Selective ribosome profiling reveals a role for SecB in the co-translational inner membrane protein biogenesis

The chaperone SecB has been implicated in de novo protein folding and translocation across the membrane, but it remains unclear which nascent polypeptides SecB binds, when during translation SecB acts, how SecB function is coordinated with other chaperones and targeting factors, and how polypeptide engagement contributes to protein biogenesis. Using selective ribosome profiling, we show that SecB binds many nascent cytoplasmic and translocated proteins generally late during translation and controlled by the chaperone trigger factor. Revealing an uncharted role in co-translational translocation, inner membrane proteins (IMPs) are the most prominent nascent SecB interactors. Unlike other substrates, IMPs are bound early during translation, following the membrane targeting by the signal recognition particle. SecB remains bound until translation is terminated, and contributes to membrane insertion. Our study establishes a role of SecB in the co-translational maturation of proteins from all cellular compartments and functionally implicates cytosolic chaperones in membrane protein biogenesis.


INTRODUCTION
Efficient synthesis of folded, correctly localized proteins is critically important for cellular integrity and is supported by multiple chaperones and targeting factors that often engage nascent polypeptides during translation. [1][2][3][4] Bacteria employ three major routes of protein maturation that are specific for the different cellular compartments. About 70%-80% of newly synthesized proteins fold in the cytosol, generally supported by a network of molecular chaperones. The remaining 20%-30% are either inserted into the inner membrane by the co-translational translocation pathway or translocated across the inner membrane to the periplasm or the outer membrane by post-translational translocation. 5 Triaging and the initiation of folding or translocation begin early during protein synthesis and are mediated by a dynamic interplay of chaperones and targeting factors. 1,[6][7][8][9] In Escherichia coli, protein translocation into and across the cytoplasmic membrane depends on two major secretion pathways that employ specific targeting factors but converge at the ubiquitously used SecYEG translocon. 3,10 Co-translational translocation initiates early during protein synthesis by binding of the signal recognition particle (SRP) to ribosomes translating inner membrane proteins (IMPs). 11,12 SRP directly binds to ribosomes, 13 and this interaction is selectively stabilized by emerging trans-membrane domains (TMDs) of IMPs. 12 By interaction with the membrane-associated SRP receptor FtsY, SRP mediates the docking of ribosomes to the SecYEG translocon. 14 Membrane insertion of IMPs occurs co-translationally, largely energized by the translation process itself. Productive membrane integration of some IMPs furthermore requires the translocation activity of the ATPase SecA. 10,[15][16][17] Translocation of outer membrane proteins (OMPs) and periplasmic proteins (PPs) across the membrane involves the alternative, SRP-independent post-translational translocation pathway. 8,18,19 Here, substrates are recognized by virtue of an N-terminal hydrophobic signal peptide that is removed during translocation. Many of these nascent substrates are initially bound by the ribosome-associated chaperone trigger factor (TF), 20 and translocation is facilitated by the targeting factors SecA and in most proteobacteria, including E. coli, by the chaperone SecB. 21 SecA can engage nascent chains by binding to the ribosome, suggesting a co-translational initiation of substrate recognition and targeting. [22][23][24][25] This agrees with early findings that many, but not all, substrates are translocated post-translationally. 26 In contrast to the SRP-mediated co-translational translocation, the SecA-dependent translocation does not involve a direct docking of ribosomes to the translocon and is energized by the ATPase activity of the translocon-docked SecA. 27 Some chaperones have a bipartite function in protein folding and targeting, suggesting a partially functional integration of systems. One example is TF, which binds nascent cytosolic proteins (CPs), PPs, and OMPs. 20,28-31 TF may bind unfolded or loosely folded polypeptides to suppress premature and unproductive folding steps and coordinate the engagement of additional chaperones and targeting factors with the progress of translation. 1,9,32 The second, much less studied example is SecB, a non-essential, ATP-independent homotetrameric protein of 69 kDa with strong antifolding activity. 21,[33][34][35][36][37] Both chaperones, TF and SecB, bind to unfolded or largely loosely folded sections of proteins to prevent their premature folding and therefore contribute to protein homeostasis in the cell by functioning as a holdase. 21,32,35 The disc-shaped SecB exposes two major long hydrophobic grooves to bind multiple hydrophobic substrate segments and wrap the unfolded polypeptide around the SecB tetramer. 38,39 Supporting its role in post-translational translocation, SecB was originally identified in a genetic screen of mutants defective in the translocation of specific precursor proteins. 