IntroductionSecretion is an essential and conserved mechanism. In Escherichia coli more than 500 proteins cross the inner membrane through the ubiquitous Sec system to reach their destination and become functional1. To retain translocation competence, clients of the Sec pathway must remain soluble and unfolded2,3,4. While in co-translational secretion, this bottleneck is bypassed by coupling the client’s translation to translocation5, in post-translational secretion, nascent chains rely on their intrinsic features and chaperones to avoid (mis)folding or aggregation during their cytoplasmic transit.Secretory clients have been evolutionary adapted to delay their folding. N-terminal signal peptide additions (SP6,7,8,9) and mature domain sequences that promote disorder and reduce overall hydrophobicity10,11 contribute both to slow folding. Key interactions with chaperones stabilize unstable or short-lived folding intermediates12, contributing to client solubility and non-folding13. Secretory chaperone examples include: (i) Trigger Factor (TF) that sorts nascent chains at the ribosome14,15 and targets secretory clients to the translocase16, (ii) SecB that binds newly synthesized preproteins to prevent premature folding or aggregation17,18 and, (iii) SecA, the motor subunit of the Sec translocase19 which associates with nascent preproteins at the ribosome20 or in solution and targets them to the translocase21. Three main activities have been attributed to chaperones and reflect their functional classification: (i) “unfoldase”; they convert misfolded proteins into folding-competent states22, (ii) “holdase”; they hold the client in one state, preventing folding or aggregation, and (iii) “foldase”; they assist client folding, oftentimes at the expense of energy23.Insights into the mechanism of secretory client selectivity have been primarily inferred from studying the interaction of the maltose-binding protein (MBP) with SecB24,25,26. MBP is a two-domain protein with a discontinuous N-terminal and C-terminal domain. Its structure27, folding kinetics, and thermodynamics8,28 are well characterized. Folding out of urea, MBP initially collapses into a compact molten globule-like state29 and then, a ‘folding core’ (FC) made of cooperative folding units called foldons30 drives an on-pathway folding intermediate that then transitions toward the native state31,32. The interaction of MBP with the SecB chaperone has been widely studied, and the unfolded client binds to a surface-exposed groove formed between two SecB dimers, occupying at least 7–8 binding sites15, with low specificity but high affinity33. SecB seems to bind proteins with low specificity by recognizing non-conserved stretches of 9 residues enriched in aromatic and basic residues34 and does not recognize signal peptides; yet shows some selectivity as it preferentially binds secretory or aggregation-prone proteins. This selectivity has been proposed to arise via a kinetic partitioning between folding or binding to SecB and is modulated by the affinity and the folding rate of the client35. By retarding folding of the mature domain, the signal peptide increases the time window for interaction with SecB6. Given that only a handful of secretory proteins are SecB clients1,36, alternative models have been proposed; i.e., selectivity to SecB might relate to the ability of the client to fold while tethered to the chaperone or the substrate-chaperone complex might dictate a relay to a partner such as SecA34,37.How the folding properties of MBP relate to its interaction with SecB and how SecB maintains the client in a translocation-competent state that allows downstream transfer to the translocase remains elusive. Here, by combining single-molecule Förster resonance energy transfer (smFRET) and hydrogen-deuterium exchange mass spectroscopy (HDX-MS)38,39,40,41,42, we describe the folding of MBP, with contributions from signal peptide and mutations that slow down folding, identified the folding intermediates that are recognized by SecB, and explored the chaperone effects on the folding dynamics of this client. Moreover, we demonstrate that SecB maintains MBP as a translocation-competent state by acting initially as an unfoldase and then as holdase. While SecA has limited effect on MBP alone, it