Exploration of Length and Hydrophobicity Requirements for Optimal α‐Amylase Secretion in Bacillus subtilis Based on Rational Design Authors: Dicky A. Pranoto, Danilo Milo, Lorenzo de la Parra Soto, Jolanda Neef, Girbe Buist, Jan Maarten van Dijl Published: 2026 ABSTRACT ABSTRACTThe secretory (Sec) pathway is the major pathway of Bacillus subtilis for protein export from the cytoplasm to the extracellular milieu. An amino‐terminal signal peptide is required to initiate protein translocation across the cytoplasmic membrane via Sec. Generally, Sec pathway signal peptides of B. subtilis are around 20–30 amino acids long and tend to be relatively hydrophobic compared to signal peptides from other organisms. In view of the importance of signal peptides for high‐level secretory protein production, understanding how signal peptide structure and length influence the efficiency of protein secretion is crucial. Accordingly, the present study was aimed at investigating the effects of signal peptide size and, in particular the subdomain sizes of a signal peptide, in relation to protein secretion efficiency. To this end, a rational engineering approach was employed for the design of shortened signal peptides that can still drive effective protein secretion. This involved in silico interaction studies with the signal recognition particle, which is involved in protein targeting from the ribosome to the membrane. Our results show that the hydrophobic H‐domain of a B. subtilis signal peptide can be shortened from seventeen to seven hydrophobic amino acid residues without a detectable decrease in protein secretion. Introduction Bacillus subtilis is a Gram‐positive bacterium commonly found in the soil and plant rhizosphere (Earl et al. 2008). However, this bacterium has also become an important model organism for fundamental and application‐oriented studies due to its well‐characterized genetics and physiology. In addition, B. subtilis has become a superb chassis in industrial biotechnology for the production of numerous products, including various enzymes and chemicals (Westers et al. 2004; van Dijl and Hecker 2013; Cui et al. 2018; Yoshida and van Dijl 2020). Notably, B. subtilis can secrete proteins into its fermentation broth to concentrations of 20 g/L and more, which greatly facilitates their downstream processing (Ploss et al. 2016). The application of B. subtilis is also attractive from regulatory points of view as many of its products are generally regarded as safe (GRAS) by the United States Food and Drug Administration (Westers et al. 2004; Harwood et al. 2018; Yoshida and van Dijl 2020). Despite these advantages, there are also some challenges in the use of B. subtilis , such as the generally encountered difficulty in secreting heterologous proteins in adequate amounts, or the problem of secretion stress which may reduce the yield of proteins of interest (Zweers et al. 2009; Suárez et al. 2019; Öktem et al. 2024). The major protein secretion route in B. subtilis is the general secretory pathway (Sec pathway), which exports proteins from the cytoplasm, across the membrane and cell wall, into the extracellular environment (Harwood and Cranenburgh 2008). Proteins secreted via the Sec pathway possess an N‐terminal signal peptide (SP) that initiates the protein secretion process by interacting with secretion machinery components and the membrane (von Heijne 1990; Zanen et al. 2005). Although Sec‐type SPs show little conservation at the amino acid level, they have a highly conserved tripartite structure, consisting of an N‐terminal N‐domain, a central H‐domain and a C‐terminal C‐domain (von Heijne 1990). The N‐domain includes at least one positively charged amino acid residue, whereas the H‐domain consists predominantly of hydrophobic amino acid residues. The H‐domain also has the propensity to form an α‐helix (Briggs et al. 1986). The C‐domain includes an Ala‐X‐Ala motif, which is recognized by signal peptidases that cleave the SP after the C‐terminal Ala residue during or shortly after membrane translocation (Dalbey and von Heijne 1992; Tjalsma et al. 2004; Nielsen et al. 1997; Dalbey et al. 1997). The cleaved SP