40 Furthermore, SecB binds to a number of presecretory proteins and directly binds and functionally cooperates with SecA and TF. 32,41 Finally, DsecB mutants display a strong cold-sensitive (Cs) phenotype below 23 C, a moderate temperature-sensitive (Ts) phenotype above 46 C, and increased protein aggregation. 42 Suggesting that TF and SecB act in the same pathway, altered levels of TF cause distinct growth consequences in absence of secB. The deletion of tig, encoding TF, reverses the Cs, Ts, and aggregation-prone phenotype of DsecB, while the toxicity of TF overexpression is exacerbated in SecB absence. 42,43 A direct role of SecB in protein folding of cytoplasmic proteins is suggested by the aggregation of some CPs in DsecB mutants, by the chemical crosslinking of SecB to CPs, and by the finding that overexpression of secB can suppress the aggregation of CPs and alleviate the heat-sensitivity of mutants lacking TF and the major Hsp70 chaperone DnaK. 42,44 Implicating SecB in additional cellular processes, secB homologs not only exist in proteobacteria but are also found in viral genomes, on plasmids, and in gram-positive bacteria that lack outer membrane and periplasm, and a SecBlike protein controls a specific toxin-antitoxin pair in Mycobacterium tuberculosis. 21 Several findings suggest that SecB exerts its role in protein targeting and translocation by binding to nascent chains: SecB comigrates with polysomes, reversibly engages nascent chains at a length of about 200 residues, 45 and can be crosslinked to ribosome nascent chain complexes (RNCs). 44 Furthermore, SecB preferentially interacts with unfolded segments of polypeptides [46][47][48] and was recently identified in a detailed proteomics analysis of the interactome of translating ribosomes. 49 In this study, we apply two ribosome profiling strategies to explore the role of SecB in co-translational protein folding and targeting, and the coordination of SecB with other maturation factors engaging translating ribosomes. Nascent chain-specific interaction profiles of SecB demonstrate that SecB constitutes an integral part of the chaperone network supporting protein folding and a general factor of protein translocation into and across the membrane. Supporting a so-far undiscovered role in membrane protein biogenesis, SecB promotes the targeting function of SRP and is critical for productive membrane insertion. Our study reveals that SecB constitutes a potent, co-translationally acting generalized chaperone and targeting factor that functionally contributes to membrane protein biogenesis and provides functional elasticity and redundancy to the proteostasis network.

RESULTS
SecB promiscuously binds a wide repertoire of nascent substrates To identify nascent substrates of SecB in vivo by selective ribosome profiling (SeRP), we generated a strain that chromosomally encoded a C-terminally Avi-tagged SecB (SecB-Avi) under control of an isopropyl b-d-1-thiogalactopyranoside (IPTG)-inducible promoter. To reach wild-type (WT) expression levels of fully biotinylated SecB-Avi, we optimized the concentration of IPTG and biotin in the media (Figures S1A, S1B, and S1C). Growth analysis at different temperatures demonstrated that WT levels of SecB-Avi fully complemented the chromosomal secB deletion ( Figure S1D). Using these conditions, we performed SeRP following previously described protocols. 50 Briefly, we harvested exponentially growing cells by rapid filtration and immediate freezing in liquid N 2 , prepared frozen lysates by mixer milling, treated the thawing lysate with micrococcal nuclease, and purified monosomes by sucrose cushion centrifugation. One part of the isolated monosomes was used to reveal the total translatome and the other part was used to isolate SecB-engaged monosomes using streptavidin beads ( Figure S2A). The isolated ribosome footprints of both samples were converted to a cDNA library, sequenced, and bioinformatically analyzed. Obtained datasets provide snapshot of mRNA positions of all translating ribosomes (total translatome) and all SecB-bound ribosomes (SecB-bound translatome) in the cell. According to sucrose cushion centrifugations followed by SecB-specific western blotting, we estimate that, in growing cells, about 5% of total SecB or SecB-Avi is engaging ribosomes ( Figure S2B).
A treemap plot representation of the total translatome (Figure 1A, left) shows that about 80% of the total active ribosome pool translates open reading frames (orfs) encoding CPs, and the remaining 20% translate orfs encoding translocated proteins. Indicating compartment-specific nascent substrate preferences of SecB, the orf distribution among the SecB-bound RNCs ( Figure 1A, right) is different from the total pool of ribosomes. Only 60% of SecB-bound RNCs synthesize CPs, suggesting that SecB, by virtue of its holdase activity, may contribute to the folding of nascent CPs, but prefers binding to translocated nascent chains. According to its established role in protein translocation across the membrane, SecB readily engages PPs and OMPs. Unexpectedly, SecB only slightly preferred nascent OMPs over all other proteins and PPs are not specifically enriched. Instead, we found an almost 3-fold enrichment of SecB on nascent IMPs (6.5% of IMPs in total translatome compared with 18% of IMPs in the SecB-bound translatome), revealing that IMPs are the preferred nascent substrate class of SecB. This strong preference for nascent IMPs points toward a so-far undetected function of SecB in co-translational targeting or membrane insertion of IMPs.
