The signal hypothesis for targeting of nascent precursor polypeptides to the ER was put forward by G. Blobel et al. (Blobel, 1980). In later versions, it proposed that the amino-terminal SP of a nascent presecretory protein is recognized and bound by cytosolic SRP, which mediates a translational attenuation and facilitates association of the RNC-SRP complex with the heterodimeric SRP receptor (SR), which is membrane-anchored via the β-subunit (Siegel and Walter, 1988; Ng et al., 1996; Egea et al., 2005; Gamerdinger et al., 2015; Hsieh et al., 2020; Meyer and Dobberstein, 1980a; Meyer and Dobberstein, 1980b; Gilmore et al., 1982a; Gilmore et al., 1982b; Tajima et al., 1986; Jomaa et al., 2021; Jomaa et al., 2022). The interaction of SRP with SR drives the mutual hydrolysis of bound GTP and leads to transfer of the RNC to the Sec61 complex (Halic and Beckmann, 2005, Halic et al., 2006; Jomaa et al., 2021). Thus, SRP represents a precursor as well as a mRNA targeting device (Figure 2). Comparative ribosome profiling experiments addressed functionality of the bacterial and yeast SRP in vivo (Chartron et al., 2016; Schibich et al., 2016; Costa et al., 2018) and demonstrated the strong preference of SRP for TMHs regardless of their position relative to the amino-terminus of the nascent polypeptide chain. Furthermore, they demonstrated the efficient ER targeting of precursors with just cleavable SPs in absence of SRP. Thereby, these studies stretch the versatility of SRP and reconciled two important considerations. First, the comparatively low abundance of SRP as compared to the abundance of translating ribosomes can be compensated by an mRNA targeting step, probably extending the time-window for the target recognition by SRP. Second, the crowded environment at the ribosomal tunnel exit can be eased by multiple iterations for SRP recognition without being limited to recognition of the SP or first TMH.

Late in the 1980s, characterization of precursor proteins with the ability for SRP-independent ER targeting, such as small presecretory proteins in mammalian cells and TA-membrane proteins in mammalian and yeast cells suggested alternative ER targeting machineries (Müller and Zimmermann, 1987; Schlenstedt and Zimmermann, 1987; Schlenstedt et al., 1990; Kutay et al., 1993; Ast et al., 2013). In the early 2000s, some small model presecretory proteins were shown to be targeted to the mammalian ER membrane in an SRP-independent fashion by their interaction with the cytosolic protein calcium-calmodulin and its putative association with the calcium-calmodulin (Ca²⁺-CaM)-binding site in the cytosolic amino-terminus of the Sec61α protein, possibly representing yet another targeting mechanism (Shao and Hegde, 2011; Schäuble et al., 2012). With respect to pathway interconnections, it is interesting to note that Ca²⁺-CaM was found to inhibit rather than stimulate targeting of TA proteins to the ER membrane (Haßdenteufel et al., 2011).

Recently, we applied the proteomic strategy to identify precursor polypeptides that depend on SR for their targeting to the ER. Applying the established statistical analysis, we found that SRA depletion significantly affected the steady-state levels of 139 proteins: 133 negatively and 6 positively (Figure 6; Supplementary Table S1) (Tirincsi et al., 2022b). Among the negatively affected proteins, GO terms assigned 50% to organelles of the endocytic and exocytic pathways, thus representing a firm enrichment compared to the total quantified proteome (26%). Furthermore, we detected significant enrichment of precursor proteins with SP, N-glycosylated proteins, and membrane proteins. The negatively affected proteins included 24 proteins with cleavable SP, among them 14 membrane proteins plus 30 membrane proteins with TMH, including the ER hairpin membrane protein ATL2, many single-spanning membrane proteins and several multi-spanning membrane proteins, including the hairpin protein REEP3. Thus, the precursors of these negatively affected proteins with SP and TMH can be expected to be clients of the SRP and SR targeting pathway. When the SPs of SR-dependent precursor polypeptides were analyzed for hydrophobicity, GP content, and SP segmentation no significant distinguishing features were determined. Overall, SRP and SR clients showed a preference for cleavable SP (44%) or non-cleavable N-terminal targeting signals (77% of the remaining membrane protein clients) and an underrepresentation of TA, which is consistent with previous results from proximity-based ribosome profiling experiments (Chartron et al., 2016; Costa et al., 2018).

TA proteins are defined as single-spanning membrane proteins with a defining carboxy-terminal TMH (Kutay et al., 1993). Approximately 1% of the human protein-coding genome code for TA proteins. Not all of these, however, have a functional association with the secretory pathway (Borgese and Fasana, 2011; Borgese et al., 2019). TA proteins of the secretory pathway, such as the β- and γ-subunits of the Sec61 complex, the redox protein Cytochrome b₅, many apoptosis-associated proteins (including various Bcl family members) and many vesicular trafficking components (i.e., Syntaxins and VAMPs), have to be targeted to and inserted into the ER membrane (Borgese and Fasana, 2011; Borgese et al., 2019). Similar to SRP-mediated targeting, TA proteins are directed to the ER membrane via a heterodimeric ER membrane resident receptor complex, made up by WRB and CAML. The minimal cytosolic targeting machinery for TA proteins was termed TA receptor complex (TRC) in mammalian cells (Table 1). The cytosolic ATPase TRC40 binds the TA protein with its hydrophobic binding pocket and the WRB/CAML complex facilitates their efficient ER targeting. The WRB/CAML complex also facilitates the actual membrane insertion (Schuldiner et al., 2008; Yamamoto and Sakisaka, 2012; Borgese et al., 2019). Additionally, the TA targeting machinery includes a ribosome binding complex (comprising Bag6, Ubl4A, and TRC35), which acts upstream of TRC40 (Leznicki et al., 2010; Mariappan et al., 2010; Leznicki et al., 2011).

