62 gene as most efficient in silencing the gene when used individually at a concentration of 20 nM (Fig. 3A). The maximum silencing (~90%) was observed after 96 hours of incubation. In contrast to the SEC61A1 gene the SEC62 gene might not be essential in HeLa cells, or possibly a small amount of Sec62 is sufficient for growth of HeLa cells (Greiner et al., 2011). The protein transport activity of the SEC62-silenced cells was monitored using the above described transport assays. HeLa cells were incubated with the two SEC62 siRNAs or the negative control siRNA for 96 hours and treated with digitonin. Then, transport assays were carried out for three types of precursor protein: the co-translationally transported precursor protein ppl, the small presecretory protein ppcecA and the tail-anchored protein cytb5. In contrast to the other two types of protein, ppcecA was less efficiently processed and imported after SEC62 silencing under co- and post-translational assay conditions (Fig. 3B,C; Table 2). Thus, silencing of the SEC62 gene led to a reduced ability for post-translational import of a small presecretory protein, but had no effects on co-translational transport or on post-translational membrane integration of TA proteins (Fig. 3D,E). In order to further substantiate this conclusion we attempted expression of the SEC62 cDNA, lacking the UTR of the SEC62 gene, in the presence of the SEC62-UTR siRNA and observed that it rescued the phenotype of SEC62 silencing, i.e. it partially restored the transport activity (Fig. 3F; Table 2). Based on western blot analysis the overexpression of the SEC62 gene under these conditions was ~600%. Thus the observed effect of SEC62 silencing on ppcecA transport was indeed due to the SEC62 gene.

We also employed this experimental system for Sec63. First, an efficient protocol for silencing of the SEC63 gene had to be developed. HeLa and NIH/3T3 cells were treated twice with candidate siRNAs for various times. Western blot analysis characterized three different siRNAs that were directed against the coding (SEC63/Sec63) and SEC63-UTR regions of the SEC63/Sec63 gene as most efficient at silencing the gene when used individually at a concentration of 20 nM (Fig. 4A). The maximum silencing (~90%) was observed after 96 hours of incubation. The growth and viability for up to 168 hours of SEC63/Sec63-silenced cells was indistinguishable from that of control siRNA-treated cells (not shown). Thus, in contrast to the SEC61A1 gene the SEC63/Sec63 gene is not essential in human and murine cells. This notion was further corroborated by the fact that there was no difference in cell growth and viability between Sec63^+/+ and Sec63^−/− cells (see below). We note, however, that knockout of the Sec63 gene in mice led to early embryonic lethality (Fedeles et al., 2011).

The protein transport activity of the Sec63 in the silenced cells was monitored using the above established transport assay. HeLa or NIH/3T3 cells were incubated with the SEC63/Sec63 siRNAs or the negative control siRNA for 96 hours and then treated with digitonin for permeabilization. Equivalent amounts of cells were used for the protein transport activity assay. First, ppl was synthesized in reticulocyte lysate in the presence of [³⁵S]methionine. The translation reactions were supplemented with the different semi-permeabilized siRNA-treated cells; after incubation, all samples were subjected to protease treatment and SDS-PAGE and phosphorimaging analysis. In the presence of human or murine cells incubated with the negative control siRNA or with either of the two SEC63/Sec63 siRNAs, ppl was cleaved to the mature pl (87±6% to 102±16% of the control siRNA-treated cells) that was protected against externally added protease by the ER membrane (Table 3). Similar results were obtained for the precursor of immunoglobulin κ (Table 3). By contrast, in the presence of cells incubated with the SEC63/Sec63 siRNAs, the precursor of the ER lumenal protein ERj3 was less efficiently processed and glycosylated to gERj3 (between 48±6 and 56±10% of the control-siRNA-treated cells) that was protected against externally added protease by the ER membrane (Fig. 4B; Table 3). Thus, transport of pERj3 but not of ppl and preimmunoglobulin κ into the ER lumen involves Sec63.