Next, we explored how nascent chain engagement is coordinated with translation on a metagene level ( Figure 1B), which reveals the averaged binding behavior of SecB to RNCs resolved by nascent chain length (further details are provided in the STAR Methods). Relevant SecB binding starts when ribosomes have reached codon 200 (and nascent chains reached a length of 200 residues). Considering that the C-terminal 30 residues are buried in the polypeptide exit tunnel of the ribosome, about 170 residues of the newly synthesized polypeptides are exposed on the ribosome surface. The SecB enrichment increases with ongoing translation and remains high until translation termination. This implies that SecB prefers longer nascent chains and contributes to protein maturation until translation is completed and possibly also post-translationally. By comparing metagene profiles of different subcellular localizations, we explored whether the length preferences of SecB may differ between substrate classes, indicating varying SecB binding behavior to nascent proteins of different subcellular localizations ( Figure 1B). The metagene binding curves of CPs, PPs, and OMPs rise slowly, exceed an enrichment value of 1 only after 200-300 codons, and reach the highest values at the end of translation. In contrast, SecB engagement of ribosomes translating IMPs is sharper, reaches higher values, and initiates much earlier, when nascent chains have a length of only about 80-100 amino acids (50-70 residues exposed). These results support established roles of SecB in coordinating the folding of CPs and translocation of PPs and OMPs by virtue of its holdase function, 42,44 and provide additional indications for a potential role of SecB in IMP targeting and insertion.

Quantitative analysis of SecB interactions with the nascent proteome
We next set out to define the nascent chain interactome of SecB, using a position-resolved binding analysis of SecB that employs a sliding window to calculate the 95% confidence interval (CI) of the enrichment for each position of a transcript. Calculating the CI along transcripts eliminates experimental noise and reveals length-resolved quantitative binding information. Furthermore, it allows determining the maximal binding of SecB to a nascent chain, termed binding score, defined as the highest value of the lower boundary of the CI. The binding score therefor reflects the general equilibrium constant of SecB binding to a nascent chain, as well as the sequencing depth (i.e., the confidence in the enrichment) of the given footprint. Based on binding scores of nascent chains, we attempted to define SecB substrates, similar to a previous study analyzing substrates of SRP. 12 However, unlike for SRP, the SecB binding scores gradually decrease across the nascent proteome ( Figure S2C), which makes it impossible to rationally define a threshold to distinguish SecBbound from all other nascent chains. We therefore decided to not classify proteins as SecB substrates, but instead generated a ranked list of all proteins sorted by SecB binding scores (Figures 1C and Table S1). Ranking proteins according to their maximum enrichment better reflects the function of generalized chaperones, which bind a wide range of nascent chains with different efficiencies. The distribution of proteins in the substrate list shows that membrane proteins in general, and in particular IMPs, are strongly enriched among the top interactors of SecB. In fact, among the 100 highest-ranked substrates are 85 membrane proteins (75 IMPs, 10 OMPs) and only 13 soluble proteins (12 CPs, one PP, two proteins of unknown localization). The strong bias persists among the 500 top SecB interactors (binding scores >3.048; Figures 1D and 1E). The enrichment of IMPs is not caused by the use of detergents during SeRP sample preparation and remained the same when detergents were omitted ( Figures S2D-S2F). Below rank 500, CPs dominate. Importantly, the binding profiles of proteins ranked below position 500 still reveal well-defined SecB interaction periods (Figures S2G-S2I), suggesting that SecB reproducibly engages a large repertoire of nascent chains, among them many CPs.