will subsequently be degraded by signal peptide peptidases (Ichihara et al. 1986; Bolhuis et al. 1999; Tjalsma et al. 2004; Dalbey et al. 2012; Neef et al. 2017). The overall hydrophobicity of the SP has been associated with SP efficiency, as hydrophobic residues in the SP are recognized by the signal recognition particle protein (SRP) in the cytoplasm and the SecA/SecYEG protein translocation machinery in the membrane (Hartl et al. 1990; Gelis et al. 2007; Freudl 2018; Kaushik et al. 2022). In B. subtilis , the main components of the SRP‐dependent protein export pathway are Ffh and FtsY (Honda et al. 1993; Oguro et al. 1995; Bunai et al. 1996; Zweers et al. 2008). Two lines of evidence support the notion that the SRP pathway is involved in protein secretion in B. subtilis . Firstly, depletion of Ffh or FtsY leads to defects in protein secretion as shown by proteomic analyses (Zanen et al. 2005). Secondly, Zanen et al. (2005) showed that B. subtilis Ffh recognizes and binds the SP from the α‐amylase AmyQ of Bacillus amyloliquefaciens in vitro. Thus, it seems plausible that Ffh of B. subtilis binds to Sec‐type SPs, thereby contributing to the secretion of the respective proteins. Judged by previous studies, which were mostly done on the Ffh of Escherichia coli , Ffh recognizes the hydrophobic core of the H‐domain and transfers the SP together with the ribosome‐nascent chain complex to FtsY (Takamatsu et al. 1997; Zweers et al. 2008; Reinau and Otzen 2009; Neef et al. 2021). The FtsY‐ribosome‐nascent complex would be localized to the cytosolic side of the membrane, where membrane translocation of the secretory precursor protein via the Sec channel is then initiated. Of note, the binding of a SP to Ffh will not only depend on the hydrophobicity of the H‐domain, but also on the length of the SP. Thus, different SPs may present different binding affinities towards the SRP due to differences in hydrophobicity and length, which may consequently influence SP performance and the efficiency of protein secretion. Over the years, many different efforts were focused on identifying optimal SPs to enhance protein secretion. This was necessary, because different proteins of interest (POIs) seem to impose different SP requirements. One approach that was followed to overcome this limitation involves the generation of SP libraries that are fused to the POI, followed by selection of strains showing the highest secretion level for the respective POI (Brockmeier, Caspers, et al. 2006; Brockmeier, Wendorff, and Eggert 2006; Hemmerich et al. 2016; Fu et al. 2018; Heinrich et al. 2019; Grasso et al. 2023). However, this approach requires high‐throughput screening, and the selected SP may not be generally applicable to direct efficient secretion of other POIs. Other strategies focused on the modification of existing SPs to enhance secretion of various POIs (Zanen et al. 2005; Caspers et al. 2010; Zhou et al. 2016). These studies have offered valuable insights on how alteration of the SP, specifically the N‐, H‐ and C‐subdomains, affect the protein secretion efficiency. These insights have laid the groundwork for rational design of synthetic SPs to enhance protein secretion in B. subtilis (Mhiri et al. 2000; Grasso et al. 2023; Jeon et al. 2024). The present study was aimed at investigating the relationship between SP length, hydrophobicity and secretion efficiency in B. subtilis with the overarching objective of determining whether and to what extent B. subtilis SPs can be shortened. The incentive to address these questions is based on the early observation that B. subtilis and other Gram‐positive bacteria have longer and, therefore, more hydrophobic SPs compared to Gram‐negative bacteria such as E. coli (von Heijne and Abrahmsén 1989; Tjalsma et al. 2000; Tjalsma et al. 2004; Neef et al. 2026). On the other hand, lipoproteins of B. subtilis have relatively short SPs, whereas they are considered to follow the same Sec‐dependent export route from the cytoplasm as secreted proteins, which only diverges at the stage of SP cleavage by type II (LspA) and type I signal peptidases (SipSTUVW), respectively (Tjalsma et al. 2000; Tjalsma