specifically recognizes MBP:SecB and assembles as a stable quaternary super-assembly forming the next step towards translocase targeting.ResultsMBP refolding reveals a high-FRET intermediateThree derivatives of the maltose binding protein (MBP) were used in this study: the mature domain (MBP), the preform carrying the signal peptide (proMBP) and a slow-folding mutant [MBP(Y283D)], identified in an in vivo screening of a defective signal peptide43 that was later shown to delay folding in vitro and in vivo7,44. A cysteine pair was introduced in all derivatives (36C/352C44, MBP numbering; Supplementary Fig. 1A.i45,46), and these proteins are used across the study.For smFRET, proteins were stochastically labelled with donor/acceptor (Alexa555/Alexa647) probes. The apparent FRET distribution of native MBPT36C/S352C (hereafter MBP) is centered at Eapp = 0.54 (N; Fig. 1A and Supplementary Fig. 1A.ii; visualized by color code as indicated), while unfolding MBP in 8 M urea resulted in low FRET values (U, Eapp = 0.16). To monitor refolding kinetics, the unfolded MBP was diluted in native buffer (200-fold, to 40 mM urea) and recording started immediately; the first ~5 seconds were routinely missed due to mixing/ loading time (Fig. 1A, right panel). Within seconds, the Eapp = 0.16 of unfolded MBP shifted sharply to higher FRET values, in a likely molten globule state (hereafter MG31) and converted within minutes to a state with lower FRET distribution similar to the native state (Eapp = 0.54, Fig. 1A), that could bind maltotriose (Supplementary Fig. 1A.ii). Attaching the probes to a different cysteine pair did not alter the folding behavior of MBP (Supplementary Fig. 1B.i–iii).Fig. 1: MBP folds through a highly disordered intermediate.A–C Refolding of urea-unfolded MBP, proMBP, MBP(Y283D), upon 200-fold dilution in native buffer (to 40 mM urea). The apparent FRET values of proteins in the course of folding time (right panels), and those of native (N) and urea-unfolded (U) steady-states (left slices; see Supplementary Fig. 1A–D) used as controls, were represented by kernel-density estimation and are visualized as color maps; purple-yellow indicates low-high numbers of events. D Evolution of the native fraction as a function of time, extracted from the folding experiments (points), together with single-exponential fits (lines). E Refolding of urea-unfolded proteins in native buffer (30-fold dilution, to 0.2 M urea) was monitored by HDX-MS (see Supplementary Fig. 2 A and Supplementary Data 1). The degree of folding (DOF) per residue of MBP (top; 10 s), or MBP(Y283D) (bottom; 10 min), is represented by a color code ranging from yellow (0 DOF: unfolded protein in 6 M urea) to purple (1 DOF: native protein) (see Supplementary Data 2). Grey: prolines/non-identified regions. Greek letters indicate foldons in order of appearance (see Supplementary Fig. 2A, B); black arrow indicates the Y283D mutation on the linear sequence. F The folding core of MBP is mapped on the structure of MBP (PDB: code: 1ANF).Full size imageSimilarly, unfolded proMBPT36C/S352C (proMBP) started from a low-FRET state (U, Eapp = 0.16) and refolded through a high-FRET MG to a state with native-like Eapp, but the transition was slower than MBP (Fig. 1B and Supplementary Fig. 1A.iii). MBPT36C/S352C/Y283D [hereafter MBP(Y283D)] showed similar distributions of the native and unfolded states (Fig. 1C and Supplementary Fig. 1A.iv). However, this mutation delayed for hours the MG conversion to a state with Eapp = 0.53 that could bind maltotriose (Supplementary Fig. 1A.iv).Although the folding pathway was similar in all derivatives, it occurred within different timescales. Folding data were fitted to a Gaussian mixture model to extract the fraction of the native state as a function of time (Fig. 1D). In all derivatives, the native fraction at t0 is above zero, indicating that a fraction of molecules “escaped” the high-FRET MG intermediate. Similar findings by ensemble methods have described a “burst-phase”8, i.e., within a folding ensemble, some chains fold directly while others explore an intermediate. The remaining folding kinetics were fitted by a single exponential to obtain an apparent folding rate. The MBP kapp = 8.2 (±0.2) × 10−3 s−1 matched previously reported values8. The