To further validate the SeRP findings with an independent approach, we performed an liquid chromatography-mass spectrometry (LC-MS) analysis of ribosomes purified by Strep-Tactinaffinity purification from WT cells and cells encoding SecB-Avi. As expected, SecB was the strongest enriched protein in the fraction purified from SecB-Avi-expressing cells, with over 120-fold change (FC) enrichment over an untagged control (Table S2). Although ribosomal proteins dominated the mass spectrometry samples, we additionally identified 84 proteins in the SecB-Avi pull-down sample that were significantly enriched (adjusted p value <0.05) and detected as SecB interactors in the SeRP data (binding scores higher than 3.048). Among these proteins were 52 CPs, 17 IMPs, six OMPs, two PPs, and seven proteins of unknown cellular localization. Again, IMPs were among the most strongly enriched nascent proteins in SecB-Avi-bound RNCs (more than 15-FC over control). Examples include the inorganic phosphate transporter PitA (protein FC of 3.5; SeRP binding score 10.75), the lipid core flippase MsbA (protein FC of 2.3; SeRP binding score 10.13), and the protein translocase subunit SecY (protein FC of 9.9; SeRP binding score 6.14). Another significantly enriched protein engaged with ribosomes and SecB was SecA (6.7 FC). SecA is not a strong nascent interactor of SecB (SeRP binding score 1.34), implying that mature SecA and SecB cooperate at the ribosome to support protein translocation.
SecB enhances the membrane engagement of IMP translating ribosomes The overrepresentation of IMPs among the most prominent SecB interactors suggests a role of SecB in co-translational targeting and membrane insertion. To understand how SecB contributes to this process, we developed a ribosome profilingbased protocol that determines the membrane association of translating ribosomes (MARseq). Similar to SeRP, we generated two deep-sequencing libraries, one comprising all ribosomes and the second one comprising only membrane-bound ribosomes, isolated by membrane flotation after gentle cell lysis. The ratio between both datasets reports for all nascent proteins whether and when RNCs engage the cell membrane. Revealing the high performance of the technique, we detected membrane engagement only for ribosomes synthesizing translocated proteins (IMPs, PP, OMPs), but not CPs (Figure 2A). The metagene profile of OMPs and PPs showed that, with ongoing translation, ribosomes were gradually enriched in the membrane fraction, indicating that translocation sometimes initiated co-translationally. As expected, the membrane engagement of ribosomes synthesizing IMPs was much more prominent and initiated earlier, when about 60 codons had been translated. After that, membrane enrichment steeply increased until it reached a stable level near codon 150, in agreement with the co-translational insertion of IMPs. Comparing MARseq and SeRP data showed that SecB binding and membrane engagement coincide at the metagene level and the level of single genes ( Figures 2B and 2C), suggesting that SecB rapidly binds to membrane-docked RNCs. To elucidate the role of SecB, we performed MARseq of DsecB mutants. Comparing MARseq metagene profiles of WT and DsecB mutants showed that the onset of the engagement of nascent IMPs at the cellular membrane was largely independent of SecB while the overall enrichment was reduced ( Figure 2D). This finding indicates that SecB may not be involved in the initial targeting but exerts its function when RNCs are already associated to the membrane.
SecB deletion reduces the efficiency of IMP membrane insertion The stabilizing effect of SecB binding on the translocon engagement of ribosomes opens the possibility that the chaperone contributes to IMP membrane insertion. To directly elucidate the contribution of SecB to IMP biogenesis, we quantitatively assessed the distribution of the cellular proteome among different compartments in WT and DsecB mutants by using a spatial proteomics screen based on the recently developed LOPIT-DC (localization of organelle proteins by isotope tagging after differential ultracentrifugation) technology. 51 This method has been shown to be highly effective in drawing spatial proteome maps of human, yeast, and bacteria. Moreover, the technology allows interrogating protein localizations that are difficult or impossible to reliably isolate with classical biochemical methods and has been shown to be capable of detecting more subtle changes in protein localizations (e.g., incomplete translocations or double localizations). 51 Briefly, we performed a series of centrifugation steps with increasing speed to separate total cell lysates into multiple fractions that were subjected to LC-MS (Figure 3A). The distribution patterns of confidently defined marker proteins among the collected fractions were then used to train a support vector machine that subsequently assigned protein localizations to all other proteins based on their specific distribution among the obtained fractions, as described in the pRoloc algorithm. 52 Comparing our localization information of 1,100 confidently quantified proteins of WT cells with experimentally verified protein localizations reported in STEPdb revealed that our method predicted the correct compartment with high accuracy. We next applied the method to explore protein localizations in DsecB cells. Interestingly, the sedimentation pattern of strong and weak SecB interactors among the CPs, OMPs, and PPs remained largely unchanged ( Figures 3B and 3C). In contrast to this, we detected a significant drift in the distribution of 24 IMPs that are prominently bound by SecB during translation (Figure 3D). As illustrated by principal-component analysis, the sedimentation behavior of these IMPs shifted from an IMP-typical pattern toward a pattern of proteins that are peripherally associated with the cytosolic side of the membrane ( Figure 3E). Changes of sedimentation behavior were also detectable for other proteins, but to a reduced extent. It is reasonable to assume that these proteins incompletely relocate, due to lower dependence on SecB activity. These data indicate that SecB binding to ribosomes synthesizing IMPs facilitates the integration of multiple nascent IMPs into the cytoplasmic membrane.