We applied our experimental strategy to identify precursor polypeptides that depend on WRB for their ER targeting (Tirincsi et al., 2022b). Applying the established statistical analysis, we found that WRB depletion significantly affected the steady-state levels of 296 proteins: 144 negatively and 152 positively (Figure 6; Supplementary Table S1). Among the negatively affected proteins, GO terms assigned 45% to organelles of the pathways of endocytosis and exocytosis. Some enrichment of precursor proteins with SP and membrane proteins was also detected. The identified precursors included 13 proteins with cleavable SP (including six membrane proteins) and 14 membrane proteins without SP, including the ER hairpin protein REEP3. When the SPs of WRB-dependent precursor polypeptides were analyzed for hydrophobicity, GP content, and SP segmentation, no significant distinguishing features were determined. However, when more WRB clients were identified under conditions of simultaneous depletion of WRB and hSnd2 (Figure 6) a preference of WRB for multispanning membrane proteins became visible and more WRB membrane protein clients were observed to have relatively more central and more carboxy-terminal TMH as compared to SRA dependent membrane proteins (Tirincsi et al., 2022b). Taken together, these results on the client spectrum of WRB point towards a more general targeting role of the TRC pathway than previously anticipated and may explain why pathogenic variants of TRC35 or TRC40 as well as CAML are linked to Congenital disorders of glycosylation in humans (Wilson et al., 2022). Notably, first hints towards this end already came from previous reports that small human presecretory proteins can be targeted to the ER of semi-permeabilized human cells by SR, WRB and hSnd2 (Haßdenteufel et al., 2018; Haßdenteufel et al., 2019) and that the cytosolic TRC pathway-component SGTA, which works upstream of Bag6, Ubl4A, and TRC35, is cotranslationally recruited to ribosomes, which synthesize a diverse range of membrane proteins, including those with cleavable SP (Leznicki and High, 2020).

Although roughly one dozen genes coding for yeast TA proteins were characterized as essential, knock-out strains for the TA targeting components are viable, suggesting at least one further targeting route (Schuldiner et al., 2008). Indeed, in 2016 a high-throughput screening approach in yeast by M. Schuldiner and coworkers identified a novel targeting pathway, termed SRP-independent (SND) (Aviram et al., 2016). Three components of this pathway were identified and named Snd1, Snd2, and Snd3 (Table 1). Two hallmarks of the SND targeting pathway emerged. First, similar to the SRP and TA targeting pathways, precursor polypeptides were targeted via the combination of a cytosolic factor (named Snd1) and a heterodimeric receptor in the ER membrane (termed Snd2 and Snd3). Previously, Snd1 had already been described as a ribosome-binding protein. Second, the SND pathway showed a preference for substrates with a central, rather than an amino- or a carboxy-terminal TMH. Furthermore, the SND pathway was able to provide an alternative targeting route for clients with a TMH at their amino- or carboxy-terminus (Aviram et al., 2016). Subsequent sequence comparisons identified the ER membrane protein TMEM208 as putative human Snd2 ortholog (named hSnd2) (Aviram et al., 2016). In experiments, combining siRNA-mediated gene silencing with protein transport into the ER of semi-permeabilized human cells in cell-free assays, hSnd2 appeared to have a similar function as its yeast ortholog (Casson et al., 2017; Haßdenteufel et al., 2017; Haßdenteufel et al., 2018), i.e., the TA membrane protein Cytochrome b₅ as well as some small presecretory proteins were targeted to the Sec61 complex in the mammalian cell-free assay. Briefly, the human hormone precursor proteins preproapelin and prestatherin can use Sec62 as well as SR for ER targeting in the cell-free assay. Interestingly, prestatherin preferred SRα over Sec62-mediated targeting, whereas preproapelin did the opposite, which was attributed to the higher hydrophobicity of the prestatherin SP (apparent ΔG −0.91 versus −0.19). Taken together with the observation that carboxy-terminal extension (by 187 amino acid residues) of preproapelin or prestatherin by the cytosolic protein DHFR leads to Sec62 independence, our data support the notion that small presecretory proteins use the SRP pathway for Sec61 targeting in human cells inefficiently, because the corresponding nascent chains are prone to be released from ribosomes before SRP can interact (Müller and Zimmermann, 1987; Schlenstedt and Zimmermann, 1987; Müller and Zimmermann, 1988; Schlenstedt et al., 1990; Lakkaraju et al., 2012; Haßdenteufel et al., 2018). Therefore, these precursors rely on alternative targeting pathways. In addition to SR and Sec62, co- and posttranslational targeting of preproapelin and prestatherin can also involve the TRC- and the SND-pathway, albeit with different efficiencies (Haßdenteufel et al., 2018). An ortholog of Snd1 has not yet been characterized in human cells. Recently, precursors of TRPC6 and various GPI-anchored proteins, such as CD55, CD59, and CD109 were added to the growing list of SND-clients (Talbot et al., 2019; Yang et al., 2021).

