A small molecule allosterically activates SecA dependent secretion

Abstract

The Sec pathway is an essential protein secretion route for all organisms. In bacteria, the SecA ATPase peripherally associates with the SecYEG channel to form the translocase that mediates preprotein export. Activation of the translocase depends strictly on the synergy of signal peptide and mature domain binding. Thus, client selectivity, translocase activation and protein secretion are coupled by one mechanism. We show here that a previously identified small molecule (HSI#6) binds SecA, modulates its intrinsic dynamics and allosterically activates the translocase in the absence of clients. By uncoupling translocase activation from preprotein binding, HSI#6 transformed the translocase into a promiscuous nanomachine that lost client selectivity and secreted unfolded pre- mature- and cytoplasmic- proteins with high efficiency in vivo or in vitro. To our knowledge, HSI#6 is the first activator of the Sec pathway and might offer unique opportunities for the discovery of new antibacterials.

A small molecule that binds on SecA uncouples activation of the translocase from preprotein binding, transforming the Sec translocase into a promiscuous nanomachine that lost client selectivity and secretes any unfolded protein in vivo or in vitro.

Introduction

Secretion is a fundamental, conserved life process. The Sec pathway is the main essential secretion route for all organisms. In bacteria, the SecYEG membrane channel together with the peripherally associated SecA subunit mediate the post-translational export of the large majority of the secretome^1–3.

SecA, a large dimeric protein, interacts with nucleotides, multiple clients, chaperones, lipids and SecYEG and functions as the ATPase of the system¹. SecA protomers comprise four main domains⁴ that associate to distinct functions. ATP is sandwiched between two domains (NBD1 / 2) that form the helicase motor^5–7. The preprotein binding domain (PBD) sprouts out of NBD1 through a Stem region, interacts with both signal peptides and mature domains⁸ and oscillates between three distinct states^9,10. The C-domain physically associates with the helicase motor suppressing its ATPase while the C-tail folds over the signal peptide cleft and stem regions to limit illicit client access¹¹.

In the cytoplasm SecA₂ is ADP bound, has low ATPase activity^5,12–14 and the PBD is predominantly in the wide-open state¹⁰. Docking on SecYEG only primes the motor; it allosterically alters the dynamics of the motor/stem without releasing ADP nor lowering its activation energy^10,15–17. Signal peptide binding lowers the activation energy of SecA¹⁵ whereas the synergy from mature domain binding secures sufficient motor dynamics to release ADP, stimulate ATP turnover and shift PBD towards the closed state^10,13. This sophisticated mechanism couples preprotein binding to translocase activation and restricts client selectivity to legitimate clients only. During secretion SecA also exploits quaternary dynamics. Alongside docking on SecYEG, lowering the activation energy and activating ATP catalysis, SecA slides through slightly different dimer conformers towards monomerization, that occurs after client-trapping¹⁶. Structures of SecA bound onto SecYEG with or without a client are available^18,19 and the motions and intrinsic dynamics of SecA in various steps of the translocation reaction have been detailed^{7,10,17,20–23}.

As SecA is a key component in protein secretion and ubiquitous in bacteria it makes a bona fide target for novel antibiotics^24–28. In one such effort, we came across a small molecular weight compound that exhibits antibacterial activity, HSI#6²⁹.

Here, we show that HSI#6 (Fig. 1A) is the first to our knowledge activator of the Sec translocase. HSI#6 binds onto SecA and allosterically increases the catalytic efficiency of the holoenzyme. By modulating the intrinsic dynamics of SecA HSI#6 activates the translocase and uncouples known secretion-limiting steps, like ADP release, lowering the activation energy and ATP catalysis activation, from preprotein binding. By stabilizing a constitutively activated holoenzyme in a client independent manner, HSI#6 transforms the Sec translocase into a promiscuous nanomachine that lost client selectivity and can secrete any client (preproteins, mature domains, cytoplasmic proteins) with similar efficiency.

Fig. 1. HSI#6 binds onto SecA and activates the wild-type translocase enhancing preprotein secretion in vitro and in vivo.

A The chemical structure of HSI#6 (see also Supplementary note). B Effect of 10 μM HSI#6 addition on the basal (0.4 μM SecA), membrane (plus 1 μM SecYEG-IMVs) and translocation (further addition of 10 µM proPhoA) ATPase (1 mM ATP; 10 min; 37 °C) of SecA. n = 3. Error bars represent standard deviation. C, E, G The basal (C), membrane (E) and translocation (G) ATPase of SecA (y axis; nmol Pi min^-1; as in B) was measured at the indicated HSI#6 and ATP (x axis) concentrations (37 °C; 10 min basal and membrane / 1 min translocation ATPase). n = 3. Best fitting to experimental data (Eq. 1a; Methods) is shown. For K_m and V_max values see Supplementary Table 1. D, F, H Effect of the HSI#6 concentration (x axis) on the K_cat values (left y axis; blue) and K_m (right y axis; black) of basal (D), membrane (F) and translocation (H) ATPase of SecA. Best-Fit values with SEM are shown (see Supplementary Table 1). Apparent K_d (D, F, H) and AC₅₀ (F) values are shown. I Effect of 10 μM HSI#6 addition on the SecA dependent secretion of proPhoA, in vitro (0.4 μM SecA; 1 μM SecYEG-IMVs; 10 μM proPhoA; 1 mM ATP, 10 min, 37 °C), monitored by immunostaining (α-PhoA). n = 7. Quantification (top; error bars represent standard deviation) and a representative western blot (bottom) are shown. Lane 1: 20% of proPhoA input; lane 2: No HSI#6; lanes 3-5:10 μM HSI#6; lane 3: No SecA; lane 5: Triton X-100 control. J Effect of 10 μM HSI#6 addition, ±NaN₃, on the in vivo Alkaline Phosphatase (AP) activity of cells overexpressing proPhoA. n = 3. Error bars represent standard deviation. K Effect of HSI#6 on the expression level of proPhoA. Samples corresponding to J, lanes 1, 3 were immunostained (α-PhoA). Signals were quantified using a standard PhoA curve. B–K Wild-type SecA and wild-type SecYEG were used. All samples contained the same final DMSO concentration.

Results

HSI#6 binds onto SecA and allosterically activates the translocase

HSI#6 (Fig. 1A; Supplementary note) has been identified during a screen that monitored bacterial protein secretion as an indicator of growth inhibition²⁹. To probe whether HSI#6 specifically affected protein secretion, we tested its effect on the ATPase of the Sec system in vitro.

SecA is known to display three distinct ATPase activities: Basal; low ATPase, characteristic of the enzyme in solution (Fig. 1B, lane 1; 0.4 µM SecA₂). Membrane; a slightly increased ATPase when the holoenzyme is formed [lane 3; addition of inverted membrane vesicles (IMVs) containing SecYEG (1 µM SecYEG-IMVs)]. Translocation; a fully stimulated ATPase achieved by the holoenzyme when a functional preprotein is added (lane 5; further addition of 10 µM proPhoA). Addition of 10 µM HSI#6 had barely an effect on the basal ( ~ 12%; compare lane 2 to 1), stimulated the membrane ATPase by ~3folds (compare lane 4 to 3) and no effect on the translocation (compare lane 6 to 5).