SecB and SRP synergistically act on translating ribosomes
To understand SecB's role in membrane insertion, we first focused on the coordination of SecB binding with the RNC bind-ing of SRP, which is the first step in IMP targeting. A previous SeRP study showed that SRP binding is closely correlated with the emergence of a TMD 12 from the ribosomal tunnel and that SRP sometimes skips the most N-terminal TMD of specific substrates. Some binding profiles of multi-spanning membrane proteins indicate multiple SRP engagement periods, agreeing with a mechanism that ribosomes transiently dissociate from the translocon 12,53-55 (for example, while cytoplasmic or periplasmic parts are translated) and get retargeted to the translocon by SRP when another TMD emerges. Comparing SecB and SRP binding profiles in a metagene profile (Figure 4A), in a binding map ( Figure 4A), and at the level of individual nascent IMPs (Figure 4C) revealed that SecB binding to RNCs generally occurs after the initial SRP engagement, also if SRP skips the first TMD of specific nascent IMPs 12 ( Figure S3). The binding profiles furthermore show that SecB remains associated with the ribosome when SRP dissociates or rebinds and until the termination of translation. Assuming that the first SRP engagement efficiently targets all nascent IMPs, these data suggest that SecB binds to RNCs that are already engaged with the membrane. The overlapping binding periods of SRP and SecB suggest a functional interplay of both factors and principally agree with two different scenarios: either the functional cooperation of both factors on the same nascent chain or the exclusive binding of both factors to different RNCs and substrate competition. To distinguish between both possibilities, we simultaneously overexpressed ffh, which encodes the protein component of SRP, and ffs, encoding the RNA component of SRP, as well as ftsY encoding the SRP receptor, and tested by SeRP the impact on SecB RNC binding ( Figure 5). We did not find any indication for a competition of SRP and SecB for nascent IMPs. Instead, elevated SRP and FtsY levels further enhanced the fraction of IMPs among the nascent SecB interactors from about 20% in WT cells to about 33% in SRP-overexpressing cells, indicating cooperation ( Figure 5A top and bottom). SRP overexpression did not delay the onset of SecB binding to nascent IMPs at the metagene and single-gene level (Figures 5B-5E). We next performed SeRP analyzing the SRP function in secB overexpressing cells. Similar to the impact of elevated SRP levels on SecB function, the overexpression of secB enhanced the binding frequency of SRP to nascent IMPs from about 30% to 45%, while OMPs, PPs, and CPs are less frequently bound ( Figure 5F top and bottom), and SecB overexpression did not affect the onset of SRP binding to emerging TMDs (Figures 5G-5I). Revealing that this impact of SecB on SRP function is directly caused by SecB binding to these nascent chains, the changes of SRP profiles were only detected for nascent IMPs that are bound by SecB and not for other nascent chains, including OMPs ( Figure 5J) SRP engagement for initial targeting and predominantly enhanced downstream SRP engagements, when IMP translating RNCs are already associated with the membrane (Figures 5G and 5J). These data agree with the model that elevated levels of SecB enhance the specificity of SRP binding to later TMDs of nascent IMPs. Finally, we analyzed SRP binding to RNCs in DsecB cells by SeRP. SRP binding properties were not detectably changed (Figure S4A), in agreement with the model that other chaperones may facilitate SRP binding to RNCs at the membrane. Supporting this possibility, multiple cytosolic chaperones, including DnaK, are overexpressed in DsecB cells and DnaK functionally overlaps with SecB in protein translocation: DnaKJ overexpression rescues the export of SecB substrates in SecA mutant cells 56 and the Cs of DsecB mutants. 9,42 Furthermore, SecB and DnaK share similar substrate binding preferences 48 and also bind an overlapping pool of substrates. 57 Importantly, our spatial proteomics data demonstrate that these chaperones cannot fully compensate SecB function in IMP maturation (Figure 3).