We applied the established proteomic strategy to identify precursor polypeptides that depend on hSnd2 for their targeting to the ER (Tirincsi et al., 2022b). Applying the established statistical analysis, we found that transient and partial hSnd2 depletion significantly affected the steady-state levels of 76 proteins: 43 negatively and 33 positively (Figure 6; Supplementary Table S1). Among the negatively affected proteins, GO terms assigned roughly 47% to organelles of the endocytic and exocytic pathways. We also detected a small enrichment of N-glycosylated proteins and a large one of membrane proteins. The negatively affected proteins included three proteins with cleavable SP (all being membrane proteins), and nine membrane proteins with TMH, including TA membrane proteins (such as Cytochrome b₅) plus single-spanning and multi-spanning membrane proteins (such as TRPM7), thus confirming previously observed classes of hSnd2 clients (TRPC6, Cytochrome b₅). Thus, there seems to be a preference of the human SND system for membrane protein precursors (Figure 6). When the SPs of hSnd2-dependent precursor polypeptides were analyzed for hydrophobicity, GP content, and segmentation, no significant distinguishing features were determined. However, when more hSnd2 clients were identified under conditions of simultaneous depletion of hSnd2 and WRB (Figure 6) a preference of hSnd2 for multispanning membrane proteins became visible and more hSnd2 membrane protein clients were found to have relatively more central or carboxy-terminal TMHs than SRA dependent membrane proteins (Tirincsi et al., 2022b), the latter two aspects being consistent with results for the yeast SND targeting pathway (Aviram et al., 2016). These observations are consistent with the fact that only little overlap between SRA clients and clients of the SND and TRC pathways was detected (Figure 6) and may explain why the latter two pathways can partially substitute for each other. In addition to SND clients, simultaneous depletion of hSnd2 and WRB negatively affected the ER membrane protein TMEM109, which was subsequently characterized as the hitherto elusive hSnd3 in experiments that were addressing its interaction with hSnd2 as well as its role in ER protein import (Tirincsi et al., 2022b).

Furthermore, recent work characterized the PEX19/PEX3-dependent pathway as a fourth pathway for targeting of precursor polypeptides to the ER (Schrul and Kopito, 2016; Yamamoto and Sakisaka, 2018). PEX3 was originally characterized as peroxisomal membrane protein, which cooperates with the cytosolic protein PEX19 in targeting of peroxisomal membrane proteins to pre-existent peroxisomes and in facilitating their membrane insertion (Erdmann et al., 1989; Hettema et al., 2000; Schmidt et al., 2012). As it turned out, however, PEX3 is also present in discrete subdomains of ER membranes and is involved in targeting of certain precursor proteins to ER membranes and most likely in their membrane insertion (Schrul and Kopito, 2016; Yamamoto and Sakisaka, 2018). These precursor proteins include membrane proteins, which either remain in the ER (the two-hairpin or reticulon-domain containing proteins ARL6IP1, RTN3A, RTN4C) or are pinched off in lipid droplets (such as the hairpin protein UBXD8) (Schrul and Kopito, 2016; Yamamoto and Sakisaka, 2018). These observations raised the question if this pathway, too, plays a more global role in protein targeting to the ER (Schrul and Schliebs, 2018; Jansen and Klei, 2019; Dhimann et al., 2020; Goodman, 2020).

Therefore, we addressed the client spectrum of PEX3 in ER protein targeting in human cells and asked if the PEX19/PEX3 pathway to the ER can also target precursor polypeptides to the Sec61 complex (Zimmermann et al., 2021). Here, the approach involved PEX3-depleted HeLa cells and chronically PEX3-deficient Zellweger patient fibroblasts (Schmidt et al., 2012). The negatively affected proteins found in the PEX3 knock-down or knock-out cells included seven peroxisomal membrane proteins and two hairpin proteins of the ER (ATL1, RTN3), thus confirming the two previously identified classes of PEX19/PEX3 clients for ER targeting in human cells (Figure 6; Supplementary Table S1). In addition, 18 membrane proteins (including TA proteins) and 28 proteins with SP (most notably 14 collagens plus collagen-related proteins as well as five membrane proteins) and belonging to the secretory pathway were negatively affected. The latter findings support the notion that PEX3 indeed represents a fourth pathway for targeting of precursor polypeptides to the Sec61 complex. Furthermore, it may suggest a hitherto unknown spatial or at least physical relationship between ER subdomains that are involved in ER shaping and the budding of peroxisomal precursor vesicles, large cargo vesicles, and lipid droplets. Thus in analogy to KTN1, the key observation may be that PEX3 plays a role in targeting of certain precursor polypeptides to Sec61 complexes in ER subdomains.