To further probe into these results we examined the effect of varying the HSI#6 concentration on the V_max and K_m of the SecA ATPase (V_max=maximum enzymatic velocity; K_m=substrate concentration at half V_max). To this end, the basal, membrane and translocation ATPases were measured (0.4 µM SecA₂, 1 µM SecYEG-IMVs; 10 µM proPhoA; 37 °C; 10 min basal, membrane / 1 min translocation) at different ATP and HSI#6 concentrations (Fig. 1C, E, G; as indicated). Experimental measurements were fitted to: i. Equations 1a, 1b (Methods) to determine the V_max, K_m and K_cat (ATP turnover) for each HSI#6 concentration (Fig. 1C–H; Supplementary Table 1); ii. Equation 2 (Methods) to calculate the α and β parameters (Supplementary Fig. 1A) that determine changes in K_m and V_max respectively and allow classification of the HSI#6 mode of action; iii. Equation 3 (Methods) to get apparent K_d values for the binding of HS#6 onto SecA (K_{d app}; Fig. 1D, F, H) and; iv. Equation 4 (Methods) to get an AC₅₀ value (the HSI#6 concentration required to increase ATP turnover by 50%; Fig. 1F).

HSI#6 decreased the K_m of the basal ATPase (α ~ 0.4) in a dose dependent manner, without affecting the V_max (β ~ 1) (Fig. 1D; Supplementary Fig. 1A and Supplementary Table 1). The K_m decrease indicated improved affinity for ATP and suggested allosteric enzyme activation similar to previously reported activators^30–33. The K_{d app} that was determined for the basal reaction ( ~ 4.88 µM; Fig. 1D), similar to those determined for membrane ( ~ 5.99 µM; Fig. 1F) and translocation ( ~ 6.13 µM; Fig. 1H), suggested relatively tight compound-to-SecA binding. According to the general modifier mechanism³⁴ ligands that activate enzymes without being essential for the enzymatic activity can be K- or V-/mixed-type (affecting K_m or V_max/ both, respectively)^34–37. HSI#6 is a non-essential, K type activator for soluble SecA₂ (Supplementary Fig. 1A).

The SecYEG channel acts itself as an allosteric activator causing both a K_m decrease and a slight K_cat increase (at 0 µM HSI#6: membrane K_m ~ 14 µM / K_cat ~ 6.7 min^-1 versus basal K_m ~ 37.2 µM / K_cat ~ 5.4 min^-1; Supplementary Table 1)³⁸. On top of this activation, HSI#6 altered both the K_m (α ~ 0.33) and the V_max (β ~ 2.9) of the membrane ATPase in a dose dependent manner, switching into being a mixed type activator for the holoenzyme (Fig. 1F; Supplementary Fig. 1A and Supplementary Table 1). Using non-linear curve fitting on the K_cat changes we determined an AC₅₀ = 3.76 ( ± 0.72) µM (Fig. 1F).

The preprotein, a known allosteric activator of the translocase¹⁵, sharply increased the V_max without affecting the K_m (at 0 µM HSI#6: translocation K_cat ~ 125.9 min^-1 versus membrane K_cat ~ 6.7 min^-1; Supplementary Table 1), in agreement with previous work³⁹. However, unlike HSI#6, the preprotein (10 µM) does not activate soluble SecA₂ (basal; +proPhoA; Supplementary Table 1). HSI#6 decreased the K_m (α ~ 0.55) of the translocation ATPase yet, notably less than the change it induced in the membrane ATPase (α ~ 0.33) and had no significant effect on the V_max (β ~ 1) of the translocation ATPase (Fig. 1H; Supplementary Fig. 1A and Supplementary Table 1).

HSI#6 did not affect the in vitro activity of any other enzyme tested by us [Supplementary Fig. 1B, DnaK (lanes 1-2) and Alkaline phosphatase (AP; lanes 3-4) are shown], or others⁴⁰. Therefore, we assume it to be SecA-specific.

Our results demonstrated that HSI#6 binds with micromolar potency onto SecA, and acts as a non-essential, allosteric activator, affecting all forms of the enzyme; soluble, membrane-bound and ‘at-translocation’.

HSI#6 enhances SecA dependent preprotein secretion, in vivo and in vitro

To determine whether HSI#6 affects protein secretion per se we used in vitro translocation assays¹⁵. SecA, SecYEG-IMVs and preprotein were incubated,  ± 10 µM HSI#6 (0.4 µM SecA_2; 1 µM SecYEG-IMVs; 10 µM proPhoA; 1 mM ATP; 10 min; 37 °C). Samples were treated with proteinase K, TCA-precipitated and analyzed by SDS-PAGE. Proteins that were translocated into the lumen of IMVs survived the Proteinase K treatment, unless Triton X-100 was added, were immuno-stained with anti-PhoA antibodies and signals quantified using ImageJ. The amount of translocated proPhoA was expressed relative to the input amount of proPhoA (set as 100%) and plotted for different conditions (Fig. 1I). Under routine conditions proPhoA was translocated up to ~20% (lane 2) of the input (lane 1). In the presence of 10 µM HSI#6 this amount reached 40% of the input (lane 4). No translocation occurred in the absence of SecA (lane 3). Translocated proPhoA became susceptible to Proteinase K in the presence of Triton X-100 (lane 5). The amount of translocation was time- and HSI#6 concentration- dependent (Supplementary Fig. 1C, D). Moreover, the Sec translocase became more tolerant to sodium azide (NaN₃), a known inhibitor of SecA-dependent secretion in vitro and in vivo^41,42, in the presence of 10 µM HSI#6 (Supplementary Fig. 1E). The HSI#6-effect on preprotein translocation was similar for a wild-type or a Prl translocase (protein localization^23,43; see below; Supplementary Fig. 1F).

To test whether HSI#6 affected protein secretion in vivo we monitored the Alkaline Phosphatase (AP) activity. AP becomes enzymatically active only after its preform, proPhoA, is secreted in the periplasm hence, this method has been used to report on the in vivo secretion of proPhoA derivatives^15,44. To this end, Enteropathogenic E. coli (EPEC) cells, which are naturally resistant to the HSI#6 antibacterial effect²⁹, were transformed with a plasmid carrying proPhoA under the control of an arabinose promoter. Following induction of expression in the absence/presence of 10 µM HSI#6, or/and 4 mM NaN₃, the AP activity of 10^8 cells was measured using a colorimetric assay⁴³ (Fig. 1J; as indicated). The activity in the absence of inducer, due to chromosomal expression (Supplementary Fig. 1I, lane 1), was subtracted from all samples. The AP activity in vivo increased by ~3.5 folds in the presence of HSI#6 (Fig. 1J, compare lane 3 to lane 1) and was NaN₃ sensitive (compare lanes 2 to 1 or 4 to 3) demonstrating its dependency on SecA⁴¹. The increase on the AP activity was HSI#6 dose dependent (Supplementary Fig. 1G), even in the presence of 4 mM NaN₃ (Supplementary Fig. 1H).

To examine whether HSI#6 affected the expression of proPhoA in vivo, the same amount of cells from the previous measurements,  ± HSI#6, were analyzed on SDS-PAGE, immuno-stained (α-PhoA) and signals quantified using a standard curve of purified PhoA. No significant difference in the PhoA amount between the two conditions was observed (Fig. 1K). Since HSI#6 did not affect the expression of proPhoA (Fig. 1K) nor altered the enzymatic activity of AP (Supplementary Fig. 1B) and AP activity is SecA dependent (Fig. 1J), we presume that AP activity changes likely reflect changes in proPhoA secretion in vivo.

HSI#6 enhanced SecA dependent preprotein secretion, in vivo and in vitro.