SecB binding to RNCs is independent of SecA
Previous data revealed that SecA binds translating ribosomes and also contributes to the productive membrane insertion of some IMPs. 10,[15][16][17][22][23][24][25] Considering that SecA and SecB interact, we tested whether the interaction with SecA may be critical for the SecB recruitment to RNCs by analyzing the RNC engagement of the mutant SecB E77K , which has reduced affinity for SecA. [58][59][60] We did not detect any relevant differences in RNC engagement of WT SecB and SecB E77K , suggesting that the selective binding of SecB to RNCs does not depend on high-affinity binding to SecA but may be an intrinsic property of SecB ( Figure S4B).

TF coordinates SecB binding to cytosolic and translocated proteins
In agreement with the established role of SecB in the folding of CPs and the membrane translocation of PPs and OMPs, 42,44 our SeRP showed that SecB prominently engages nascent proteins from all major compartments (Figures 1C and 1D Table S1). To better understand how SecB executes this function during translation and in the context of other co-translationally acting factors, we analyzed the temporal coordination of SecB with TF, which engages most CPs, OMPs, and PPs. 20 Metagene profiles as well as binding maps indicated extended binding periods of both chaperones ( Figure 6A). Interestingly, the binding periods of TF and SecB were largely distinct, suggesting that nascent chains are bound by only one of the chaperones at a time ( Figure 6A right). To explore the interplay further, we performed SecB-SeRP in Dtig cells lacking TF. In the absence of TF, SecB binding was shifted to earlier time points during translation, taking over translation periods normally covered by TF ( Figures 6B and 6C). This indicates that TF prevents SecB binding to nascent substrates by competing for an overlapping pool    (Tables S1 and S3): in Dtig cells, CPs and translocated proteins generally shifted to higher ranks of the SecB substrate list, implying that the SecB chaperoning capacity is redirected toward protein folding and translocation ( Figure 6D). This dynamic shift very likely compensates for the loss of TF function in co-translational folding. The extended SecB binding to nascent CPs also explains how SecB overexpression can compensate the deleterious effect of a simultaneous deletion of tig and dnaK in protein folding. 44 We did not find any indication that TF may facilitate SecB binding to specific nascent chains, which would be indicated by a loss or delay of SecB binding in Dtig cells.
We also tested whether a direct substrate handover from TF to SecB may facilitate TF dissociation, by performing TF-SeRP in WT and DsecB mutant cells. However, metagene profiles (Figure S5A) as well as analysis of TF-binding onset ( Figure S5B) and offset ( Figure S5C) showed no difference in the binding behavior of TF, indicating that TF substrate interaction is independent of SecB. Finally, we analyzed how both chaperones control the membrane targeting of nascent OMPs and PPs by performing MARseq of Dtig and DsecB mutants ( Figure 6E). Similar to IMP membrane targeting, the onset of SecB engagement correlated with membrane association of translating ribosomes. Unlike what we detected for nascent IMPs, the membrane enrichment of nascent OMPs and PPs was delayed in DsecB mutants, suggesting that SecB facilitates the timely targeting of RNCs translating PPs and OMPs or stabilizes early RNC-translocon complexes ( Figure 6E). In contrast to this, MARseq of Dtig mutants showed that the lack of TF promoted   Article ll OPEN ACCESS the membrane engagement of ribosomes translating OMPs and PPs by increasing the prevalence of membrane-coupled translation and by allowing earlier membrane engagements. We conclude that TF and SecB reciprocally control the interaction of nascent PPs and OMPs with the membrane-embedded SecY translocon.

DISCUSSION
Our study reveals the multi-faceted activity of SecB in cotranslational protein maturation. We describe the redundant roles of SecB and the chaperone TF in folding of cytoplasmic proteins and the targeting of OMPs and PPs to the translocon and demonstrate that SecB contributes to IMP biogenesis ( Figure 7).
Accumulating genetic and biochemical evidence suggests that SecB is an integral part of the chaperone network implicated in folding of CPs. SecB overexpression suppresses the toxicity of a simultaneous deletion of tig and dnaK, 44 DsecB mutant cells accumulate protein aggregates that contain CPs, 61,62 and the simultaneous deletion of secB and dnaK (or dnaJ) is synthetically lethal. 42 Our data revealed that SecB engages a large number of nascent CPs (Figures 1D and 1E), suggesting that the folding support by SecB often initiates co-translationally. This is also supported by the finding that about 5% or more of total SecB in the cell engage RNCs ( Figure S2B). The majority of SecB is not engaged with translating ribosomes, in agreement with established roles in post-translational protein maturation, including distinct chaperone-like as well as targeting functions at various stages of protein biogenesis. We detected a strictly hierarchical architecture of the network with TF controlling the onset and prevalence of SecB nascent chain engagement. Control of the prevalence of nascent chain engagement by TF was previously detected for DnaK, which engages 2-to 3-fold more nascent  Figure 7. SecB is a general chaperone that acts on nascent chains of all cellular compartments SecB engages nascent proteins of all cellular compartments. For IMPs, the engagement coincides with SRP-mediated ribosomal docking at the membrane and it promotes membrane retargeting via SRP. SecB engagement is persistent until the end of translation and membrane integration. Cytoplasmic, periplasmic, and outer membrane nascent chains are first engaged by TF and later handed to SecB, which either promotes folding (CPs) or translocation (PPs and OMPs). chains in Dtig cells compared with WT cells. 63,64 Considering the toxicity of a simultaneous deletion of dnaK and secB, this suggests that DnaK and SecB may both be similarly controlled by TF, act in later stages of translation, and also have overlapping function in co-translational folding.