The heterotrimeric Sec61 complex provides an entry point for precursor polypeptides with SPs into the ER. In the course of co- and posttranslational membrane translocation, the SPs of precursor polypeptides first approach the Sec61 channel (Gumbart and Schulten, 2007; Lang et al., 2017; Lang et al., 2019; Bhadra and Helms, 2021). Subsequently, they begin to sample the cytosolic funnel of the Sec61 channel. According to molecular dynamics simulations, sampling in the Sec61 channel is affected by various properties of precursors or their mRNAs, i.e., deleterious charges, hydrophobicity, mature protein length, arrest peptides or poly-proline motifs in the precursor polypeptides and translation speed, which is dependent on pause sites, rare codons or hairpins in the mRNAs (Zhang and Miller, 2012). For productive SP insertion into and simultaneous full opening of the Sec61 channel, comparatively high hydrophobicity and, therefore, low apparent ΔG value for the H-region was found to be conducive (Gumbart and Schulten, 2007; Zhang and Miller, 2012; Bhadra and Helms, 2021). H-region hydrophobicity of the SP or TMH is supposed to be recognized by the hydrophobic patch formed by four residues of Sec61α TMHs 2 and 7, which line the lateral gate of the channel (Voorhees et al., 2014; Voorhees and Hegde, 2016).

In our opinion, gating of the Sec61 channel can best be described in analogy to an enzyme-catalyzed reaction where the precursor polypeptides with their SPs are the catalysts and the channel is their substrate (Figures 1D, 7) (Haßdenteufel et al., 2018): Channel opening and closing represent two energetically un-favorable reversible reactions and the clients with or without support from auxiliary components or allosteric effectors (TRAP, Sec62/Sec63, see below) are the co-catalysts, which lower the activation energy for the required conformational transitions by binding to the Sec61 complex (Lang et al., 2017; Lang et al., 2019). Interestingly, there are SP mutations of certain precursor polypeptides, such as preproinsulin, preprorenin, as well as SEC61A1 mutations that can cause the same hereditary diseases, such as Diabetes mellitus and ADTKD (Lloyd et al., 2010; Guo et al., 2014; Bolar et al., 2016; Devuyst et al., 2019; Sicking et al., 2022). Again, this can best be described by an energy diagram for Sec61 channel gating (Figure 7). Accordingly, substitutions of crucial amino acids in either SPs or the pore-forming α-subunit of the Sec61 channel may increase the activation energy for Sec61 channel opening and, therefore, slow down ER import of the particular precursor polypeptide or a whole group of precursor polypeptides, which is particularly dependent on a certain amino acid residue in the Sec61 channel. Notably, SEC61A1 mutations that cause ADTKD are discussed in the context of additional Sec61-channelopathies in more detail below in the Discussion and were recently reviewed by Sicking et al. (2021a).

As mentioned above, the depletion of Sec61α originally served as a proof-of-principle for the proteomic approach (Figures 4, 8; Supplementary Table S1). Among the negatively affected proteins that included all three subunits of the Sec61 complex, GO terms assigned 61% to organelles of the pathways of endocytosis and exocytosis, thus representing a firm enrichment compared to the value for the total quantified proteome (26%) (Nguyen et al., 2018). Furthermore, significant enrichment of precursor proteins with SP (6.8-fold), N-glycosylated proteins (5.6-fold), and membrane proteins (3.0-fold) was detected for the negatively affected proteins (Supplementary Table S1). This suggests that the precursors of these negatively affected proteins, 198 with SP (including 80 membrane protein precursors) and 90 with TMH, represent clients of the Sec61 channel and, therefore, were degraded by the cytosolic proteasome upon Sec61 depletion (Figure 8). As also expected, the positively affected proteins included potential compensatory components, including the two subunits of the SRP receptor (Nguyen et al., 2018). When we analyzed the physicochemical properties of the SPs of the Sec61 clients, precursors with less-hydrophobic SPs were more strongly affected by Sec61 absence, i.e., over-represented in the negatively affected polypeptides, suggesting that precursor polypeptides with a higher SP hydrophobicity are more efficient in Sec61 channel opening than those with lower hydrophobicity. Comparison of Sec61 clients with those of the membrane protein insertase EMC confirmed the preference of the latter for membrane protein precursors (Figure 8) (Shurtleff et al., 2018; Tian et al., 2019). However, the EMC data would also be consistent with the idea that EMC may also be able to facilitate Sec61 channel opening for precursors of soluble proteins with weak SPs, i.e., in analogy to TRAP and Sec62/Sec63 (see below).