HSI#6 allows the wild-type translocase to bypass the need for signal peptides, in vivo and in vitro

Clients without signal peptides cannot be secreted by the Sec translocase, since signal peptide binding is needed to lower the activation energy of SecA¹⁵. Only Prl mutants, on secA (prlD) or secY (prlA), that structurally mimic the low activation energy state of the translocase, can secrete signal peptide-less clients^43,45–47. Nevertheless, as the synergy of signal peptide and mature domain binding is required for post-activation events, Prl mutants fail to restore mature domain secretion to preprotein levels^15,47.

In vitro translocation assays, similar to those used for proPhoA (Fig. 1I), show that PhoA or MBP are translocated by the PrlA4 translocase [0.4 µM SecA_2; 1 µM SecY(PrlA4)EG-IMVs; 10 µM client; 1 mM ATP; 10 min; 37 °C] to ~10% of their input (Fig. 2A, lane 2), whereas proPhoA is secreted to ~40% of input by the same translocase (quantification in Supplementary Fig. 2B). In the presence of 10 µM HSI#6 translocation increased to ~50% of input for PhoA and ~30% of input for MBP (Fig. 2A, lane 4; Supplementary Fig. 2A, B) indicating that the compound likely bypassed more secretion-limiting steps, compared to the prl phenotype¹⁵.

Fig. 2. HSI#6 allows the wild-type translocase to secrete signal peptide-less secretory proteins, in vitro and in vivo.

A, B in vitro, SecA dependent secretion of PhoA or MBP (as indicated) by the PrlA4 (A) or wild-type (B) translocase,  ± 10 µM HSI#6 (as in 1I). Lane 1: 20% of PhoA/5% of MBP input; lane 2: No HSI#6; lanes 3-5:10 μM HSI#6; lane 3: No SecA; lane 5: Triton X-100 control. n ≥ 3. A representative experiment is shown. Quantifications in Supplementary Figs 2-3. C in vivo AP activity (top) and expression level (bottom) of cells overexpressing PhoA (lanes 1-2), or PhoA plus secY(PrlA4)EG (lanes 3-4),  ± 10 μM HSI#6 (as in 1 J). n ≥ 3. Error bars represent standard deviation. A-C All samples contained the same final DMSO concentration.

As anticipated, PhoA or MBP cannot be secreted by the wild-type translocase (Fig. 2B, lane 2) [0.4 µM SecA_2; 1 µM SecYEG-IMVs; 10 µM client; 1 mM ATP; 10 min; 37 °C]. In sharp contrast, in the presence of 10 µM HSI#6 both are secreted (lane 4) (Supplementary Fig. 3A, B for quantification). HSI#6 "made"| the wild-type translocase "behave" like a Prl one, allowing it to bypass the need for a signal peptide. Secretion of signal-less proteins by the wild-type translocase in the presence of HSI#6 remained SecA- (Fig. 2B, lane 3), ATP- and physiological temperature- dependent and required unfolded clients (Supplementary Fig. 3C). Accumulation of translocated PhoA was time dependent (Supplementary Fig. 3D) and became Proteinase K susceptible in the presence of Triton X-100 (Fig. 2B, lane 5).

To test whether HSI#6 affected mature domain secretion by the wild-type translocase, in vivo, we monitored the in vivo AP activity. EPEC cells were transformed with a plasmid carrying PhoA under the control of an arabinose promoter. PhoA expression was induced,  ± 10 µM HSI#6 and the AP activity was measured (as in Fig. 1J). We observed that in the presence of HSI#6 the AP activity significantly increased (Fig. 2C, top, compare lane 2 to 1, top) in a time dependent manner (Supplementary Fig. 3E). No difference in the expression level of PhoA  ± HSI#6 was observed (Fig. 2C, bottom, compare lane 2 to 1, as in 1 K).

To study whether HSI#6 affected mature domain secretion by the Prl-translocase, in vivo, the AP activity was measured in EPEC cells carrying PhoA on a plasmid that were co-transformed with a second plasmid carrying the secY(prlA4)EG operon under the control of T5 promoter. Gene expression was induced by concomitant addition of 10 μM arabinose and 0.05 mM IPTG, - / 10 µM HSI#6, and the AP activity was measured (as in 1 J). We observe that the AP activity becomes significantly enhanced in the presence of HSI#6 (Fig. 2C, top, compare lane 4 to 3). The difference between lane 3 and 1/lane 4 and 2 is attributed to secretion by the PrlA4 translocase. No difference in the expression level of PhoA between  ± HSI#6 was observed (Fig. 2D, bottom, compare lane 4 to 3; as in 1 K).

In the presence of HSI#6 the wild-type translocase can secrete mature domains as efficiently as preproteins, in vivo and in vitro, bypassing known limiting steps of "canonical" secretion.

HSI#6 allows the wild-type translocase to secrete unfolded cytoplasmic proteins, in vivo and in vitro

Multiple checkpoints secure that cytoplasmic proteins are not secreted: The Sec translocase secretes unfolded clients and cytoplasmic proteins typically fold fast⁴⁸. Preprotein targeting signals (signals peptides and mature domain regions) are absent from cytoplasmic proteins⁸. Even if cytoplasmic clients reach the site in unfolded states, in the absence of a signal peptide they fail to lower the activation energy of the translocase, activate ATP catalysis and become trapped or secreted¹⁵.

As anticipated, under routine in vitro translocation conditions [0.4 µM SecA; 1 µM SecYEG or SecY(PrlA4)EG; 10 µM unfolded in 6 M urea cytoplasmic protein; 1 mM ATP; 10 min; 37 °C] neither the PrlA4 (Fig. 3A, lane 2) nor the wild-type translocase (Fig. 3B, lane 2) can secrete the cytoplasmic proteins PpiB or Rs12 (as indicated), nor SecB, TF, DnaK and DnaJ (Supplementary Figs. 2A and 3A, respectively). Surprisingly, in the presence of HSI#6 both translocases can secrete all proteins, almost equally well to preproteins (Fig. 3A, B, lane 4; Supplementary Figs. 2-3, A-B).

Fig. 3. HSI#6 allows the wild-type translocase to secrete cytoplasmic proteins, in vitro and in vivo.

A, B In vitro, SecA dependent secretion of PpiB or Rs12 (as indicated) by the PrlA4 (A) or the wild-type (B) translocase,  ± 10 μM HSI#6 (as in 1I). Lane 1: 20% of PpiB/Rs12 input; lane 2: No HSI#6; lanes 3-5:10 μM HSI#6; lane 3: No SecA; lane 5: Triton X-100 control. n = 3. A representative experiment is shown. More cytoplasmic protein examples and quantification in Supplementary Figs 2-3. C in vivo AP activity (top) and expression level (bottom) of cells overexpressing PpiB_PhoA (lanes 1-2), or Rs12_PhoA (lanes 3-4), - / + 10 μM HSI#6 (as in 1 J). n = 3. Error bars represent standard deviation. A–C All samples contained the same final DMSO concentration.

To study the effect of HSI#6 on the secretion of cytoplasmic proteins in vivo, we took advantage of a prior observation. A cytoplasmic protein that was fused N-terminally to PhoA became the "determinant" for the secretion of the fusion protein⁴⁹. To repeat this observation, EPEC cells were transformed with a plasmid that carried PhoA, or PpiB that was fused N-terminally to PhoA (hereafter PpiB_PhoA), or/and co-transformed with a second plasmid carrying the secY(prlA4)EG operon (as in Fig. 2C). Following induction of gene expression (10 μM arabinose; 0.05 mM IPTG; 30 min, 37 °C) the AP activity of 10^8 cells was measured (as in Fig. 1J). Unlike PhoA, PpiB_PhoA could not be secreted by the PrlA4 translocase (Supplementary Fig. 3F)⁴⁹.