et al. 2004; Neef et al. 2026). We therefore hypothesized that B. subtilis SPs do not need to be long in order to function properly. To verify this idea, the α‐amylase AmyE of B. subtilis was chosen as a reporter protein. Furthermore, we selected the SP from the B. subtilis lipoprotein MntA (SPMntA) as a prototype for a short SP that is highly effective in driving lipoprotein export (Antelmann et al. 2001; Kouwen et al. 2007), and asked the question whether it can also drive AmyE secretion. In brief, the results show that SPMntA can drive AmyE secretion and this served as a starting point for further SP minimization. Importantly, to give direction to the SP minimization effort, we devised an in silico approach, where the B. subtilis Ffh protein was used as a ‘molecular mould’ to dock mutant SP designs, in order to predict whether they might be able to drive effective AmyE secretion in B. subtilis . Materials and Methods Bacterial Strains and Growth Conditions The bacterial strains used in this study are listed in Table 1. The B. subtilis strains are all based on B. subtilis TS10. This strain carries the Pmtl‐ComKS cassette for mannitol‐induced development of super competence, which has been inserted in the chromosomal yvcA gene (Schilling et al. 2023). All bacterial strains were regularly grown in lysogeny broth (LB; Oxoid Limited, Hampshire, UK) at 37°C with shaking at 250 rpm. Where appropriate, the media to culture E. coli strains were supplemented with ampicillin (100 μg/mL; Sigma‐Aldrich, Massachusetts, USA), and the media to culture B. subtilis strains with kanamycin (20 μg/mL; Sigma‐Aldrich, Massachusetts, USA). E. coli strains carrying the pXTC plasmid (Darmon et al. 2006) and B. subtilis strains with the chromosomally integrated XTC cassette were grown in media supplemented with 10 μg/mL tetracycline (Sigma‐Aldrich, Massachusetts, USA). For pulse‐labelling studies the bacteria were grown in MBU medium (Neef et al. 2017). TABLE 1List of bacterial strains and plasmid used in this study.CharacteristicsReferencesStrains Bacillus subtilis TS10168 trpC + ∆yvcA::Pmtl‐comKS Schilling et al. (2023) B. subtilis TS10 amyE::tet B. subtilis TS10 with a tetracycline resistance marker inserted in the chromosomal amyE gene, tetR This studyPlasmidspBSMulI Escherichia coli ‐ B. subtilis shuttle vector for gene expression in B. subtilis driven by the PHpaII promotor; pUB110 ori for replication in B. subtilis ; ampR (for E. coli ), kanR (for B. subtilis )Brockmeier, Caspers, et al. (2006); Brockmeier, Wendorff, and Eggert (2006)pBSMulI‐SPAmyE‐AmyEpBSMulI‐derived plasmid, encodes the native AmyE SP fused to AmyEThis studypBSMulI‐SPMntA‐AmyEpBSMulI‐derived plasmid, encodes the MntA SP with Ala‐X‐Ala motif fused to AmyEThis studypBSMulI‐SPAmyENdom‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with a mutated N‐domain fused to AmyEThis studypBSMulI‐SPAmyEHdom‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with a mutated H‐domain fused to AmyEThis studypBSMulI‐SPAmyECdom‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with a mutated C‐domain fused to AmyEThis studypBSMulI‐SPAmyENHCdom‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with mutated N‐, H‐, C‐domains fused to AmyEThis studypBSMulI‐SPAmyEHdom2‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with seven leucine residues in the H‐domain fused to AmyEThis studypBSMulI‐SPAmyEHdom3‐AmyEpBSMulI‐derived plasmid, encodes the AmyE SP with five leucine residues in the H‐domain fused to AmyEThis studypBSMulI‐SPAmyE7‐AmyEpBSMulI‐derived plasmid, encodes a shortened AmyE SP with 7 hydrophobic amino acid residues in the H‐domain and lacking 3 hydrophobic amino acid residues from the N‐domain fused to AmyEThis studypBSMulI‐SPAmyE8‐AmyEpBSMulI‐derived plasmid, encodes a shortened AmyE SP with 10 hydrophobic amino acid residues in the H‐domain and lacking 3 hydrophobic amino acid residues from the N‐domain fused to AmyEThis study List of bacterial strains and plasmid used in this study. SP Engineering and Plasmid Construction The prediction of SP domains and translocation pathways was conducted with SignalP 6.0 (https://services.healthtech.dtu.dk/services/SignalP‐6.0/; Teufel