While the Sec61-complex mediates import of most precursor polypeptides into the ER, the Sec61-associated Sec62/Sec63 heterodimer supports ER protein import in a client-specific manner. Direct interaction between the Sec61 complex and Sec63 was demonstrated by co-immunoprecipitation as well as in living human cells (Tyedmers et al., 2000; Sicking et al., 2021b). Recently, four studies addressed the architecture of the posttranslationally acting translocon complex in yeast by cryo-electron microscopy (cryo-EM) (Itskanov and Park, 2019; Wu et al., 2019; Weng et al., 2021; Itskanov et al., 2021). This particular translocon represents a heptameric protein ensemble, termed the SEC complex (Deshaies et al., 1991). In the SEC complex the trimeric Sec61 complex is assembled with the tetrameric Sec62p/Sec63p complex. The latter comprises two essential, evolutionarily conserved subunits, the membrane proteins Sec62p and Sec63p, and two non-essential subunits, Sec71p and Sec72p. The data provided insights into the mechanism how the SEC complex allows gating of the Sec61 complex and supports ER protein import. Most informative were the observed interactions between Sec63p and the Sec61 complex, which include contacts in the cytosolic, membrane and lumenal domains. Strikingly, the cytosolic Brl domain of Sec63p contacts loops 6 and 8 of Sec61α, thereby blocking the ribosome binding site. Interestingly, as was structurally predicted for the interaction of the TRAPα/β subunits with the Sec61 complex (Pfeffer et al., 2017) and supported by Alphafold 2 (Jumper et al., 2021), the Brl domain of Sec63p represents a canonical beta-sandwich fold to allow an antigen-antibody-like binding to loop 6 of Sec61α. In the membrane, Sec63p (specifically TMH 3 of Sec63) contacts all three subunits of the Sec61 complex in the hinge region opposite of the lateral gate, including TMHs 5 and 1 of Sec61α as well the TMHs of Sec61β and Sec61γ (Figure 1D). Additionally, the short lumenal amino-terminus of Sec63p intercalates on the lumenal side of the channel between the hinge loop 5 of Sec61α and Sec61γ (Itskanov and Park, 2019; Wu et al., 2019; Itskanov et al., 2021; Weng et al., 2021). Apparently, binding of Sec62p/Sec63p to the Sec61 channel causes wide opening of the lateral gate (Van den Berg et al., 2004; Voorhees et al., 2014; Voorhees and Hegde, 2016). The functional implications for the SEC translocon as a consequence of gating by the Sec62p/Sec63p are that SP of many substrates are less hydrophobic and, therefore, have a lower chance to enter the lateral gate and trigger complete opening of the channel. Thus, in the SEC complex binding of the Sec62p/Sec63p induces a fully opened channel that readily accommodates even “weak” or inefficiently gating SPs (Ng et al., 1996; Trueman et al., 2011). Consistent with the concept of the Sec62p/Sec63p inducing wide opening of the lateral gate, yeast Sec62p was found to be able to aid in membrane topology of moderately hydrophobic signal anchor proteins, in particular single-spanning type II membrane proteins, which perform the energetically unfavorable 180° flip turn for correcting their initial type I orientation (Reithinger et al., 2013; Jung et al., 2014; Jung and Kim, 2021).

Similar to yeast, analyses of protein transport in mammalian cells showed a client-specific role of Sec62 in ER protein import. According to in vitro experiments with model proteins from insects (such as preprocecropin A) and humans (such as preproapelin and prestatherin), the ER import of presecretory proteins with a content of less than 100 amino acid residues (termed small precursor proteins) into the mammalian endoplasmic reticulum (ER) can occur posttranslationally (Schlenstedt et al., 1990; Shao and Hegde, 2011; Lakkaraju et al., 2012; Johnson et al., 2013; Haßdenteufel et al., 2018) and involves various targeting mechanisms (Haßdenteufel et al., 2018) as well as the ER-membrane proteins Sec62 and Sec63 (Lakkaraju et al., 2012; Lang et al., 2012; Johnson et al., 2013; Haßdenteufel et al., 2017, 2018, and 2019). In case of preprocecropin A, posttranslational ER import has been observed in intact human cells (Shao and Hegde, 2011) and Sec62-dependence of small human presecretory proteins was observed in intact human cells (Lakkaraju et al., 2012).

In contrast to yeast, however, the mammalian Sec62 protein apparently experienced a gain of function, i.e., it can interact with ribosomes near the ribosomal tunnel exit and can support cotranslational transport of certain clients, such as the precursors of ERj3- and prion-protein with 358 and 253 amino acid residues, repectively (Müller et al., 2010; Ziska et al., 2019; Schorr et al., 2020). Therefore, crosslinking experiments with stalled precursor polypeptides in transit through the mammalian translocon observed the dynamic recruitment of allosteric Sec61 channel effectors, such Sec62 (Conti et al., 2015). In contrast, the model precursor bovine preprolactin triggered Sec61 recruitment of accessory factors such as the allosteric effector TRAP and the auxiliary translocating chain-associating membrane (TRAM) protein. However, when ERj3- or prion-protein were used as model transport substrates the Sec62/Sec63 instead of TRAP and TRAM were recruited to the channel in order to allow translocation of substrates having a weak or inefficiently gating SP (Conti et al., 2015). In other studies, another dynamic transition of the translocon was observed for Sec62 and the SRP receptor. To allow cotranslational targeting the SR can displace Sec62 from the Sec61 complex, thereby switching the Sec61 channel from Sec62- to SRP-dependent translocation (Jadhav et al., 2015). According to the above mentioned crosslinking approach, however, SR and Sec62 can also act sequentially, namely after SRP-dependent targeting of precursors of ERj3- and prion-protein, Sec62 can displace SR from the Sec61 channel and -together with Sec63- support channel opening. Furthermore, Sec63 has to “take over” loops 6 and 8 of Sec61α from the ribosome. The cryo-EM structures of the yeast SEC complex may support the idea of a dynamic transition and flexibility of Sec62/Sec63. Both Sec62p and the ER lumenal J-domain of Sec63p could not be sufficiently resolved in the single particle analysis and this might have been due to their structural flexibility and dynamic assembly into the SEC translocon.