Next, the precursor of ERj1 (type I ER membrane protein; also known as DnaJ C1) was analyzed using the transport assay (Table 3). In the presence of cells incubated with the negative control siRNA or with either one of the SEC63/Sec63 siRNAs, pERj1 was processed to ERj1 (between 96±5% and 101±6% of the control-siRNA-treated cells). The model membrane proteins and glycoproteins, ivc and AQP2 were also analyzed in the transport assay (Fig. 4B; Table 3). In the presence of cells incubated with the negative control siRNA, both precursors were glycosylated and partially protected against the externally added protease by the ER membrane, demonstrating that they had entered the ER membrane. However, in the SEC63/Sec63-silenced cells, the ivc and AQP2 were glycosylated less efficiently (between 40±0% and 75±4% of the control-siRNA-treated cells), and very little of the ivc and AQP2 was protected from the protease. Thus, insertion of AQP2 and invariant chain into the ER membrane, but not ERj1, involves Sec63.

Finally, ppcecA, the model post-translationally transported presecretory protein, was examined in the transport assay (Fig. 4B; Table 3). The precursor protein was synthesized in the absence of membranes and subsequently incubated with siRNA-treated semi-permeabilized cells as indicated. The samples were then subjected to protease treatment, SDS-PAGE and phosphorimaging. In the presence of cells with intact Sec63, ppcecA was inserted into the Sec61 complex (i.e. processed to mature pcecA) and imported. By contrast, in the SEC63/Sec63-silenced cells, ppcecA was less efficiently processed (between 69±10% and 71±6% of the control-siRNA-treated cells).

To rule out unspecific effects of SEC63/Sec63 silencing, we studied membrane integration of the TA proteins cytb5-ops28 and Syb2-ops13. SEC63/Sec63 silencing had no effect on membrane integration of these proteins as shown by unimpaired glycosylation in the SEC63/Sec63-silenced cells (Fig. 4B; Table 3). Thus, silencing of the SEC63/Sec63 gene led to a reduced membrane integration or import of certain precursors, but had no overall effects on the integrity of the ER membrane, i.e. on glycosylation activity and membrane integration or import of other precursor proteins.

In order to further substantiate this conclusion we also induced a rescue of the SEC63 silencing phenotype in HeLa cells by expressing the SEC63 cDNA lacking the SEC63 UTR in the presence of the SEC63-UTR siRNA (Fig. 4C). The glycosylation efficiency of ERj3 was increased by SEC63 cDNA expression by a factor of about two (n=3) and the complementation efficiency of the SEC63 expression plasmid was ~97±8% (n=3). Thus, the observed effects of SEC63 silencing were a direct result of a lack of Sec63.

In order to address the transport activity of mammalian Sec63 using an siRNA-independent approach, similar transport experiments were carried out with murine Sec63 knockout cells (Fedeles et al., 2011). The protein transport activity of Sec63 was monitored using the above described transport assay. Several precursor proteins (ppl, pre-immunoglobulin κ and pERj1) were not affected by the absence of Sec63, whereas transport of the other precursor proteins (pERj3, pivc, pAQP2, ppcecA) was affected to varying degrees (Fig. 4D; Table 3). We note that in the latter cases, initial insertion into the Sec61 complex was lower in the absence of Sec63, as assayed by signal peptidase processing or glycosylation. To control ER integrity in the Sec63^−/− cells (including the glycosylation machinery) we employed the glycosylatable ER TA proteins cytb5 and Syb2 (Fig. 4D; Table 3). Both proteins were glycosylated in the deficient cells as efficiently as in control cells. Overall, the effects of Sec63 depletion on transport of certain precursor proteins were more pronounced in the Sec63^−/− cells than in the SEC63-silenced cells, most probably because of the more efficient Sec63 depletion.