signal peptide delayed the transition to the native state by a factor of ~10 [proMBP kapp = 9.7 (±0.2) × 10−4 s−1], while the Y283D mutation resulted in a 43-times slower kapp [1.9 (±0.5) × 10−4 s−1].The high-FRET intermediate is highly disordered and folds toward the native state through defined foldonsTo describe MBP folding at near-residue level, we used local HDX-MS42,47. While the aggregation-prone proMBP could not be investigated, urea-unfolded MBP or MBP(Y283D) were allowed to refold by dilution in native buffer (30-fold, to 200 mM urea), pulse-labelled in D2O (10 s) at different folding time points, quenched, digested, and analyzed by MS. As secondary structural elements are progressively formed during folding, less hydrogen amides can be exchanged with deuterium (D). Hence, D-uptake as a function of folding time can report on protein folding31. The experimentally determined D-uptake per peptide (Supplementary Data 1) was converted to “degree of folding” (DOF) by normalization between fully denatured (DOF = 0) and native (DOF = 1) controls (Supplementary Data 2). Peptide-level DOF values were mapped to residues by weighted averaging, using PyHDX48, and are visualized as linear color gradients plotted against refolding time (Supplementary Fig. 2A).Since the MG intermediate is short-lived in MBP compared to MBP(Y283D) (Fig. 1A, C), we focused on the 10 s (MBP) or 10 min MBP(Y283D) of folding (Fig. 1E). Most of the MBP chain is highly dynamic and disordered (yellow), except for five regions that exhibit relatively advanced folding (green) and are indicated with Greek letters in order of appearance (for details about the folding rates of foldons, see Supplementary Fig. 2B, C). While distant in primary sequence, these regions are proximal in the native structure (Fig. 1F), as anticipated for foldons31.The overall folding of MBP(Y283D) was extremely slow (Supplementary Fig. 2A–C). The mutation has significantly modified the protein’s folding core (FC) (Fig. 1E, bottom); at 10 min the chain is mostly disordered (yellow) and only limited regions showed some DOF. As expected, the Y283D mutation abolished the formation of foldon δ, close to where the mutation lies, but also showed long-distance effects as foldon γ is absent despite being located about 250 residues upstream in the primary structure. This is explained by the proximity of these two foldons in the folding core of the protein. Only foldon ɛ seemed unaffected by the mutation, however, only at earlier timepoints.The high FRET intermediate of MBP is highly disordered, like previously described molten globule (MG) states29,32,49,50 and folding proceeds through a folding core (FC) of 5 well-defined foldons. For MBP(Y283D), the HDX-MS bulk folding rate [1.7 ( ± 0.2) × 10−4 s−1] was comparable to the apparent folding rate found by smFRET. In contrast, HDX-MS and smFRET folding rates for MBP differ by a factor ~9. This difference is attributed to the abolishing of the FC intermediate by the Y283D mutation, such that both methods directly observe native state formation. In MBP, chains forming the folding core acquire native-like Eapp, therefore yielding faster apparent folding rates.SecB holds its client in an expanded stateSecretion of proMBP is thought to be SecB dependent24,25. To investigate the effect of the chaperone on client folding, unfolded MBP was diluted in native buffer that contained SecB4 at physiological concentration (1.6 µM51). Under these conditions, a new low-FRET population (Eapp = 0.24) was observed (Fig. 2A.i), distinct from the U-state (Eapp = 0.18), indicating that the probes are now distant, despite the absence of urea (~9 nm as calculated from accurate FRET corrections52). Size-exclusion chromatography coupled with multiple-angle light scattering (SEC-MALS) confirmed MBP:SecB4 complex formation under similar conditions (Fig. 2B, top panel). The low-FRET population represents a SecB-bound (B) state that eventually converted to Eapp = 0.54 with a folding rate of kapp = 1.09 ( ± 0.02) × 10−3 s−1 ~ 8-times slower than in the absence of SecB (Fig. 2C, blue dashed line). Attaching the probes to a different cysteine pair did not alter the folding behavior of MBP in the presence of SecB (Supplementary Fig. 1B.iv). The apparent lifetime of the B-state decreased