Next, EPEC cells were transformed with a plasmid that carried PpiB_PhoA or Rs12_PhoA. Following induction of gene expression,  ± 10 μM HSI#6, the in vivo AP activity was measured and is plotted after subtraction of the chromosomal AP activity (as in Fig. 1J). We observe that in cells that overexpressed either fusion the AP activity was significantly increased in the presence of HSI#6 (Fig. 3C, top). No difference in the expression level of the fusion proteins -/+ HSI#6 was observed (Fig. 3C, bottom; as in 1 K).

HSI#6 allowed secretion of cytoplasmic proteins by the Sec translocase, suggesting that client-selectivity-restrictions that are imposed on canonical secretion are now bypassed, in vivo or in vitro.

HSI#6 uncouples activation of ATP catalysis from preprotein binding

In the canonical secretion SecYEG binding primes the SecA motor without increasing catalysis whereas, preprotein binding allosterically activates ATP catalysis and couples it to secretion^6,10. To visualize this, the ratio of K_cat values for the different SecA ATPase activities (Supplementary Table 1) was plotted. The ratio of Membrane / Basal (M / B) K_cat is ~1 (Fig. 4A, lane 1) indicating no-activation of ATP catalysis by SecYEG binding while the Translocase / Membrane (T / M) K_cat ratio is >>1 (lane 2) indicating activation of ATP catalysis by the preprotein.

Fig. 4. HSI#6 activates ATP catalysis of the holoenzyme in the absence of clients.

A Effect of 10 µM HSI#6 (as indicated) on the stimulation of ATP turnover by SecYEG or preprotein addition (Membrane / Basal; Translocation / Membrane K_cat ratios respectively; calculated from Supplementary Table 1 values). B Effect of 10 µM HSI#6 (as indicated) on the stimulation of ATP turnover (top; Translocation / Membrane K_cat ratios) or the in vitro secretion (bottom; quantification with error bars representing standard deviation from Supplementary Figs 3A-B; n ≥ 3) using the indicated clients. Top: For proPhoA, PhoA and PpiB K_cat values (Supplementary Table 1) were used; for the remaining clients apparent K_cat values with errors representing standard deviation are shown. C Effect of HSI#6 concentration (x axis) on the ATP turnover stimulation (Translocation / Membrane K_cat ratios; from Supplementary Table 1) by the indicated clients. D. Effect of HSI#6 concentration (x axis) on the catalytic efficiency of SecA (as indicated; K_cat / K_m; Supplementary Table 1). A-D Wild-type SecA and wild-type SecYEG were used. All samples contained the same final DMSO concentration.

Client inability to stimulate catalysis, as is the case for mature domains, results to no-secretion^15–17. As anticipated, the ATP turnover by the holoenzyme was not stimulated by the addition of mature domains or cytoplasmic proteins, - / + 10 µM HSI#6 (T / M K_cat ratio ~1; Fig. 4B, top). Despite this, the same clients were secreted in the presence of HSI#6 (Fig. 4B, bottom). Using two examples, the mature domain PhoA and the cytoplasmic protein PpiB, it was shown that the ATP turnover of the holoenzyme was not stimulated in a range of HSI#6 concentrations (T / M K_cat ratio ~1; Fig. 4C; Supplementary Table 1). Additionally, it was observed that while HSI#6 decreased the preprotein stimulated ATP catalysis (Fig. 4B, top, proPhoA), in a dose dependent manner [Fig. 4C, proPhoA; T / M K_cat ratio decreased instead of remaining constant], its secretion increased (Fig. 4B, bottom, proPhoA). A similar effect has been reported for PrlA4^23,47.

To explain how secretion was allowed despite the absence of client-stimulated ATP catalysis (Fig. 4B, top) a step back is needed. Channel binding in the presence of HSI#6 was not restricted to priming; it extended into ATP catalysis activation (M / B K_cat ratio ~3; Fig. 4A, lane 3). The drastic increase in the catalytic efficiency (i.e. K_cat / K_m ratio) of the enzyme at the level of SecYEG binding, in an HSI#6 dose dependent manner, demonstrated a catalytically-active-holoenzyme in the absence of client (Fig. 4D; Supplementary Table 1). Unlike the preprotein that drastically enhances the K_cat to maximize the catalytic efficiency of SecA, HSI#6 allows SecA to achieve similar catalytic efficiency by causing only a moderate change in the K_cat that is coupled to a drastic K_m change. By shifting the translocase activation upstream of the client-induced-activation the client-selectivity is lost, because there is no need for preprotein synergistic binding anymore. We propose that HSI#6 allosterically activates ATP catalysis of the holoenzyme and this activated-translocase does not need to await client-specific-activation anymore, it can secrete any unfolded client that comes along.

HSI#6 enhances ADP release from SecA

HSI#6 binding likely allows the translocase to overcome a rate - limiting step. ADP release is such a step^5,13,39,50, and in the physiological reaction is achieved by the synergistic, and not the individual, binding of signal peptide and mature domain¹⁷.

We monitored the effect of HSI#6 on the release of MANT-ADP from SecA₂:SecYEG. The fluorescence intensity of 1 µM MANT-ADP increased upon addition of the holoenzyme (Fig. 5A, black arrow; 0.4 µM SecA_2, 1 µM SecYEG-IMVs preincubated for 2 min, ice) and remained stable (grey trace), indicating tight ADP binding^10,51. Addition of 50 µl buffer - 1 µM MANT-ADP without (grey trace) or with DMSO (yellow trace) had no significant effect on the fluorescence of MANT-ADP. Addition of 10 µM HSI#6 (50 µl buffer - 1 µM MANT-ADP; red arrow) caused a drop in the fluorescence intensity (orange), similarly to the effect of adding 10 µM proPhoA (purple; 50 µl buffer - 1 µM MANT-ADP), indicative of MANT-ADP release¹⁷. The effect of two higher HSI#6 concentrations can be seen in Supplementary Fig. 4A.

Fig. 5. HSI#6 causes ADP release from membrane bound or soluble SecA.

A–B MANT-ADP (1 µM; black trace) binds to the wild-type translocase (A; grey trace; 0.4 µM SecA₂; 1 µM SecYEG-IMVs) or to the wild-type SecA₂ (B; grey trace), at 25^oC, increases its fluorescence intensity and then remains relatively stable. The black arrow indicates time of translocase (A) or SecA₂ (B) addition. Subsequent addition of 1 µM MANT-ADP buffer (50 µl; red arrow) containing preprotein (10 µM; purple trace), or HSI#6 (10 µM; orange trace), caused an immediate drop in the fluorescence intensity, while addition of only buffer (grey trace), or buffer with DMSO (yellow trace), did not have that effect. All samples, except the grey trace-sample (buffer), contain the same final DMSO concentration. n = 3. Repeats were normalized (to 0% = MANT-ADP intensity; 100% = MANT-ADP plus partner intensity) and averaged. The effect of higher HSI#6 concentrations is shown in Supplementary Fig. 4.