Quantitative western blotting of Sec63^+/+ and Sec63^−/− cells confirmed the substrate specificity of Sec63 for transport of proteins into the ER in vivo: we observed no ERj3 and hardly any AQ2 in Sec63^−/− cells, whereas the amounts of other resident ER proteins such as the membrane proteins ERj1, Sec61α, Sec62, SRβ and calnexin and the lumenal proteins BiP, Grp170, protein disulfide-isomerase (PDI) and calreticulin were not reduced in Sec63^−/− cells (supplementary material Table S2).

The experiments described above suggested a precursor-specific role of Sec63 in the early phase of protein transport. The concept of substrate-specific gating of the Sec61 channel mediated by signal sequences was originally put forward for the translocation of the prion protein (PrP) into the ER (Kim et al., 2002). Therefore, we included murine PrP in our analysis of the effects of Sec63 depletion. Two variants of prion protein were employed that lack the central hydrophobic region (also termed potential transmembrane domain; amino acid residues 113–133; ΔHD) and the GPI acceptor region (residues 231–254, ΔGPI), respectively (Winklhofer et al., 2003). Sec63^−/− cells and the negative control Sec63^+/+ cells were treated with digitonin. The translation reactions were supplemented with the two types of semi-permeabilized cells, and after incubation, all samples were subjected to protease treatment and SDS-PAGE and phosphorimaging analysis (Fig. 5; Table 3). In the presence of Sec63^−/− cells the precursors of the two PrP variants were less efficiently glycosylated to gPrP that was protected against externally added protease by the ER membrane (average glycosylation: 36±8%; n=4). Thus, the lack of Sec63 led to a defect in transport of PrP into the ER.

Transport of nascent polypeptides into the membrane or lumen of the mammalian ER involves the Sec61 complex (Görlich and Rapoport, 1993) that forms both a ribosome receptor (Kalies et al., 1994) and a gated pore in the ER membrane (Kim et al., 2002; Wirth et al., 2003). So far, this interaction had been shown only in vitro. Here we confirmed with a cell-based demonstration that the product of the SEC61A1 gene is involved in ribosome binding and in co-translational transport of soluble and membrane proteins into the ER. Our findings must be considered in the context of the existence of the SEC61A2 gene in human cells. The product of this gene shows 95% sequence identity to SEC61A1 and should not be depleted under our conditions of SEC61A1-siRNA-mediated gene silencing. Thus, the SEC61A2 gene could not complement the loss of function of the SEC61A1 gene, either because the SEC61A2 gene is not expressed in HeLa cells to any extent or its product has a different function to the product of the SEC61A1 gene. We favor the first explanation because expressed sequence tagging and array databases, as well as our preliminary RT-PCR experiments, indeed suggest that the SEC61A2 gene is expressed at a very low level, even after silencing of the SEC61A1 gene (data not shown).

Our experimental approach also allowed us to directly address the question of SEC61A1 gene involvement in post-translational protein transport into the mammalian ER. This kind of protein transport has gained interest recently in the context of the biogenesis of TA membrane proteins and associated cytosolic factors. Here, we provide the first demonstration that the product of the SEC61A1 gene also is involved in post-translational transport of signal peptide-containing precursor proteins into the human ER. This transport mechanism was previously shown to be ribonucleoparticle independent and to involve cytosolic molecular chaperones, such as Hsc70 plus Hsp40 (Schlenstedt and Zimmermann, 1987; Müller and Zimmermann, 1987 Müller and Zimmermann, 1988; Schlenstedt et al., 1990; Klappa et al., 1994). Furthermore, our experiments represent the first cell-based demonstration that the product of the SEC61A1 gene is not involved in post-translational integration of TA proteins into the mammalian ER membrane. We note that the latter was observed not only for cytb5, which can be delivered to the ER membrane by Hsc70 plus Hsp40 (Abell et al., 2007; Rabu et al., 2008) and spontaneously inserts into the membrane (Brambillasca et al., 2005; Brambillasca et al., 2006), but also for Sec61β and Syb2, two model TA proteins that can be delivered to the ER membrane by the SRP and SR (Abell et al., 2004) or TRC and the TRC receptor (WRB) (Stefanovic and Hegde, 2007; Rabu et al., 2008; Vilardi et al., 2011). Therefore, we propose that the previously observed cross-links between TA proteins and subunits of the Sec61 complex must have resulted from the close association of SR and the Sec61 complex rather than from a role of the Sec61 complex in TA membrane insertion (Abell et al., 2003). We note that our findings on TA biogenesis in mammalian cells are entirely consistent with the corresponding observations in yeast cells (Steel et al., 2002; Yabal et al., 2003).