in a competition assay upon addition of MBP excess (2.5 µM unlabeled-unfolded; Fig. 2A.ii, C, blue dotted line) and increased upon addition of SecB4 excess (16 µM; Supplementary Fig. 3A, top). To rule out artefactual effects from crowding or aggregation due to excess unlabeled client, we performed control experiments in which labeled MBP derivatives were folded in the presence of 2.5 µM unfolded and unlabeled MBP in the absence of SecB4. MBP folding rates these conditions were comparable to those of labeled MBP folded alone, indicating that the accelerated folding observed in the competition assay reflects a shift of the labelled client/chaperone equilibrium by sequestering free SecB4 by unlabeled material. Additionally, aggregation is unlikely since the protein is known to be soluble in the concentration used41.Fig. 2: SecB holds clients in a disordered state preventing their further folding.Ai-iii-v Refolding of unfolded (U, left slices) MBP, proMBP, or MBP(Y283D) in the presence of SecB4 (1.6 µM, right panels). A ii-iv-vi Unfolded MBP, proMBP, or MBP(Y283D) were allowed to refold in the presence of SecB4 for 5 s (B: bound state, left slices) before adding unlabeled/unfolded client excess (2.5 µM; competition; right panels), as in Fig. 1A. B SEC-MALS analysis of protein complexes. 15 µM unfolded MBP or MBP(Y283D) was mixed with SecB4 (15 µM) and analyzed either directly (solid line) or after two hours incubation (dashed line). SecB4 and native MBP controls are shown (grey line). C Evolution of the native fraction as a function of time, in three different conditions per protein. Solid line: Refolding of apoprotein (Fig. 1A–C). Dashed line: Refolding in the presence of SecB (A i-iii). Dotted line: Once the B-state is formed unlabeled/unfolded MBP was added (A iv-vi). D, E Refolding of MBP, or MBP(Y283D), in the absence/presence of SecB4, for the indicated amount of time, was monitored by HDX-MS; as in Fig. 1E.Full size imageproMBP exhibited similar folding behavior in the presence of SecB (Fig. 2A.iii). The low-FRET B-state (Eapp = 0.24) shifted to Eapp = 0.55 with a folding rate of kapp = 2.39 ( ± 0.09) × 10-4 s-1, ~4-times slower than in the absence of chaperone (Fig. 2C, orange dashed line). The presence of signal peptide stabilized the client-chaperone interaction. Unlike MBP, labeled proMBP cannot be significantly replaced upon addition of client excess (Fig. 2A.iv). Presumably the signal peptide provides for additional SecB binding sites34,53. Like MBP, addition of chaperone excess to the unfolded proMBP further populated this state, slowing down overall folding of the client (Supplementary Fig. 3A, middle).MBP(Y283D) showed a striking folding behavior in the presence of SecB. The low-FRET B-state was observed up to 24 h (Fig. 2A.v) and barely transitioned to the native state (kapp 40-fold (Fig. 1E).Secretion of MBP is SecB dependent24,25. Here, we showed that SecB recognizes the MG and FC intermediates of MBP (Fig. 3A and Supplementary Fig. 4) but not native states (Supplementary Fig. 3C) and demonstrated for the first time that SecB denatured partially folded secondary structural elements and then held the client, consequently preventing formation of critical foldons (Fig. 2 and Supplementary Fig. 2D). Known unfoldases, like DnaK63 or ClpB64 are dedicated to prevent off-pathway folding. Usually, conformational changes induced in such chaperones unfold misfolded client elements or negate aggregation at the expense of energy22,65. For example, upon ATP binding, ClpB assembles into hexameric rings competent for substrate binding and then unfold and translocates the substrate through its channel64. In the Hsp70 system, DnaJ recognizes misfolded clients, recruits DnaK that uses ATP hydrolysis to mechanically unfold them66,67, and then GrpE acts as a substrate-release agent by releasing the nucleotide from DnaK68.In contrast, the role of secretory chaperones like SecB, SecA, and TF is to prevent on-pathway folding and maintain the client in a translocation-competent state. Preproteins have evolved intrinsic features that delay their folding, i.e., signal peptides and mature domain regions that promote disorder and disperse hydrophobicity41 minimizing the ‘’unfolding task at hand” for secretory chaperones. Here, SecB had