The preprotein cannot release MANT-ADP from soluble SecA₂ (Fig. 5B, purple; 10 µM proPhoA in 50 µl buffer - 1 µM MANT-ADP)^17,38. On the contrary, HSI#6 can (orange; 10 µM; 50 µl buffer - 1 µM MANT-ADP). Addition of buffer without (grey) or with DMSO (yellow) had no significant effect on the fluorescence of MANT-ADP. The effect of two higher HSI#6 concentrations can be seen in Supplementary Fig. 4B.

In the canonical pathway, client-induced-ADP-destabilization lies behind the ATP turnover stimulation during translocation (Fig. 4A, lane 2)¹⁷. Likewise, HSI#6-induced-ADP-destabilization leads to ATP turnover stimulation at the level of SecYEG binding (Fig. 4A, lane 3) and lies behind the drastic increase in the catalytic efficiency of the holoenzyme in the absence of a client (Fig. 4D; membrane). The fact that ADP destabilization on soluble SecA₂ (Fig. 5B) does not lead to activation of ATP catalysis (Fig. 1D; K_cat) argues that additional barriers towards activation can be bypassed only by the holoenzyme.

HSI#6 activates the translocase by modulating the intrinsic dynamics of SecA

To gain insight on the mechanism by which HSI#6 allosterically activates the translocase we used local Hydrogen Deuterium Exchange Mass Spectrometry (HDX-MS)^10,17. SecA₂ was preincubated with SecYEG-IMVs (1:1.5 molar ratio) in the presence of ADP (2 mM final concentration) for 2 min on ice. The D-uptake of SecA₂:ADP bound on SecYEG, - / + 20 µM HSI#6, at 30^oC, was determined (Supplementary Data 1). The % of D-uptake for either condition was calculated by normalization to a fully deuterated control (set as 100%) and the differential D-uptake between the two conditions determined (ΔD; %ΔD; Supplementary Data 1). %ΔD values were plotted on the ecSecA_2VDA structure using a color code that indicates regions with enhanced / suppressed dynamics (green / purple respectively) in the presence of HSI#6 (Fig. 6A), quantified as minor or major (%ΔD; 10-20% light hues; >20% dark hues).

Fig. 6. HSI#6 affects the local dynamics of membrane bound SecA.

A–B The D-uptake differences (∆%D) caused by the presence of 20 µM HSI#6 on the wild-type SecA bound either as a dimer (A) or as a monomer (B) on wild-type SecYEG (1:1.5 molar ratio), determined by HDX-MS, are shown on the ecSecA structure (PDB: 2VDA) using a color code (as indicated). Decreased / increased dynamics: purple / green respectively; no difference: white; light / dark hues: ΔD = 10-20% / > 20% respectively. Domains / sub-structures of interest and ADP are indicated. Pictograms: “reference” (top); “test” (bottom); orange circle: ADP. All samples contained the same final DMSO concentration. n = 2 biological repeats, each with technical triplicates (see Supplementary Data 1 for HDX data / experimental conditions). To enable comparison with the changes a preprotein causes on the wild-type dimer SecA bound on wild-type SecYEG, a reprint of Fig. 1C from¹⁷ is included in Supplementary Fig. 5.

HSI#6 increased the dynamics (green) of SecYEG-bound SecA₂ specifically at three regions of the Preprotein Binding Domain [helix α10, 3β-tip with part of α13 and Stem_in (β12)], the proximal IRA1 tip and residues of the far away α0 / α1 elements, required for quaternary conformational transitions¹⁶ (Fig. 6A). The rigidification (purple) on three turns of the scaffold helix, a region typically destabilized by SecYEG binding¹⁰, might suggest a proximal binding site for the compound. Given that our functional assays show that HSI#6 activated the holoenzyme, the absence of increased dynamics within the ATPase motor¹⁷ surprised us. Furthermore, high PBD dynamics suggested increased motions, a signature of translocase activation and monomerization^10,17. In the canonical pathway monomerization is a late event; docking on SecYEG and subsequent preprotein activating events occur with dimeric SecA¹⁶.

We entertained the possibility of HSI#6 inducing early monomerization and compared the dynamics of HSI#6 bound SecA:SecYEG:ADP to mSecA:SecYEG:ADP. These dynamics seem more consistent with an activated translocase (Fig. 6B): increased dynamics in Motif IVa, the signature for triggered translocase¹⁷; scaffold dynamics that are identical to the ones seen with preprotein (increased at the kink / decreased towards NBD2; compare to Supplementary Fig. 5); increased dynamics in parts of α0 / α1 (involved in quaternary conformational transitions¹⁶); increased dynamics in the Stem_out (β6) that start from α10 and extend to the bottom of Gate 2 within the motor, that are a unique HSI#6 signature; increased dynamics in PBD that are restricted to α10 (as with preprotein-induced low activation energy; compare to Supplementary Fig. 5). A rigidification in the Stem_in (β12) that extends into PBD and the proximal IRA1 region, both below the scaffold kink, might be due to a proximal HSI#6 binding site).

Our results demonstrate that HSI#6 binding alters the dynamics of the holoenzyme distinctly from preprotein binding (compare Fig. 6A, B to Supplementary Fig. 5). Presumably, these changes lie behind the altered holoenzyme properties (K_m, V_max, K_cat). While the HSI#6 induced dynamics on a SecA₂– bound to SecYEG cannot explain how secretion occurs in the absence of increased Motif IVa dynamics¹⁷, a mSecA- model can. Monomerization of SecA₂ bound on the channel by HSI#6 is an intriguing hypothesis that needs to be tested experimentally.

Discussion

Protein secretion is a fundamental process of life, with the Sec pathway being the main essential secretion route. Ubiquitous in bacteria and essential for protein secretion, the SecA subunit of the translocase makes an attractive target for antibiotics^26–28. In a recent screen a potent small-molecule-antibacterial, HSI#6, has been identified²⁹.

Here, we studied the effect of HSI#6 on bacterial secretion and found that it activates the soluble, membrane-bound and at-translocation-site SecA states (Fig. 1C–H; Supplementary Fig. 1B and Supplementary Table 1). The fact that HSI#6 activated the SecA subunit of the translocase in minimal, in vitro basal ATPase assays that only employ the purified enzyme, demonstrates a direct interaction between the two, with micromolar affinity (K_{d app} ~ 5-6 µM). HSI#6 shares features with non-essential activators and switches from being a K-type activator for soluble SecA_2, to mixed type activator for the holoenzyme, and then back to K-type activator in the presence of a preprotein (Fig. 1D, F, H; Supplementary Fig. 1B and Supplementary Table 1)³⁶.

Non-essential activators, like HSI#6, bind allosterically and achieve enzyme upregulation through diverse mechanisms of action. The simplest/most common means to activate dormant enzymes is to bind on an allosteric site of the catalytic domain, often cooperatively with the substrate, and induce a conformational change that stabilizes an active enzyme conformation³⁵. Examples include activators for the Glucokinase^52,53, sirtuins and the phosphoinositide dependent protein kinase 1^36,54. Alternatively, activators bind to a regulatory subdomain and promote the activity of the catalytic domain, like the AMP-activated protein kinase^55–58 or the Protein Kinase A⁵⁹ activators. Finally, activators may facilitate oligomerization which in turn promotes enzyme activation, as is the case with RNase L activators⁶⁰.