Here, we provide the first demonstration that the product of the human SEC62 gene is involved in protein transport into the mammalian ER. We note that ppcecA can partially use the SRP–SR pathway when ER membranes are present during its synthesis and that its post-translational transport is SRP and SR independent (Schlenstedt et al., 1990). Therefore, the partial inhibition of ppcecA transport under co-translational conditions and the substantial inhibition of its post-translational transport suggest that only the SRP- and SR-independent transport of ppcecA involves Sec62. These observations are consistent with corresponding observations in yeast (Lyman and Schekman, 1997; Plath et al., 1998), but also raise the question of how they can be reconciled with the presence of a ribosome-binding site in mammalian Sec62 (Müller et al., 2010). We propose that the ribosome-binding site in Sec62 is involved in coordinating co- and post-translational protein transport in the mammalian ER, e.g. by preventing synthesis of precursor proteins near the Sec61 complex, whereas the Sec61 complex is engaged in post-translational transport. However, many more precursor polypeptides will have to be analyzed under these conditions to deduce any general rules.

Our findings are also the first demonstration of a role for Sec63 in co-translational protein translocation into the mammalian ER, and add to the observed defect in the biogenesis of certain plasma membrane proteins in a Sec63^−/− model of human polycystic liver disease (Fedeles et al., 2011). Sec63 plays a substrate-specific role in the initial insertion of certain precursors into the Sec61 complex, an early step of translocation. We suggest that in the transport of certain precursors, such as ppl, the signal peptides are strong enough to trigger opening of the Sec61 channel, i.e. after priming by the ribosome (Kim et al., 2002), whereas other precursors require additional action of Sec63 (or even Sec63 plus BiP). At least in the first case, the cytosolic Brl domain of Sec63 appears to be involved. We note that an early role for Sec63 and BiP in co-translational translocation of the precursor of dipeptidyl-aminopeptidase B into the ER, as well as a contribution of the cytosolic domain of Sec63, were previously observed in yeast (Young et al., 2001). Of further note, an early and precursor-specific role has also been reported for the mammalian TRAM protein (Voigt et al., 1996) and TRAP complex (Fons et al., 2003), which raises the question of whether or not the same precursor polypeptides involve Sec63, TRAM and TRAP.

What could distinguish Sec63-dependent from Sec63-independent precursor polypeptides with respect to the initial phase of translocation? One distinguishing factor could be the properties of the cleavable or non-cleavable signal peptides. Compared with the cleavable signal peptides of the other tested precursors (22–42 residues, four to eight charges), the cleavable signal peptides of ppcecA (22 residues, one charge), pERj3 (22 residues, no charged side chain) and pPrP (22 residues, no charged side chain) are short and apolar. However, more precursor polypeptides and chimeric precursors will have to be analyzed to deduce the underlying principles. Another distinguishing feature could be that some signal peptides insert into the Sec61 complex in loop conformation (Shaw et al., 1988) whereas others insert in a ‘head-on’ orientation that is later followed by a ‘flip-turn’ (Devaraneni et al., 2011). The latter might require a larger internal diameter of the Sec61 channel, i.e. wider opening of the lateral gate, and therefore, help from Sec63 (or Sec63 plus BiP). We note that a role of Sec63 and BiP in Sec61 channel opening was previously shown for the early phase of post-translational protein transport into the yeast ER (Lyman and Schekman, 1997). However, this idea was subsequently disputed by a model translocation reaction that allowed precursor movement through the yeast Sec61 complex in detergent solution (Matlack et al., 1997). We suggest that in this experimental system Sec61 channel gating might have been facilitated by the detergent.