only to revert partial folding of secondary structures (Fig. 3F), without energy input. In the absence of an ATPase activity, the unfoldase activity of SecB can only result from passive interactions between hydrophobic stretches in the disordered client69 and the SecB hydrophobic groove15. Then the client is held in a completely disordered state, lacking secondary or tertiary structures (Fig. 2), by a dynamic catch-and-release interplay (Fig. 2A.ii and Supplementary Fig. 4A) as previously proposed70 and in complete agreement with reported anti-folding activity24,25,71 and the structure of unordered clients wrapped around a chaperone15.The signal peptide stabilized the client:SecB interaction (Fig. 2A.iii, iv), possibly by offering additional chaperone binding sites53,72, yet we observe that slow-folding mutants feature increased SecB holdase activity, in part through a decrease in off rate and absence of on-chaperone folding. These mutants increase SecB holdase activity without offering additional binding sites, which suggests that client folding drives chaperone off rates, yet released clients are released as folding intermediates and not native state. However, off-rates obtained from a global fit of SecB titration (Supplementary Fig. 4A, B) cannot explain why SecB-bound Y283D client could not be replaced in a competition assay, even after 24 h. (Fig. 2A.vi).Clients of the Sec pathway will ultimately be transferred to SecA, the ATPase motor subunit of the translocase either in solution or at the membrane-site73. SecA alone did not recognize MBP intermediates (Fig. 4A). Instead, it required a pre-assembled MBP:SecB4 complex (Fig. 4B) to form stable quaternary super-complexes in solution (Fig. 4C, D), explaining why secretion of MBP is dependent on the non-essential SecB. The SecA2:MBP:SecB4 interaction is dependent on the SecA2:Zn2+:SecB4 interaction58 (Fig. 4C, D). One can only assume that the c-tail of SecA binds the flat surface of SecB, as previously reported74, with MBP wrapped around SecB15. In support of this hypothesis, the SecA to SecB interaction site is still accessible in the structure of MBP:SecB15. The client is anticipated to be sandwiched between the two chaperones, with the C-tails of SecA2 locking it onto SecB4 and preventing its release, in agreement with the longer lifetime of the low-FRET B:A state (Fig. 4B–D). In the current model of the SecA:SecB complex75, a gap of >40 Å is observed between the two chaperones and could accommodate the client. In addition, the C-tail of SecA, bound to SecB, is connected to the body of SecA by a stretch of 40 disordered residues76,77 that could theoretically extent, resulting in a gap between the two chaperones up to 70 Å, again allowing the client to be sandwiched between the two chaperones15. Such positioning might enable the transfer of the client between the two chaperones, as suggested37,57,61,78. However, in the absence of signal peptide, no such transfer was observed79 (Fig. 4C). An external signal (e.g., interaction with the SP or binding to the translocase) might trigger the client transfer between the two chaperones. A SecA2:MBP:SecB4 structure is not available yet, and future experiments are clearly due in order to determine the requirements and mechanism by which a client is relayed from SecB to SecA. Nevertheless, the observed complex, where the two chaperones act synergistically, is on-pathway for the transit of a secretory client in the translocation-competent state and likely represents a late step towards delivery to the translocase.MethodsMaterials availabilityThis study did not generate new unique reagents.PlasmidsPlasmids coding for either the maltose binding protein (MBP; malE gene; UNiprot accession: P0AE9) or the precursor form (proMBP) carrying a C-terminal His-tag for protein purification were obtained from previously published work41. The gene coding for MBP was modified to introduce two cysteine residues (T36C and S352C) for fluorophore labeling80.Molecular cloningSite-directed mutagenesis was performed using the QuickChange site-directed Mutagenesis protocol (Stratagene Agilent) using indicated vector templates and primers (Supplementary Table 1). Genes were inserted into the indicated plasmids by restriction enzyme digestion and ligation using