et al. 2022). Protein grand average of hydropathy (GRAVY) scores were calculated using the Sequence Manipulation Suite Protein GRAVY webtool (https://www.bioinformatics.org/sms2/protein_gravy.html; Kyte and Doolittle 1982; Stothard 2000). Sequences of all primers used for PCR and subsequent cloning are listed in Supplementary Table S1. The sequences of all SPs and the amyE gene were amplified using the chromosomal amyE gene of B. subtilis strain 168 as template, unless specified otherwise. Engineered SPs were PCR‐amplified using the high‐fidelity Phusion polymerase (New England Biolabs, Massachusetts, USA). To generate short engineered SPs, complementary long primer pairs were used. Such long primer pairs were mixed with T4 DNA ligase buffer (New England Biolabs, Massachusetts, USA), heated at 90°C for 15 min and allowed to anneal by slowly decreasing to room temperature. For plasmid construction, some of the primers contained 18 bp overhanging sequences. All plasmids used and generated in this study are listed in Table 1. For cloning, the plasmid backbone was linearized and amplified by PCR. Subsequently, the NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs, Massachusetts, USA) was used to assemble the linear plasmid backbone with engineered SP sequences and the amyE gene lacking the coding sequence of the original SP. Transformation of Strains E. coli strain TG1 was used to select plasmids encoding the engineered SP‐AmyE fusions upon transformation by heat‐shock (Mandel and Higa 1970; Froger and Hall 2007). All SP‐AmyE fusions were verified by PCR amplification from the different plasmids and subsequent sequencing (Eurofins, Konstanz, Germany). Plasmids encoding the correct SP‐AmyE fusions were isolated using the innuPREP Plasmid Mini Kit 2.0 (Analytik Jena, Jena, Germany). These plasmids were subsequently introduced into B. subtilis strain TS10 using mannitol‐induced competent cells (Rahmer et al. 2015). Western Blotting and Quantification of AmyE Protein Production Bacterial cultures were grown overnight, normalized to the same optical density at 600 nm (OD600) and centrifuged to separate the cells and growth media. Cells were resuspended in lithium dodecyl sulphate (LDS) gel loading buffer (Invitrogen, Massachusetts, USA) containing dithiothreitol (DTT; Invitrogen, Massachusetts, USA). The cells were then lysed by bead‐beating (Bertin Technologies, Montigny‐le‐Bretonneux, France). Proteins in the growth media were precipitated using 10% (w/v) trichloroacetic acid. Precipitated proteins were washed with ice‐cold acetone before resuspension in LDS gel loading buffer with DTT. Subsequently, the cellular and extracellular protein samples were heated at 95°C for 10 min. The protein samples were then separated by LDS polyacrylamide gel electrophoresis (PAGE) on 10% Bis‐Tris NuPage gels (Invitrogen, Massachusetts, USA). Lastly, the separated proteins were transferred to Amersham Protan 0.45 μm nitrocellulose membranes (GE Healthcare, Chicago, USA) using a Power Blotter XL System (Invitrogen, Massachusetts, USA). The membranes were blocked for 1 h with 5% (w/v) skim milk resuspended in phosphate‐buffered saline (PBS) and washed twice with PBS containing Tween‐20 (PBS‐T). The membranes were then incubated for 1 h with polyclonal rabbit antibodies against AmyE at a 1:5000 dilution. After incubation, membranes were washed three times with PBS‐T and incubated for 45 min with a 1:10,000 diluted secondary goat anti‐rabbit IgG that had been labelled with IRDye800CW (Li‐Cor Inc., Nevada, USA). The membranes were then washed three times with PBS‐T and twice with PBS. Membrane visualization was performed using an Amersham Typhoon Biomolecular Imager (Cytiva, Massachusetts, USA). Quantification of the AmyE protein was conducted by Western blotting using the pure AmyE protein (MP Biomedicals, California, USA) to prepare a calibration curve. Different dilutions of cell extracts and growth medium samples were separated by reducing LDS‐PAGE on the same gel as different dilutions of the pure AmyE. The separated proteins were then transferred to a nitrocellulose membrane for