Notably, for the prion protein precursor, Sec62/Sec63-dependent ER-import has also been demonstrated in a genetic screen in human cells (Davis et al., 2015). Subsequent in vitro import assays, using a full-length prion protein precursor, demonstrated SRP-dependence and the fact that Sec63-dependence is not only due to the SP but also due to a polybasic motif, which is downstream of the SP in the mature region (Ziska et al., 2019) and was missing from the artificial prion protein construct in previous work (Rane et al., 2009; Davis et al., 2015). Furthermore, these in vitro import assays demonstrated that Sec63-dependence of the small preproapelin and the precursors of ERj3 and prion protein is related to gating of the Sec61 channel to the open state and coincides with BiP´s involvement, which was linked to the combination of a weak SP plus, in case of preproapelin (₃₇RRK) and the prion protein precursor (₁KKRPK), a positively charged cluster in the mature region (Haßdenteufel et al., 2018; Ziska et al., 2019). Notably, loss of Sec63 protein function in the liver of a subset of human patients with polycystic liver disease was also interpreted in light of a client specific function of Sec63 in ER protein import (Fedeles et al., 2011; Lang et al., 2012). Interestingly, Sec63-dependence of ERj3 import into the ER was indirectly confirmed in murine SEC63 null cells, which were generated as an animal model for the human disease (Lang et al., 2012). These SEC63 ^−/− cells lacked ERj3 while the levels of various other ER proteins were unchanged compared to murine SEC63 ^+/+ cells (Fedeles et al., 2011; Schorr et al., 2020).

With our proteomic approach, we determined the rules for engagement of Sec62/Sec63 in ER-import in intact human cells (Schorr et al., 2020). Applying the statistical analysis, we found that Sec62 depletion significantly affected the steady-state levels of 351 proteins: 155 negatively and 196 positively (Figure 9; Supplementary Table S2). The identified precursors included 18 proteins with cleavable SP and six proteins with TMH. The proteins positively affected by transient Sec62 depletion included both SRP- receptor subunits (SRPRA, SRPRB) and the TRAP ß-subunit (coded by the SSR2 gene). We assume that these short-term compensatory mechanisms may have contributed to the comparatively low number of negatively affected proteins. The Sec63-depletion significantly affected the steady-state levels of 34 proteins: 21 negatively and 13 positively (Figure 9; Supplementary Table S2). The identified precursors included four proteins with cleavable SP and six proteins with TMH and only one of the proteins with SP had also been negatively affected by Sec62-depletion (TGFBI). Upon closer inspection of the potential overlap between Sec62 and Sec63 depletion in HeLa cells, four additional precursor polypeptides with SP were negatively affected by Sec63-depletion in HeLa cells, which, however, did not meet the significance threshold (ERj3 –coded by the DNAJB11 gene-, MAGT1, PDIA5, SDF2) (Schorr et al., 2020).

To identify additional substrates, we performed similar analyses after Sec62 or Sec63 knock-out, employing CRISPR/Cas9 treated HEK293 cells compared to HEK293 control cells (Fumagalli et al., 2017; Schorr et al., 2020). Here, we found that Sec62 deficiency significantly affected the steady-state levels of 329 proteins: 208 negatively and 121 positively (Supplementary Table S2). Of the negatively affected proteins, GO terms assigned ∼48% to organelles of the endocytic and exocytic pathways. We also detected significant enrichment of proteins with SP (5.5-fold), N-glycosylated proteins (4.5-fold), and membrane proteins (1.8-fold). The identified precursors included 74 proteins with cleavable SP (including 19 membrane proteins) and 29 proteins with TMH. As expected (Conti et al., 2015), ERj3 was negatively affected. After Sec63 knock-out in HEK293 cells, we found that Sec63-deficiency significantly affected the steady-state levels of 302 proteins: 199 negatively and 103 positively (Supplementary Table S2). GO terms assigned ∼37% of the negatively affected proteins to organelles of the endocytic and exocytic pathways. We also detected significant enrichment of proteins with SP (1.9-fold), N-glycosylated proteins (2.4-fold), and membrane proteins (1.8-fold). The identified precursors included 24 proteins with cleavable SP (including ten membrane proteins) and 38 proteins with TMH (Figure 9). Here, 22 precursor polypeptides were negatively affected by Sec62 as well as Sec63 deficiency, 11 each with SP (including six membrane proteins) and with TMH, all belonging to the secretory pathway (Figure 9). Upon closer inspection of the potential overlap between Sec62 and Sec63 in HEK293 cells, six additional precursor polypeptides with SP were negatively affected by Sec63-deficiency, which did not meet the significance threshold (MAGT1, PDIA5, RNASET2, SDF2, SIL1, TMED5) (Schorr et al., 2020). Interestingly, the analysis also identified 54 precursors with SP (including 11 membrane proteins) plus 17 precursors with TMH, which showed a requirement for Sec62 but not for Sec63, as well as 12 precursors with SP (including four membrane proteins) plus 26 precursors with TMH, which showed a requirement for Sec63 but not for Sec62, consistent with the previous observation in vitro that the two proteins can also support ER protein import independently of each other (Haßdenteufel et al., 2018). Notably, the overlap was probably underestimated since the analysis was done for mixed guide RNA clones in the case of Sec63, i.e., the cell pool did not represent a SEC63 knock-out (Schorr et al., 2020).