TF also prevented the premature membrane engagement of RNCs translating OMPs and PPs ( Figure 6E). The impaired membrane binding of ribosomes in DsecB cells ( Figure 6E) implies that TF may at least partially exert this function by limiting the binding of SecB to RNCs. The role of SecB in promoting membrane engagement may involve membrane-bound SecA, which recruits RNCs to membranes by engaging ribosomes and/or SecB. Supporting this model, we find that SecA is significantly enriched in SecB-bound RNCs compared with all ribosomes (Table S2) and previous data showed that the affinity of SecA and SecB increases if SecA is membrane associated 65 and that SecB promotes the SecA binding to periplasmic loops of IMPs. 22 Our data corroborates and extends the established model of protein translocation across the membrane: TF, by virtue of its holdase function, 66,67 maintains the translocation competence of nascent PPs and OMPs in early stages of translation. When the nascent chain grows longer, TF dissociates and is replaced by SecB, which maintains the unfolded state of the nascent protein. 35,68,69 Translocation may start co-translationally, initiated by TF dissociation, SecB recruitment, and binding to the membrane-associated SecA. SecA and SecB nascent chain engagement very likely persists beyond translation termination to also support later steps of translocation. This model provides an explanation for how tig deletion relieves the temperature-sensitive phenotype of secB mutants 42 : In SecB mutants, the membrane engagement of some RNCs may be delayed or defective. Subsequent to the TF binding period (which is not extended by the SecB absence; Figure S5B), the nascent chain may be unprotected by chaperones, which may trigger misfolding and also inhibit translocation. Deleting tig in DsecB cells may compensate for this detrimental effect of DsecB, by allowing an earlier SecA binding to short nascent chains and early translocon engagement of RNCs, which will limit the time while nascent chains are unoccupied and prone for misfolding.
Current models imply that co-translational membrane targeting of nascent IMPs relies on the SRP-mediated targeting and stable docking of the translating ribosome to the translocon, and the translation-coupled insertion of the protein into the lipid bilayer. 70 Questioning this simple model, several studies have previously suggested a contribution of SecA to IMP targeting and translocation. SecA directly binds to ribosomes 23,24 and has elevated affinity to ribosomes translating IMP-encoding mRNAs. 25 Furthermore, SecA was shown to contribute to the translocation of a specific subset of IMPs, in particular IMPs containing extended periplasmic loops. 10,15,16,22 We now also implicate SecB in the IMP translocation machinery. The binding of SecB to nascent IMPs seems independent of SecA but is critical for accurate membrane insertion of co-translationally engaged SecB substrates ( Figure 3E and S4B). SecB binding depends on a preceding SRP engagement and occurs simultaneous with membrane association, implying mutual dependence of SRP-mediated RNC engagement with the translocon and SecB binding. In agreement with this model, we find that (1) the overexpression of SRP and its receptor FtsY shifts SecB away from other substrates and toward nascent IMPs, and (2) SecB overexpression stimulates the SRP rebinding to nascent IMPs at the membrane, while the initial SRP binding is not affected.
Considering that SecB association persists during IMP insertion and until the end of translation, we propose that SecB contributes to correct membrane insertion of IMPs and the translocation of periplasmic segments across the membrane by binding to RNCs that dissociated from the translocon. The role of SecB may be to protect nascent chain segments that are exposed in the cytosol through its holdase function to facilitate subsequent SRP engagement 12 (Figure 7). A lack of this protecting function of SecB could prevent the retargeting of dissociated ribosomes and allow unproductive nascent chain folding outside the membrane, resulting in failed biogenesis of IMPs. Supporting this model, we find that the deletion of secB selectively weakened the co-translational membrane engagement of the top-ranked nascent IMP interactors but not of other proteins ( Figure 3E). Similar to the translocation of PPs and OMPs, SecB may in addition collaborate with SecA in the maturation of IMPs; for example, by supporting the translocation of large periplasmic segments. Unraveling the molecular details of SecB function at the membrane is an important open question for future studies. Answering this question will require a combination of mechanistic studies of the membrane insertion process and structural analysis of the ternary or quaternary complexes of SecB with ribosomes that are additionally engaged with either the translocon, SecA, or SRP.