immunodetection of AmyE as described above. The signal intensity of AmyE‐specific protein bands was measured using Fiji software (Schindelin et al. 2012), and the signal intensities for AmyE in cellular or growth medium samples were correlated to the signal intensities of known AmyE amounts in the samples for the standard curve. Starch Plate Assay for α‐Amylase Activity Cultures were grown overnight and diluted to reach an OD600 of 0.5. Subsequently, 1.5 μL of each diluted culture was placed on LB agar supplemented with 1% soluble starch. The plates were then incubated for at least 16 h at 37°C. After colonies had formed, the starch in the plates was stained with the vapour from iodine granules to obtain a blue‐coloured iodine‐starch complex. α‐Amylase activity was detectable by the formation of an uncolored halo around the respective colonies. α‐Amylase Enzyme Activity Assay Measurement of AmyE activity in culture supernatants was conducted using the α‐Amylase Assay Kit (Ceralpha method) from Megazyme (Megazyme, Bray, Ireland). B. subtilis strains were grown overnight and normalized to the same OD600. Then the cultures were centrifuged to separate the cells and culture supernatants. Ten microliter of supernatant was mixed with 490 μL of Buffer B containing 0.1 M maleic acid, 0.1 M NaCl, 2 mM CaCl2.2H2O and 0.01% NaN3. The mixture was specified as ‘original extract’. The original extract was diluted, when necessary, before being mixed with substrate and then incubated at 40°C for 10 min. A stopping reagent supplied by the manufacturer was used to stop the reaction. Lastly, the optical density at 400 nm (OD400) was measured using a Cary 60 UV–Vis spectrophotometer (Agilent Technologies, California, USA). The AmyE activity was calculated according to the kit manufacturer's directions. All measurements were conducted in triplicate. Ffh Structure and SP‐Ffh Docking Analyses Interactions between the various engineered SPs and Ffh were predicted using the HADDOCK v2.4 webserver (https://rascar.science.uu.nl/haddock2.4/) for protein docking (Honorato et al. 2021; Honorato et al. 2024). To this end, the protein structures of Ffh and the investigated SPs were first predicted using AlphaFold3 (https://alphafoldserver.com/; Evans et al. 2022; Mirdita et al. 2022; Abramson et al. 2024). The respective predicted template modelling (pTM) scores are presented in Table S2. Generally, a pTM score higher than 0.5 indicates that the predicted structure closely resembles the original structure (Xu and Zhang 2010). However, it is important to note here that the structures of only a limited number of SPs have been determined experimentally. This may lead to an underestimation of the pTM scores that are generated when predicting the structures of different SPs. To predict the structure of B. subtilis Ffh, the respective amino acid sequence was retrieved from the SubtiWiki repository (February 2025; Pedreira et al. 2022), while the sequence of E. coli Ffh was retrieved from UniProt (March 2025; UniProt Id: P0AGD7) and the sequence of the 4.5S RNA was retrieved from the NCBI repository (ID:8302906; February 2025). Comparison of the B. subtilis and E. coli Ffh proteins was conducted by overlaying the respective predicted protein structures with UCSF ChimeraX downloaded from (https://www.cgl.ucsf.edu/chimerax/; Pettersen et al. 2021). This verified the high structural similarity between the B. subtilis and E. coli Ffh proteins and, therefore, further analyses were based on the predicted structure of B. subtilis Ffh. For docking of SPs to Ffh using the HADDOCK webserver, the amino acid residues 300–446 of Ffh and the H‐domains of the SPs were selected as the ‘active site’. The remaining regions of Ffh and the SPs were automatically selected as the ‘passive site’. Subsequently, the predicted model of B. subtilis Ffh with a bound SP was compared with the crystal structure of the B. subtilis Ffh protein as deposited in the PDB database (PDB Id: 4UE4 and 7O9F; Figure S1). Additionally, the interaction of Ffh and the 4.5S RNA with the docked SP was predicted using the HADDOCK webserver and AlphaFold3 for comparison and to predict whether