Our observations are also relevant in the context of polycystic liver disease. Autosomal dominant polycystic liver disease (ADPLD) is a rare human disorder, characterized by the progressive development of biliary epithelial liver cysts. On the genetic level, ADPLD is heterogeneous involving at least two different genes: PRKCSH, encoding the β-subunit of glucosidase II, and SEC63 (Drenth et al., 2003; Li et al., 2003; Davila et al., 2004). Using mutant mice, it was observed that loss-of-function mutations in Prkcsh or Sec63 result in reduced levels of the plasma membrane protein polycystin-1 and, therefore, cyst formation in the liver (Fedeles et al., 2011). Here we showed that Sec63 can facilitate integration of the model plasma membrane protein AQP2 into the ER membrane of semi-permeabilized cells and that AQP2 is present at reduced levels in Sec63^−/− cells. We note that AQP2 levels are also reduced in Prkcsh^−/− cells, but that transport of Sec63-dependent precursor proteins such as AQP2 into the ER is not affected by the absence of PRKCSH (not shown). Thus, these results confirmed the previous suggestion that PRKCSH and Sec63 act at different stages in the biogenesis of proteins such as AQP2 and – by extrapolation – polycystin-1.

Complete SEC63 deletion results in pre-implantation defects because we never observed Sec63^−/− embryos, even at E7.0. Therefore, we generated a conditional Sec63^flox allele with loxP sites flanking exon 2 and a neo cassette flanked by FRT sites inserted into IVS2 (Fedeles et al., 2011). Removal of exon 2 by Cre recombinase activity results in complete loss of function (as documented by northern and western blotting). Sec63^flox/flox mice were crossed with the ImmortoMouse interferon-γ inducible H-2K^b-tsA58 SV40 temperature-sensitive transgene. Renal tubules (of TAL origin) from these mice were microdissected and the cells cultured in DMEM/Ham's F-12 medium (PAA, Pasching, Austria) with 1% fetal bovine serum (FBS), 10 μl/ml insulin–transferrin–selenium-X, 3.4 μg/ml 3,3′,5-triiodo-L-thyronine, 50,000 Units penicillin and streptomycin, and 10–20 Units interferon-γ under humidified conditions at 33°C in 5% CO₂. The resultant ‘parental’ cells were converted to a null cell line ex vivo by transfection with a Cre-recombinase plasmid (where Cre was fused to GFP) followed by either FACS sorting or limiting dilution cloning to obtain several clones of null cells (Sec63^−/−). In order to confirm complete deletion of the protein, western blotting with anti-Sec63 antiserum was performed.

HeLa cells (ATCC no. CCL-2) and NIH/3T3 (Irene Schulz, Homburg) cells were cultured in DMEM (Gibco-Invitrogen, Karlsruhe, Germany) containing 10% FBS (Biochrom, Berlin, Germany) and 1% penicillin and streptomycin (PAA, Pasching, Austria) in a humidified environment with 5% CO₂ at 37°C. Sec63 control cells (derived from the Sec63floxed mouse) and Sec63 null cells were cultured in DMEM/Ham's F-12 medium (PAA) containing 1% FBS, 1% insulin–transferrin–selenium-X, 10 IU/l interferon-γ from mouse, 1 μg/l 3,3′,5-triiodo-L-thyronine, and 1% penicillin and streptomycin under humidified conditions at 33°C in 5% CO₂. Growth rates and viability were determined using a Countess® Automated Cell Counter (Invitrogen). Cell viability was also evaluated using Nuclear-ID™ Blue/Green cell viability reagent (Enzo Life Sciences, Lausen, Switzerland) according to the manufacturer's instructions.