Along with confirming ERj3 as a client, 30 novel Sec62/Sec63-clients were identified under these in vivo-like conditions, 18 with SP (including eight membrane proteins) and 12 with TMH (Figure 9; Supplementary Table S2). These previously unknown substrates have in common less hydrophobic SPs with longer but less hydrophobic H-regions and lower C-region polarity (Figures 10A,C). Further analyses with four substrates, ERj3 in particular, revealed the combination of a weak SP and a translocation-disruptive positively charged cluster of amino acid residues within the mature part (₂₀KKAYRK) as decisive for the Sec62-/Sec63-requirement (Figure 10E) (Schorr et al., 2020). This is reminscent of preproapelin and prion protein import (Haßdenteufel et al., 2018; Ziska et al., 2019) and in all three cases these features were found to be responsible for an additional BiP-requirement and for sensitivity towards the Sec61 channel inhibitor CAM741. Thus, human Sec62/Sec63 may support Sec61 channel opening for precursor polypeptides with weak SPs by direct interaction with Sec61α and/or via recruitment of BiP and its interaction with the ER-lumenal loop 7 of Sec61α, which we supposed to lower the activation energy for channel opening (Figure 7).

ERj1 belongs to the class of ribosome-associated membrane proteins (RAMPs) (Dudek et al., 2002; Blau et al., 2005; Dudek et al., 2005; Benedix et al., 2010). However, its ribosome association, appears to be more dynamic as compared to the classical RAMPs, i.e., Sec61, TRAM and TRAP (Görlich et al., 1992b; Görlich and Rapoport, 1993). This was microscopically confirmed by fluorescence microscopy using fluorescently labeled antibodies against ERj1 in permeabilized MDCK cells (Snapp et al., 2004; Benedix et al., 2010). According to cryo-EM, the cytosolic domain of ERj1 binds at the ribosomal tunnel exit and involves expansion segment 27 (ES27) of the 28S rRNA (Blau et al., 2005). ERj1 was proposed to play a role in ER protein import as a possible functional homolog for Sec62/Sec63, combining the cytosolic ribosome binding activity of Sec62 with the ER lumenal Hsp40-type co-chaperone activity of Sec63 in one polypeptide (Dierks et al., 1996; Skowronek et al., 1999; Mayer et al., 2000; Tyedmers et al., 2000; Dudek et al., 2005; Tyedmers et al., 2005; Müller et al., 2010; Lang et al., 2012; Schäuble et al., 2012). Notably, human Sec62 was microscopically confirmed as RAMP, too, by fluorescence microscopy using fluorescently labeled antibodies against Sec62 in permeabilized MDCK cells (Snapp et al., 2004; Müller et al., 2010). Interestingly, the cytosolic domain of ERj1 is able to allosterically inhibit translation at the stage of initiation when its ER lumenal J-domain is not associated with BiP but allows translation when BiP is bound (Benedix et al., 2010). Thus, ERj1 would be perfectly able to allow initiation of protein synthesis of precursor polypeptides on ER bound ribosomes when BiP is available on the ER lumenal side of the membrane.

Employing the statistical analysis, we found that transient and partial ERj1 depletion significantly affected the steady-state levels of 172 proteins: 92 negatively and 80 positively (Figure 5; Supplementary Table S2) (Bhadra et al., 2021a). Of the negatively affected proteins, GO terms assigned almost 30% to organelles of the pathways of endocytosis and exocytosis. The identified precursors included seven proteins with cleavable SP, among them two membrane proteins, and eight membrane proteins with TMH and were discussed above in the context of KTN1-dependent mRNA targeting to the ER.

Originally, TRAP was characterized as signal-sequence receptor (SSR) complex (Wiedmann et al., 1987). Furthermore, it had been crosslinked to nascent polypeptides at late translocation stages (Conti et al., 2015) and had been demonstrated to associate with Sec61 (Menetret et al., 2008; Dejgaard et al., 2010; Pfeffer et al., 2017). As mentioned in the Introduction, the ribosome-associated Sec61-complex and the TRAP form a stable stoichiometric super-complex called a translocon (Menetret et al., 2008; Bano-Polo et al., 2017; Pfeffer et al., 2017). In vitro transport studies showed that the TRAP stimulates protein translocation depending on the efficiency of the SP in transport initiation (Fons et al., 2003); Sec61 gating efficiency and TRAP dependence were inversely correlated. Recent studies in intact cells suggest that TRAP may also affect TMH topology (Sommer et al., 2013), reminiscent of Sec62/Sec63 in yeast (Reithinger et al., 2013; Jung et al., 2014).

To identify TRAP dependent precursors, we combined siRNA-mediated TRAP depletion in HeLa cells, label-free quantitative proteomics, differential protein abundance analysis, and statistical analysis. We found that TRAPβ depletion significantly affected the steady-state levels of 257 proteins: 180 negatively and 77 positively. Of the negatively affected proteins, GO terms assigned ∼40% to organelles of the endocytotic and exocytotic pathways and included all four subunits of TRAP. We also detected significant enrichment of proteins with SP (3.3-fold), N-glycosylated proteins (2.7-fold), and membrane proteins (2.1-fold). The identified precursors included 38 proteins with cleavable SP and 22 proteins with TMH, and represented N-glycosylated proteins and non-glycosylated proteins (Figure 9; Supplementary Table S2). For TRAP deficient fibroblasts from patients with a Congenital disorder of glycosylation the steady-state levels of 318 proteins were altered: 279 negatively and 39 positively (Supplementary Table S2). Of the negatively affected proteins, GO terms assigned 36% to organelles of the pathways of endocytosis and exocytosis. The identified precursors included 34 proteins with cleavable SP and 41 proteins with TMH. Taken together, TRAP knock-down and knock-out identified 59 membrane proteins with TMH and 66 proteins with SP (including 20 membrane proteins), all belonging to the secretory pathway, as TRAP clients (Figure 9). Interestingly, six of the TRAP clients were found among the Sec62/Sec63 clients and nine among the Sec62 substrates, consistent with the view that these two allosteric Sec61 channel effectors have overlapping but non-identical functions.