T4 DNA Ligase (Promega) or using Gibson cloning (Seamless assembly, Codex). Restriction sites/overlapping regions for the gene of interest and mutations were added using PCR with PFU (Ultra) Polymerase (Stratagene).Protein expression. MBP (and derivative) expressionFor MBPT36C/S352C, MBPV8G/T36C/S352C, MBPT36C/A276G/S352C, and MBPT36C/Y283D/S352C expression, E. coli BL21(DE3) (T7 RNA polymerase gene under the control of the lac UV5 promoter81) cells were transformed with plasmids containing the relevant protein derivative gene (Supplementary Table 1) and grown at 37 °C in 1 L of LB medium until the OD600 reached a value around 0.8. Gene expression was induced with 0.2 mM IPTG, and cells were grown 3 h at 37 °C. Cells were then collected by centrifugation (4000 × g; 15 min, 4 °C) and the resulting cell pellet was stored at −20 °C. proMBP expression. E. coli BL21(DE3) cells were transformed with the plasmid coding for proMBP(T36C/S352C)-his and grown at 37 °C in 1 L. of LB medium supplemented with 5 mM MgCl2. When the OD600 reached a value around 0.6, the medium was supplemented with 4 mM sodium azide, and 1 mM PMSF to reduce the cleavage of the signal peptide, incubated 15 minutes at 30 °C, and gene expression was induced with 0.2 mM IPTG for three hours at 30 °C. Cells were then collected by centrifugation (4000 × g; 15 min, 4 °C) and the resulting cell pellet was stored at −20 °C. Chaperones expression. E. coli BL21(DE3) cells were transformed with the plasmid coding for secA or secB and grown at 37 °C in 2.5 L. of LB medium. When the OD600 reached a value around 0.6, gene expression was induced with 0.2 mM IPTG, and cells were grown 3 h at 37 °C. Cells were then collected by centrifugation (4000 × g; 15 min, 4 °C and the resulting cell pellet was stored at −20 °C.Protein purificationAll protein purification steps and centrifugations were performed at 4°C. MBP (and derivative) expression purification. The bacterial pellet was solubilized in 35 mL of Buffer A (50 mM Tris-HCl pH 8.0, 400 mM NaCl, 5 mM imidazole, and 5% glycerol, Supplementary Table 2) and then passed five times through a pre-chilled French Pressure Cell (SLM-Aminco) at 1000 PSI. The resulting suspension was then centrifugated at 14.000 × g for 15 min, and the supernatant was loaded on a Ni2+-NTA-Resin column (5 mL) previously equilibrated with 5 column volumes (CV) of buffer A. The column was then washed with 5 CV of buffer B (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 5 mM imidazole, and 5% glycerol) and eluted with buffer C (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 200 mM imidazole, and 5% glycerol). Purified proteins were concentrated and then dialyzed 2 h against buffer D (50 mM Tris-HCl pH 8.0, 50 mM NaCl, and 5% glycerol) and overnight against buffer E (50 mM Tris-HCl pH 8.0, 50 mM NaCl, and 50% glycerol); then aliquoted and stored at −20 °C. proMBP purification. The bacterial pellet was resuspended in 35 mL of Buffer A, passed five times through a pre-chilled French Pressure Cell (SLM-Amico) at 1000 PSI. The resulting suspension was then centrifuged (14.000 × g; 15 min) and the pellet was washed with 35 mL buffer F (50 mM Tris-HCl pH 8.0, 1 mM EDTA, 50 mM NaCl, 1 M urea, and 1% Triton X-100) and centrifuged at 14.000 × g for 15 min. The pellet containing inclusion bodies was then solubilized overnight in 35 mL buffer G (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 1 mM DTT, 5 mM imidazole and 6 M urea) and loaded on a Ni-NTA-Resin column (5 mL) previously equilibrated with 5 CV of buffer G, and eluted with buffer H (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 200 mM imidazole, and 6 M urea). Purified proteins were concentrated, dialyzed 2 h against buffer I (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, and 6 M urea), and then stored at −20 °C. SecB purification. The bacterial pellet was dispersed in 35 mL of Buffer J (50 mM Tris-HCl pH 8.0, 400 mM NaCl, and 5% glycerol) supplemented with 2 mM PMSF and 50 µg/mL of DNAse I, passed five time through a pre-chilled French Pressure Cell (SLM-Amico) at 1000 PSI, the centrifugated at 14.000 × g for 15 min. The supernatant was first loaded on a Q-Sepharose fast-flow resin (10 mL, equilibrated in buffer D), diluted twice to reduce the salt concentration to 200 