For SEC61A1 silencing, 5.2×10⁵ HeLa cells were seeded per 6 cm culture dish in normal culture. The cells were transfected with SEC61A1 siRNA (5′-GGAAUUUGCCUGCUAAUCAtt-3′; Applied Biosystems, Darmstadt, Germany), SEC61A1-UTR siRNA (5′-CACUGAAAUGUCUACGUUUtt-3′; Applied Biosystems), or control siRNA (AllStars Negative Control siRNA; Qiagen, Hilden, Germany) at a final concentration of 20 nM using HiPerFect Reagent (Qiagen) as described previously (Erdmann et al., 2011; Lang et al., 2011). After 24 hours, the medium was changed and the cells were transfected a second time. SEC62 silencing was carried out according to the same protocol with the previously described siRNAs (Greiner et al., 2011). For SEC63 gene silencing in HeLa or NIH/3T3 cells, 6×10⁵ cells were seeded per 6 cm culture dish in normal culture. The cells were transfected with human SEC63 siRNAs (SEC63: 5′- GGGAGGUGUAGUUUUUUUAtt-3′, SEC63-UTR: 5′-AGUCUAUGGUCCCAGUAAAtt-3′; Applied Biosystems), murine siRNA (Sec63: 5′-CCAGAACGCGGAGCAAAUUtt-3′; Applied Biosystems) or the same control siRNA as above at a final concentration of 20 nM using HiPerFect. For both cell types, the medium was changed after 24 hours and the cells were transfected a second time. Silencing was evaluated by RT-PCR and western blot analysis.

In order to rescue the phenotype of SEC61A1, SEC62 or SEC63 silencing in HeLa cells, the SEC61A1, SEC62 and SEC63 cDNA was inserted into the multiple cloning site (MCS) of a pCDNA3-internal ribosomal entry site (IRES)-GFP vector that contained the cytomegalovirus (CMV) promoter, the MCS, the IRES, plus the green fluorescent protein (GFP) coding sequence. Cells were treated with the respective UTR siRNA as described above for 48 hours. Subsequently, the cells were transfected with either vector or SEC61A1, SEC62 or SEC63 expression plasmid using Fugene HD (Roche Diagnostics, Mannheim, Germany). Transfection efficiency was ~80%, as determined by GFP fluorescence.

Cells were harvested and mRNA was isolated using the NucleoSpin RNA II Kit (Machery and Nagel, Düren, Germany). Reverse transcription was performed with Superscript II (Invitrogen) and oligo(dT) primer (Eurofins MWG Operon, Ebersberg, Germany). TaqMan® Gene Expression Assays (Applied Biosystems) were used for quantitative real-time PCR of SEC61A1 (Hs00273698_m1), SEC61B (Hs00606455_m1), SEC63 (Hs00273093_m1), ERJ1 (Hs01550849_m1) and BIP (Hs99999174_m1) in the HT7900 system (Applied Biosystems). Δct values were calculated using SOX (Hs01053049_s1) and BMI-1 (Hs00180411_m1) as a standard and the values were then normalized to control-siRNA-treated cells.

Rabbit antibodies were raised against the C-terminal peptides of human Sec61α (14-mer), Sec62 (11-mer) and SRβ (12-mer) or the N-terminal peptides of human BiP (12-mer), ERj3 (15-mer) and Grp170 (11-mer) plus an N- or C-terminal cysteine, and against purified PDI, calreticulin and ERj1C-ΔN21 (Dudek et al., 2005). Antibodies directed against AQP2 (New England Biolabs, Schwalbach, Germany), calnexin (Stressgene, Hamburg, Germany), β-actin (Santa Cruz Biotechnology, Heidelberg, Germany; Sigma, Taufkirchen, Germany), presenilin 1 (Millipore, Eschborn, Germany) and SERCA2 (Sigma) were obtained from commercial sources. The primary antibodies were visualized using ECL™ Plex goat anti-rabbit IgG–Cy5 conjugate or ECL™ Plex goat anti-mouse IgG–Cy3 conjugate (GE Healthcare, Freiburg, Germany) and the Typhoon-Trio imaging system (GE Healthcare) in combination with Image Quant TL software 7.0 (GE Healthcare).