An enzyme might possess allosteric activation sites, even multiple ones, but these are difficult to predict^35,61. In the absence of a SecA₂:HSI#6 structure the binding site / mechanism of action remain elusive. A rigidification seen by HDX-MS experiments for Scaffold / IRA1 regions nearby the Stem_out might indicate a candidate binding site (Fig. 6). The Stem of SecA is a central checkpoint of allosteric networks¹⁰. HSI#6 did not interrupt binding of known translocation ligands (nucleotides, SecYEG, clients; Supplementary Table 1) and increased the affinity of SecA for the nucleotide (K_m decreases; Figs. 1D, F, H). Separate measurements are necessary to determine whether it also improves the affinity for SecYEG / clients and disentangle individual from synergistic effects. Between SecYEG and HSI#6 a synergy was established; while the α remained similar in soluble- and membrane-bound-enzyme the β increased for the holoenzyme from 1 to ~ 2.9 (Supplementary Fig. 1A). This synergy is the reason for successful secretion of mature domains/cytoplasmic proteins whose addition could not level up the V_max of the ATPase (Fig. 4B, top). The addition of a third allosteric regulator in the reaction (the signal peptide) "disturbs" the synergy that was previously established between the other two regulators (HSI#6 and SecYEG); the β returns to ~1, while α remained at ~0.5 (Supplementary Fig. 1A).

Like preproteins do, HSI#6 exploits the dynamics of SecA, a behavior that is common for allosteric regulators⁶². Whether we assume that HSI#6:SecA bound to SecYEG is a dimer, as it has been demonstrated for the canonical pathway¹⁶, or monomerized early, the signature of HSI#6 induced dynamics is unique (Fig. 6). We favor the idea of HSI#6 stabilizing a monomer SecA on SecYEG, in the absence of clients, because this dynamics signature (Fig. 6B) is closer to the signature of a preprotein-activated-translocase (Supplementary Fig. 5)¹⁷. Without any doubt, the real stoichiometry under these conditions needs to be addressed experimentally.

In the canonical protein secretion, binding of SecA at SecYEG only primes the helicase motor without increasing ATP catalysis (Fig. 4A, lane 1)^5,6,10. Client selectivity comes by the mere rule that only legitimate-client binding (i.e. preproteins) can lower the activation energy of SecA, release ADP thus activate ATP catalysis and couple these events to secretion^10,15. HSI#6 "changed the rules of the game". By modulating the intrinsic dynamics of SecA, HSI#6 activated the holoenzyme to its full catalytic potential before the client arrives thus, it uncoupled all activation steps from client binding (Figs. 2B-C, 3B-C, 4D, 5, 6). By stabilizing a constitutively activated translocase in a client independent way, HSI#6 transformed the Sec translocase into a promiscuous nanomachine that lost client selectivity and would secrete any protein that comes along as long as is unfolded enough to fit into the channel (Figs. 1–3; Supplementary Figs 2-3). This HSI#6 phenotype is the first one reported for the Sec translocase. In a similar example, acyldepspeptides (ADEPs) stabilized an active form of the ClpP proteolytic chamber turning it into a constitutively degrading machinery that resulted in antibacterial action^63–66.

Many questions are still pending. An obvious one would be the effect of HSI#6 on the SecYEG channel. It is anticipated for holoenzymes that conformational changes are transmitted between partners. An "|open'' or/and more active state of the channel might also be stabilized upon SecA:HSI#6 binding. Attempts to resolve the structure of HSI#6 complexed with SecA or/and SecA:SecYEG are under way. Small-molecule activators that modulate the structural dynamics of essential enzymes like SecA can provide a unique resource for biochemical and mechanistic dissection. Such compounds can pave the way for in silico design of inhibitors that would abrogate activation mechanisms leading to successful antibacterials or/and of new activators that would enhance protein secretion directly without the cost of genetic manipulations and benefit various biotechnological applications⁶⁷.

Methods

Materials availability

This study did not generate new unique reagents.

Chemical synthesis of HSI#6

The protocol and quality controls for HSI#6 synthesis can be seen in Supplementary Note.

Strains/plasmids/DNA primers/Antibodies/Buffers

The complete list of buffers (Supplementary Table 2), strains (Supplementary Table 3), plasmids (Supplementary Table 4), DNA primers (Supplementary Table 5) and Antibodies (Supplementary Table 6), that were used in this study is provided.

Cell growth, gene overexpression and cell lysis

E. coli K12 expression cells transformed with plasmid carrying the indicated client derivative gene were grown in LB (37 °C; OD₆₀₀ = 0.6) before gene expression was induced (0.2 mM IPTG; 3 h; 30^oC). For preproteins, 4 mM Sodium Azide was added in the cell culture 5 min prior to IPTG addition, to abolish SecA-dependent secretion and prevent signal peptide cleavage⁴¹. Cells were collected (4,000 x g; 4 °C; 15 min; Avanti J-26S XPI, JLA 8.1000 rotor; Beckman) and stored at −20 °C until purification.

Purification

SecA purification: Untagged SecA was purified as previously described¹⁶. Following cell lysis, the supernatant was loaded on a pre-equilibrated home-prepared Cibacron Blue resin [2 ml / min; Sigma;⁸]. The column was washed with Buffer A (10 column volumes; ÄKTA Pure System; GE Healthcare) and SecA was eluted using a linear gradient (Buffer A to Buffer B; 4 column volumes; 2 ml/min; 5 ml fractions) and concentrated. NaCl was adjusted to 1 M final concentration and following DTT treatment (20 mM; 15 min; 4 °C), SecA was loaded on a preparative HiLoad Superdex 200 (GE Healthcare), in Buffer C. Fractions (5 ml) were collected, concentrated, treated with 10 mM DTT and re-loaded on a HiLoad Superdex 200 that was pre-equilibrated with Buffer D. Fractions containing SecA were pooled and dialyzed in Buffer E (12 h; 4 °C). Protein aliquots were stored at −20 °C. SecB purification: Untagged SecB was purified as previously described⁶⁸. Briefly, following cell lysis, the supernatant was loaded on a 5 ml column packed with Q resin (1 ml/min; GE Healthcare) and washed sequentially with 10 column volumes of Buffer F, Buffer G and Buffer N⁸. SecB was eluted with Buffer H and further purified using preparative HiLoad Superdex 200 (ÄKTA Pure System; GE Healthcare), in Buffer C. Fractions (5 ml) were collected, pooled and dialyzed in Buffer I (12 h; 4 °C) then in Buffer E (12 h; 4 °C). Protein aliquots were stored at −20 °C. Client purification: The His-tagged derivatives of clients were purified under denaturing (proPhoA) or native (PhoA, MBP, PpiB, TF, DnaK, DnaJ, Rs12) conditions, following the manufacturer’s instructions (Qiagen) and stored in buffer J or E respectively, at −20 °C. Before being added to in vitro secretion assays proteins were dialyzed in buffer J (12 h; 4 °C) and then buffer K (2 h). The quality of purification was assessed by gel filtration chromatography or / and SDS-PAGE ( > 95% purity). IMVs purification: IMVs were prepared from cells overexpressing wild-type SecYEG or SecY(PrlA4)EG, as described¹⁰. Following cell lysis by French press, unbroken cells were removed (4,000 x g; 10 min; Sigma 3-16KL; rotor 11180) and the supernatant was ultra-centrifuged (140,000 x g; 90 min; 4^oC; fixed angle 45 Ti rotor; Optima XPN-80, Beckman). The membrane pellet was resuspended in 50 mM Tris-Cl pH 8.0 using a Dounce-homogenizer, loaded (2 mL) on top of a 5-step sucrose gradient (1.9; 1.7; 1.5; 1.3; 1.1 M sucrose in 50 mM Tris-Cl pH:8.0; 6 mL each layer) and centrifuged to equilibrium (124,000 x g; 16 h, 4^oC; swinging bucket SW 32 Ti rotor, Optima XPN-80, Beckman). IMVs, collected from gradient fraction 2, were resuspended in buffer L and re-centrifuged (140,000 x g; 90 min; 4^oC; fixed angle 45 Ti rotor; Optima XPN-80, Beckman). The membrane pellet was homogenized in 6 M Urea, 50 mM Tris-Cl pH 8.0 (30 min; ice), loaded on top of equal volume buffer M and centrifuged (150,000 x g; 90 min, 4 °C; swinging bucket SW 32 Ti rotor, Optima XPN-80, Beckman). IMVs were collected, homogenized in buffer N and extruded through a 100 nm pore size filter (15-21 times; Avestin LiposoFast-Basic system), to obtain vesicles of unilamellar diameter, and stored in aliquots, at -80 °C. The SecY concentration in these preparations was determined as described¹⁶.