mM; then loaded onto a HiPrepQ FF 16/600 (25 mL, equilibrated in buffer K (50 mM Tris-HCl pH 8.0, 200 mM NaCl, 5% glycerol). The column was then washed with 5 CV of buffer, and the protein was eluted over a linear gradient of NaCl from 200 to 400 mM. Fractions containing SecB were concentrated and then loaded on a Superdex 200 26/600 (equilibrated in Buffer D), and the purified protein was concentrated, dialyzed overnight against buffer E, and then stored at −20 °C. SecA purification. The bacterial pellet was dispersed in 35 mL of Buffer L (50 mM Tris-HCl pH 7.4, 200 mM KCl, 1 mM MgCl2, and 10% glycerol) supplemented with 2 mM PMSF and 50 µg/mL of DNAse I, passed five time through a pre-chilled French Pressure Cell (SLM-Amico) at 1000 PSI; then centrifugated at 14.000 × g for 15 min. The supernatant was sequentially loaded onto a Cibacron column (25 mL, equilibrated in buffer L) and eluted over a linear gradient of KCl (0.2 to 1 M KCl); then concentrated and loaded on a Superdex S200 26/600 equilibrated with buffer M, and then buffer N (50 mM Tris-HCl, pH 7.4, 50 mM KCl, and 10% glycerol). For storage, SecA is dialyzed overnight against buffer E and stored at -20°C.Single molecule-FRET. Protein labelingHis-MBP and derivatives (100 µg, from a glycerol stock) were treated with 5 mM DTT (15 minutes on ice), diluted to 600 µL in buffer D, loaded onto a Ni-NTA column (100 µL, equilibrated with buffer D), and the resin was washed with 5 CV of buffer D to remove the excess of DTT. Alexa555-maleimide (Thermo Fisher Scientific, 25 nmol) and Alexa647-maleimide (Thermo Fisher Scientific, 25 nmol) were dissolved in 4 μL DMSO, then diluted in 500 µL of buffer A and added onto the resin to induced stochastic labeling of MBP for 2 h at 4 °C under gentle agitation. The resin was washed with (3 × 200 µL) buffer A to remove excess of dyes and proteins, were eluted with 3 × 200 µL of buffer C 3 × 200 μl and then loaded on a Superdex S200 10/300 previously equilibrated with buffer A to remove the excess of dyes. Labeling efficiency was first confirmed by a shift on SDS-PAGE and measured above 90% by absorbance on a nanodrop spectrophotometer (Nanodrop 2000, Thermo-Fisher). Then evaluated over smFRET measurements. Single-molecule fluorescence microscopy and PIE: Single-molecule Pulse-Interleaved Excitation (PIE) experiments were performed at 20 °C using a MicroTime 200 (PicoQuant, Germany) with laser lines at 520 nm (70 µW) and 637 nm (30 µW). Laser excitation pulses were interleaved at 20 MHz repetition rate. The lasers were coupled into the objective through the main dichroic (ZT532/640rpc-UF3, AHF Analysentechnik) and focused ~20 µm above the microscope coverslip. Fluorescence emission was collected by the same water objective (UPLSAPO 60x Ultra-Planapochromat, NA 1.2, Olympus), focused onto a 75 µm pinhole and separated onto two Single-Photon avalanche diodes (SPAD) with appropriate spectral filtering (donor channel: 582/64 BrightLine HC (F37-082); acceptor channel: 690/70H Bandpass (F49-691); both AHF Analysentechnik). Samples were prepared by dilution in buffer O (50 mM Tris-HCl pH 8.0, 50 mM NaCl, and 1 mM Trolox) and quickly loaded on a microscope cover slide (no. 1.5H precision cover slide, VWR; coated with 1 mg/mL BSA for at least 2 minutes to passivate the surface). To monitor folding, the labeled proteins were first diluted in buffer P (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM Trolox, and 8 M urea) for 15 minutes at room temperature; then diluted 200x in buffer O (final urea concentration at 40 mM) and immediately loaded on the confocal microscope. Burst search and thresholding: Time-tagged photon detection events were read from instrument. ptu files with the “phconvert” python library (part of the Photon-HDF5 project82), and a sliding temporal window was applied to the timestamps where a fluorescent signal was considered a burst when a total of 50 photons are identified with at least 35 neighboring photons in a 500 µs window. This sliding window filter was applied separately to photons emitted during donor and photons emitted during acceptor excitation (dual-colour burst search83), and both results reduced to overlapping intervals between both searches. Additionally, filters were applied burst-wise; number of photons > 110,