The SP analysis of TRAP-substrates demonstrated an above-average glycine-plus-proline content (GP content) and below-average hydrophobicity as the key features (Figures 11A–C). Thus, the Sec61-associated TRAP supports protein translocation in a substrate-specific manner. We suggest that high GP content and low hydrophobicity extend the dwell time of SP at the cytosolic funnel of the Sec61 channel, and that TRAP can compensate this potential problem by stabilizing SP on the cytosolic surface and by aiding in Sec61 channel gating at the lumenal side. This raises the question of how TRAP relays the presence of an SP-bearing RNC to the Sec61 channel. In an attempt to interpret our findings at the structural level, i.e., in the context of the TRAP architecture, in which individual TRAP subunits were assigned positions within the overall density of human TRAP in native ER membranes by cryo-electron tomography (CET) (Figures 11E,F) (Pfeffer et al., 2017), the ER-lumenal domains of the TRAPαβ-subcomplex contact loop 5 in the hinge region between the amino- and carboxy-terminal halves of Sec61α and, thereby, mediate Sec61 channel opening by lowering the activation energy, required for channel opening (Figure 7). TRAPγ occupies a central position in human TRAP, contacting eL38 and short rRNA expansion segment (ES) on the ribosome, thus coordinating the other TRAP subunits with the ribosome and the additional translocon components, i.e. the Sec61-complex (contacted by TRAPαβ) and OST (contacted by TRAPδ). Previously, the ribosomal components uL24 and H59, both in vicinity to eL38 and TRAPγ, were observed to coordinate SP for SRP binding in the bacterial system (Jomaa et al., 2016). Assuming a similar SP position in the human system, the amino-terminal SP tip may consequently be close enough to interact with eL38 and the cytosolic domain of TRAPγ during the “hand-over” of the SP from SRP to Sec61 (Figure 11E). According to this hypothetical scenario, TRAP may support the insertion of SP into the Sec61 channel in the productive hairpin (rather than head-first) configuration.

TRAM (since the discovery of TRAM2 (Stefanonvic et al., 2004) termed TRAM1) represents an ER membrane protein with eight TMHs. It belongs to a protein family, characterized by the TLC (short for: TRAM/LAG1/CLN8) homology domain, which is supposed to bind ceramide or related sphingolipids (Klein et al., 2020). Similarly to TRAP, it was discovered by crosslinking of nascent presecretory proteins in the context of RNCs, but in contrast to TRAP early in their translocation into the ER (Görlich et al., 1992b; Görlich and Rapoport, 1993). Subsequently, it was described to interact with nascent membrane proteins in the course of their initial integration into the Sec61 channel (High et al., 1993; Mothes et al., 1994; Do et al., 1996; Voigt et al., 1996; Hegde et al., 1998; McCormick et al., 2003; Sadlish et al., 2005; Sauri et al., 2007). Actually, TRAM1 was one of the first proteins found to provide substrate-specific support for ER protein import (Görlich and Rapoport, 1993). Furthermore, it was observed that precursor proteins with short charged amino-terminal domains in their SPs require TRAM1 for efficient insertion into the lateral gate and that TRAM1 can regulate cytosolic extrusion of nascent chain domains into the gap between ribosome and translocon (Voigt et al., 1996; Hegde et al., 1998). In addition, it was concluded that precursors with shorter than average N-regions and shorter H-regions in their SP require the help of TRAM1 for efficient insertion into the lateral gate.

By applying our unbiased proteomic approach, we identified 30 potential TRAM1 substrates that included 13 precursors with SP (including four membrane proteins) and 17 with TMH (Figure 9; Supplementary Table S2) (Klein et al., 2020). Comparing these precursors to those found for Sec61 and TRAP in similar experiments, did not point to a preference of TRAM1 for any particular type of precursor polypeptides. Furthermore, analysis of the physicochemical properties of SP and TMH of the TRAM1 substrates did not point to a specific feature, except for precursors with short N-regions in their SP. Notably, 27% of the TRAM1 substrates were also negatively affected by TRAP depletion (Figure 9). This is consistent with the co-localization of TRAM1 with Sec61 and TRAP (Görlich and Rapoport, 1993; Dejgaard et al., 2010). Considering the strong overlap in substrates of these two transport components and TRAM1’s apparent lack of precursor preference may indicate that TRAM1 does not act as a receptor for SPs and TMHs. It may rather play a supportive role in ER protein import, such as making the phospholipid bilayer conducive for accepting SP and TMH in the vicinity of the lateral gate of the Sec61 channel. This interpretation is consistent with the above-mentioned prediction that TRAM1 may be able to bind sphingolipids (Klein et al., 2020).