Determination of protein concentration

Spectroscopic measurements (Nanodrop 2000; Thermo) in the range of 0.5-5 mg/ml (buffer F, or K for preproteins), that is within the linear range of 0.1-1 for OD₂₈₀, were routinely used. The molecular weights/extinction coefficients for the various proteins were determined using the (http://web.expasy.org/protparam) server. For low extinction coefficient / low concentration protein samples the Bradford assay (Bio-Rad) was used in combination with a BSA linear control standard curve (0.5-20 µg). Ultrafiltration devices (Amicon Ultra-15, 3 K cut-off, Millipore) were used for protein sample concentration (3,000 x g; Sigma 3-16KL; 4 °C).

Analysis of protein structures

Protein structures were visualized with Swiss PDB viewer and PyMol softwares. Structures with PDB codes were downloaded from RSCB, Protein Data Bank (http://www.rcsb.org/). Images were generated with maximum quality, antialias set at 2, ray trace mode 1 and ray-traced at 1500, 1500. FASTA protein sequences were downloaded from (http://www.ncbi.nlm.nih.gov/protein).

Western blot analysis

Following SDS-PAGE proteins were transferred onto nitrocellulose (PROTRAN; Thermo Scientific) using Semi Dry Transfer (BIORAD; 20 V; 20 min) and immunostained with polyclonal specific antibodies (Supplementary Table 4). Membranes were incubated with SuperSignalTM West Pico PLUS chemiluminescent substrate (Thermo Scientific; 2 min). Images were acquired using the manufacturer’s settings (Las4000; GE Healthcare; Resolution / Sensitivity: standard mode; Exposure time 1 min; Image dimensions 210 × 140 mm; Image resolution 176 dpi), processed using canvas X GIS 2020 and signals were quantified using ImageJ (https://imagej.net).

ATPase measurements

As previously described⁶⁹. For basal ATPase, reactions (50 μl buffer P; 0.4 μM SecA or 0.6 μM DnaK; 1 mM ATP unless otherwise indicated) containing HSI#6 (as indicated) were incubated (SecA: 10 min, unless otherwise indicated; DnaK: 4 min; 37 °C). DMSO was adjusted to the same final concentration in all samples. For membrane ATPase 1 μM SecYEG was further added. For translocation ATPase 10 μM client (unless otherwise indicated) was additionally added. Hydrolysis stopped by transferring samples on ice and adding first malachite green reagent (20 min; room temperature) and then 37% citric acid (5 min, room temperature). OD₆₆₀ values were recorded (Tecan Infinite M200). Released Pi was calculated using a standard curve from known concentrations of inorganic phosphate (Sigma) and V values (nmol Pi min^-1) were determined.

Alkaline Phosphatase (AP) assays

in vivo: Enteropathogenic E.coli (EPEC) cells transformed with pBAD501 plasmids carrying proPhoA, or PhoA, or the indicated client-PhoA fusions either alone, or co-transformed with a pET610 plasmid carrying the SecY(PrlA4)EG operon⁴³ were grown to OD₆₀₀ = 0.25 (LB; 37 °C). Gene expression was induced [10 μM arabinose for PhoA derivatives; 0.05 mM IPTG for SecY(PrlA4)EG2; 30 min; 37 °C]. Cells were transferred on ice, centrifuged (1 ml; 1,500 × g, 8 min), redissolved in 1 M Tris–HCl pH 8.0 (500 μl) and OD₆₀₀ values recorded. p-nitrophenol phosphate was added (pNPP; 0.01 M) to 10^8 cells and samples were incubated at 37 °C until development of yellow color. Hydrolysis stopped by transferring samples on ice and adding 0.17 M K₂HPO₄. Cells were lyzed (0.17% Triton X-100) and debris was removed (15,500 × g; 5 min; 4 °C). in vitro: Pure AP (TSAP; Promega) was incubated - / + 10 µM HSI#6 (50 µl 1 M Tris–HCl pH 8.0 - 0.01 M pNPP; 37 °C; 4 min). Hydrolysis stopped by transfer on ice and addition of 0.17 M K₂HPO₄. In either case, OD₄₂₀ values were recorded (Tecan Infinite M200) and units of AP activity were calculated using the formula U = [∆OD₄₂₀ * 1000]/time⁴³.

in vitro translocation/secretion assays

As previously described¹⁵. Briefly, 50 μl reactions in Buffer P containing 0.4 μM SecA, 1 μM SecYEG IMVs, 10 μM unfolded in 6 M Urea buffer client, 1 mM ATP, - / + the indicated HSI#6 concentration, were incubated (37 °C, 10 min). DMSO was adjusted to the same final concentration in all samples. Following proteinase K treatment (1 mg/ml; 20 min; ice), TCA precipitation (15% w/v) and SDS-PAGE (13%) analysis, proteins were immuno-stained with client specific polyclonal antibodies (Supplementary Table 6) and signals quantified using ImageJ. The amount of client that became protease protected, following translocation into the IMVs lumen, was expressed relative to the input amount of client (set as 100%) and determined protein translocation/secretion.

MANT-ADP binding/release assays

MANT-ADP binding onto SecA results in a strong increase in fluorescence intensity¹⁰. Experiments were carried out using 1 mL cuvettes (buffer Q - 1 μM MANT-ADP), where 0.4 μM SecA₂ or SecA₂:SecYEG-IMVs (1:1.5 molar ratio; preincubated for 2 min, ice) was added as indicated, at 25 ^oC, in a Cary Eclipse fluorimeter (Agilent; λ_ex: 356 nm; λ_em: 450 nm; slit_ex: 2.5 nm; slit_em: 5 nm). Two min later, 50 µl of buffer Q - 1 μM MANT-ADP only, or + DMSO, or + 10 µM proPhoA, or + HSI#6 (as indicated) were added. All samples contained the same DMSO concentration (unless otherwise indicated) and fluorescence intensity was monitored for 10 min.