Cells were fixed with paraformaldehyde, washed and an indirect immunofluorescence staining was performed with an affinity purified rabbit antipeptide antibody directed against the C-terminal undecapeptide of human Sec62 protein (plus an N-terminal cysteine) and Alexa-Fluor-488- or Alexa-Fluor-594-coupled secondary antibody from goat (Invitrogen). We note that the anti-Sec62 antibody is specific for Sec62 under denaturing as well as native conditions (i.e. western blot and fluorescence microscopy signals were quenched after silencing of the SEC62 gene) (Greiner et al., 2011). Cells were analyzed by microscopy using an Elyra SIM (Carl Zeiss MicroImaging, Göttingen, Germany). Synthesis of the Sec62 protein was not affected by SEC61A1 silencing for 96 hours (supplementary material Table S1). Electron microscopy Cells were plated on poly-lysine-coated sapphire discs in 6-cm dishes and treated for 96 hours as described above. Sapphire discs were frozen in a high-pressure freezer and subjected to freeze substitution and embedding as previously described (Liu et al., 2010). Ultrathin (70 nm) sections were cut parallel to the cell monolayer, collected on copper grids, stained with uranyl acetate and lead citrate, and analyzed on the Tecnai 12 Biotwin electron microscope (Philips, Eindhoven, The Netherlands). Ribosome content per ER membrane surface was quantified with a LabVIEW-based program.

Precursor polypeptides were synthesized in reticulocyte lysate (nuclease treated; Promega, Heidelberg, Germany) in the presence of [³⁵S]methionine (Perkin Elmer, Rodgau-Jügesheim, Germany) plus buffer, rough microsomes (RM) (positive control) or semi-permeabilized cells for 60 minutes at 30°C (co-translational transport experiment). Alternatively, precursor polypeptides were synthesized in reticulocyte lysate in the presence of [³⁵S]methionine for 15 minutes at 30°C. After 5 minutes of incubation with RNase A (final concentration: 80 μg/ml) and cycloheximide (final concentration: 100 μg/ml) at 30°C, buffer or semi-permeabilized cells were added and the incubation was continued for 30 minutes (post-translational transport experiment). Semi-permeabilized cells were prepared from identical cell numbers according to a published procedure (Wilson et al., 1995). Their concentrations were adjusted according to the OD₂₈₀ in 2% SDS and, eventually, confirmed by SDS-PAGE and protein staining. For protease treatment, samples were divided into aliquots and incubated in the absence or presence of proteinase K (170 μg/ml) for 60 minutes at 0°C. Where indicated, Triton X-100 was present at a final concentration of 1%. Protease treatment was terminated by the addition of phenylmethylsulfonyl fluoride (10 mM). All samples were analyzed by SDS-PAGE and phosphorimaging (Typhoon-Trio imaging system). Image Quant TL software 7.0 was used for quantifications.

We are grateful to N. Borgese (Milan, Italy), S. High (Manchester, UK) and W. Skach (Portland, USA) for providing the plasmids that code for glycosylatable cytb5–ops28 and MHC class II inavariant chain, Sec61β–ops13, synaptobrevin2–ops13 and aquaporin 2, respectively. The antibodies against Sec61γ were a kind gift from K. Kapp and B. Dobberstein (Heidelberg). We thank Monika Lerner and Anika Müller (Homburg) for excellent technical assistance and Detlef Hof for writing the software that was used for EM analysis.