Limitations of the study
Based on a range of methods that analyze (1) interactions of SecB with translating ribosomes, (2) the membrane engagement of RNCs, and (3) the folding properties of IMPs in the context of living cells, our study suggests a role of SecB in IMP biogenesis in E. coli. One limitation of our study is that the mechanism by which SecB supports the membrane insertion remains enigmatic. Two non-exclusive models are (1) that SecB engages nascent chains of RNCs that transiently dissociated from the translocon to support another round of SRP-mediated retargeting, and (2) that SecB cooperates with SecA in the translocation of periplasmic loops. Another potential limitation are biases of, in particular, mass spectrometry, and to some extent also ribosome profiling-based approaches (SeRP, MARseq), toward the detection of highly abundant proteins. In particular, the low_CIbased score, which we use to assess strength of binding, will not reflect the equilibrium constant of genes/regions with low coverage. Additionally, membrane proteins are inherently more difficult to detect using proteomics, likely due to their more difficult extraction and ability to produce MS-detectable peptides. Furthermore, we cannot exclude the possibility that SeRP and MARseq data report on interactions of different sub-populations of all translating ribosomes. This possible limitation can only be overcome by the development of new approaches that can detect the properties of individual molecules within complex ensembles that simultaneously exist in the highly complex context of a living cell.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources should be directed to and will be fulfilled by the lead contact, G€ unter Kramer (g.kramer@zmbh.uni-heidelberg.de).

Material availability
Plasmid and strains generated in this study are available from the lead contact upon request.
Data and code availability d All data reported in this paper will be shared by the lead contact upon request. d All original code has been deposited at Open Science Framework repository and is publicly available as of the date of publication. DOIs are listed in the key resources table. d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
Experiments performed in this work have been conducted using Escherichia coli MC4100 strain as originally described in. 71 Transgenic derivatives of the original strain generated for the purpose of this study are listed in the key resources table.

METHOD DETAILS
Cultivation of E. coli strains E. coli cells were grown in Luria Broth (LB) or Neidhardt rich defined MOPS media by incubation at 30/37 C under rigorous shaking (110 pm). Appropriate antibiotics were added to E. coli strains carrying a plasmid with a resistance cassette or a deletion. The script to plot individual transcript profiles works similar to the generation of a ranked list. This script calculates for each transcript position the 95% confidence interval according to Agresti and Coull of the enrichment ratio (factor-bound translatome to total translatome) by using a sliding window. This information is then plotted for the two biological replicates separately for individual transcripts.
The heuristic to determine binding periods is a threshold-based method to identify periods during translation of specific transcripts where the FOI binds the RNC. This is helpful to investigate the binding behavior of different factors or to study the effect of a deletion on FOI-substrate interaction. This data can then be visualized in a binding map for several nascent proteins. The binding detection heuristic is based on the calculation of the 95% Agresti-Coull confidence interval (CI) as described for the ranked list. Contrasting the previous analysis, the algorithm used here first stores the 95% CI information for each position of each transcript. Then it determines for each position, whether obtained CI is entirely below the chosen threshold (no binding), is entirely above it (binding) or includes the threshold (binding status cannot be determined).

QUANTIFICATION AND STATISTICAL ANALYSIS
For the statistical analysis of the (selective) ribosome profiling data R (version 3.5.1) from https://www.r-project.org/foundation/ was used. Statistical details are found in the figure legends as well as in the METHOD DETAILS section. The ranked lists as well as individual gene profiles were generated by calculating the 95% confidence interval according to Agresti and Coull of the enrichment ratio (factor-bound translatome to total translatome) by applying a sliding window of 35 codons. Two biological replicates were visualized separately. Proteomics data was searched using MaxQuant (version 2.1.4.0 76 ) and further processed using R (version 3.5.1). Spatial proteomics subcellular localization data has been analyzed using the pRoloc R package. 52 Specific details of analysis can be found in the METHOD DETAILS section. Localization data were analyzed in three biological replicates. Specific description of depicted values can be found in the figure captions.