Hydrogen deuterium exchange mass spectrometry (HDX-MS)

As previously described¹⁰. Briefly, SecA was dialyzed overnight into buffer R and concentrated (∼100 μM; Vivaspin 500, Sartorius). To form the SecA:SecYEG state, IMVs were sonicated and incubated with SecA at 1:1.5 (SecA:SecY) molar ratio (2 min; ice), prior to D exchange. Isotope labeling was carried out using lyophilized buffer S (reconstituted in 99.9% D₂O;Euriso-top), with fresh TCEP [tris(2-carboxyethyl)phosphine; 2 mM]. Buffer pH_read was adjusted to 8.0 using NaOD (Sigma). D exchange buffer was pre-incubated in a 30 °C water bath, and the D exchange reaction was initiated by diluting 200 pmol of protein into D₂O buffer F at a 1:10 ratio (final D₂O concentration 90%). Final concentration of SecA was maintained at 4 μM in the D exchange reaction. Continuous labeling reaction was incubated for various time points (0.167, 0.5, 1, 2, 5, 10, 30 min), at 30 °C. Reactions were quenched by the addition of pre-chilled quench buffer T at a 1:1 ratio (final pH of 2.5), incubated (4 °C; 2 min) and centrifuged (20,000 x g; 90 s; 4 °C; Eppendorf). The supernatant containing SecA peptides (100 pmol) was immediately injected into a nanoACQUITY UPLC System with HDX technology (Waters, UK) coupled to a SYNAPT G2 ESI-Q-TOF mass spectrometer. For enhanced peptide coverage, SecA was digested in 2 steps, first with fungal protease XIII (Sigma) at the quench step, and then online using a home-packed immobilized pepsin (Sigma) cartridge (2 mm × 2 cm, Idex), at 16 °C. The resulting peptides were trapped onto a VanGuard C18 Pre-column, (130 Å, 1.7 mm, 2.1 ×5 mm, Waters) at 100 mL/min for 3 min using 0.23% (v/v) formic acid and then separated on a C18 analytical column (130 A°, 1.7 mm, 1 × 100 mm, Waters) at 40 mL/min. UPLC separation (solvent A: 0.23% v/v formic acid, solvent B: 0.23% v/v formic acid in acetonitrile) was carried out using a 12 min linear gradient (5-50% solvent B). At the end, solvent B was raised to 90% for 1 min for column cleaning. Peptide trapping-desalting and separation were performed at 2 °C. MS parameters: capillary voltage 3.0 kV, sampling cone voltage 20 V, extraction cone voltage 3.6 V, source temperature 80 °C, desolvation gas flow 500 L/h at 150 °C. Full deuteration controls were obtained by incubating SecA in buffer F containing 6 M Urea-d4 (98% D, Sigma) overnight at room temperature. D-uptake (%) was calculated using the full deuteration control D-uptake values. The back exchange for our instrumental set up was calculated to be between 20-45%, depending on peptide composition. The data have not been corrected for back exchange and are represented either as absolute D values or as a percent of the full deuteration control. Peptide identification was carried out using 100 pmol of protein diluted in protiated buffer F. The sample was quenched as described above and analyzed in the MSE acquisition mode in a nanoACQUITY UPLC System with HDX technology (Waters, UK) coupled to a SYNAPT G2 ESI-Q-TOF mass spectrometer over the m/z range 100-2,000 Da. The collision energy was ramped from 15 to 35 V. Other instrument parameters were as described above. Peptide identification was performed using ProteinLynx Global Server (PLGS v3.0.1, Waters, UK) and the primary sequence of SecA as a search template. Peptides were individually assessed for accurate identification and were only considered if they had a signal to noise ratio above 10, a PLGS score above 7.5, and if they appeared in 3 out of 5 replicate runs. Data analysis was carried out using DynamX 3.0 (Waters, Milford MA) software to compile and process raw mass spectral data and generate centroid values to calculate relative deuteration values.

Steady state data analysis

-V_max and K_m were determined by fitting three, independent experimental repeats of V_o values (nmol Pi min^-1) at various ATP concentrations, at a fixed HSI#6 concentration, in Michaelis-Menten equation:

$$v=V_{\max }S/\left(K_m+S\right)$$ 1a

V_max = maximum velocity; S = substrate concentration; K_m = Michaelis constant. The Best-Fit values (95% CI) with SEM were identified.

-To calculate K_cat, V_o values (nmol Pi min^-1) were fit to equation:

$$v=K_{cat}{E}_{t}S/\left(K_m+S\right)$$ 1b

E_t = enzyme concentration; S = substrate concentration; K_m = Michaelis constant. The Best-Fit values (95% CI) with SEM were identified.

-To classify the type of compound activation we calculated the α and β parameters of the following Equation³³:

$$v=V_{\max }S/\left\{K_m\left[\left(1+A/K_{\mathrm{A}}\right)/\left(1+\beta A/\alpha K_{\mathrm{A}}\right)\right]+S\left[\left(1+A/\alpha K_{\mathrm{A}}\right)/\left(1+\beta A/\alpha K_{\mathrm{A}}\right)\right]\right\}$$ 2

S = substrate concentration; A = compound concentration; α = the factor that modifies K_m; β = the factor that modifies V_max; K_A = half-max compound concentration. Best-Fit values (95% CI) with SEM were identified.

-For apparent K_d determination K_m values were fit to Equation:

$$K_m=B_{\max }X/\left(K_d+X\right)+K_0$$ 3

B_max = maximum binding; X = compound concentration; K_d = dissociation constant; K₀ = initial binding at zero compound concentration. The Best-Fit values (95% CI) with SEM were identified.

-For AC₅₀ determination the K_cat values for the various HSI#6 concentrations were fit to Eq. 4, with Hill number fixed to 1⁷⁰:

$$K_{cat}={K_{cat}}_{\left(\min \right)}+\left\{\frac{\left[\left({K_{cat}}_{\left(amax\right)}-{K_{cat}}_{\left(\min \right)}\right)\times A^{{\mathrm{n}}_{\mathrm{H}}}\right]}{\left[A{C}_{50}^{{\mathrm{n}}_{\mathrm{H}}}+A^{{\mathrm{n}}_{\mathrm{H}}}\right]}\right\}$$ 4

K_{cat (min)} = ATP turnover at zero compound; K_{cat (amax)} = maximum ATP turnover; AC₅₀ = compound concentration required to increase ATP turnover by 50%; A = compound concentration; n_H = Hill number.

Statistics / Reproducibility

Statistical analysis was performed using GraphPad Prism. Data points in plots represent independent measurements and are shown with ± SD. All derived parameters are reported as best-fit values ± SEM (from the fit), and 95% CIs were computed using the profile likelihood method. For comparisons between two independent groups, unpaired two-tailed Student′s t-tests were applied. p < 0.05 (*), p < 0.01(**), p < 0.001 (***), p < 0.0001 (****), p ≥ 0.05 (ns).

Reporting Summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Purification of proteins and IMVs was done by H.S., H.M.B., and S.r.K.; ATPase assays and MANT-ADP binding experiments by H.S. and H.M.B.; in vivo and in vitro secretion assays by H.S.; HDX-MS experiments by S.r.K.; compound synthesis and quality control by C.P., M.A., P.V. and V.M.; data analysis by H.S., H.M.B., S.r.K., S.K., and A.E. A.E. and S.K. conceived, managed and supervised the project. H.S., H.M.B., S.r.K., S.K., and A.E. wrote the first draft of the paper. All authors read and approved the MS.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Dr Mark Collins and Dr Ophelia Bu. A peer review file is available.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper, its Supplementary Information/Data files. Source data for HDX-MS figures can be obtained from Supplementary Data 1 and HDX-MS raw data have been deposited on Zenodo⁷¹. Source data underlying graphs can be obtained from Supplementary Data 2.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09623-w.