the 4.5S RNA interacts with SP. Since no interaction was predicted (Figure S1), and since there is no literature regarding possible interaction between Ffh‐bound SPs and the 4.5S RNA, the structure of 4.5S RNA was not used in the final Ffh‐SP docking and visualization. Of note, for docking in HADDOCK, the amino acid sequence of the SP was used as is, with the addition of amino acids from the mature AmyE protein to normalize the length for shortened SPs, such that always a stretch of 33 amino acids in total was used. The best model from HADDOCK was chosen by evaluating parameters, such as protein conformation, SP orientation, Z‐score and HADDOCK score. Since the N‐terminus of a SP is first to emerge from the ribosome, we selected the HADDOCK model for SP binding to Ffh with respect to the SP‐Ffh orientation based on the highest probability score, and on structural information derived from the B. subtilis SRP with the ribosome‐nascent chain sensor for membrane protein biogenesis MifM bound to Ffh (PDB Id: 7O5B; Czech et al. 2022). The predicted contacts between Ffh and a SP were analysed by counting predicted contacts with a cut‐off of ≥ 0 Å. Lastly, the predicted atomic interaction and complex binding stability were predicted using the PRODIGY webtool (https://rascar.science.uu.nl/prodigy/; Vangone and Bonvin 2015; Xue et al. 2016; Honorato et al. 2021). Predicted protein complexes were visualized and analysed using UCSF ChimeraX downloaded from (https://www.cgl.ucsf.edu/chimerax/; Pettersen et al. 2021). Reverse Transcriptase‐Quantitative PCR (RT‐qPCR) of amyE mRNA Bacterial cultures were grown in LB medium as described above. The bacterial cells were collected by centrifugation, resuspended in 20 mM Tris–HCl (pH 7) with 25 mg/mL lysozyme (Sigma Aldrich, Missouri, USA), and incubated for 10 min at 37°C with shaking. Total RNA was then extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The isolated RNA was treated with RNase‐free DNase (Qiagen, Hilden, Germany) for 20 min at 30°C and purified once more with the RNeasy Mini Kit reagents. Equal amounts of purified RNA were then mixed with the TaqMan reverse transcriptase master mix (Invitrogen, Massachusetts, USA) to convert the RNA into cDNA. Subsequently, equal amounts of cDNA from the different samples were used for quantitative PCR with a qTOWER3 PCR machine (Analytik Jena, Jena, Germany) using the ROTILux SYBRGreen qPCR master mix (Carl Roth, Karlsruhe, Germany) and amyE‐specific primers. For reference, qPCR was also carried out with specific primers against the gyrA gene, specifying DNA gyrase. Primers used for qPCR are listed in Table S1. Specificity of the amplified product was assessed through a melting curve analysis (Figure S2). Pulse‐Labelling of AmyE Protein Pulse‐labelling of the B. subtilis AmyE protein was performed using the EasyTag EXPRESS35S Protein Labeling Mix (35S‐L‐methionine and 35S‐L‐cysteine mix; Revvity, Massachusetts, USA) as previously described (Neef et al. 2017). In brief, B. subtilis cells expressing AmyE fused to different SPs, expressing AmyE without a SP, or control bacteria without a plasmid encoding amyE were cultured for ~18 h in MBU medium (Neef et al. 2017) at 37°C with shaking at 250 rpm. The cultures were then diluted to an OD600 of 0.7 in fresh MBU medium and growth was continued for 1 h. Subsequently, protein labelling was performed by incubating the bacteria with 25 μCi [35S]‐methionine/cysteine mix for 2 min at 37°C with shaking. Samples of 600 μL were withdrawn and kept on ice for ~20 min. Cells and growth medium were then separated by centrifugation and incubated with 10% (w/v) TCA for 1 h on ice. The TCA‐precipitated cells and extracellular proteins were washed with acetone three times before resuspension in lysis buffer (10 mM Tris pH 8, 25 mM MgCl2, 200 mM NaCl and 5 mg/mL lysozyme). The resuspended cells were incubated for 15 min at 37°C. Complete cell lysis was then achieved by adding 1% (w/v) SDS and heating for 10 min at 100°C. Immunoprecipitation of [35S]‐labelled AmyE was performed in 1× STD‐Tris buffer (10 mM Tris pH 8.2, 0.9% [w/v] NaCl, 1% [v/v] Triton X‐100 and 0.5% [w/v] sodium deoxycholate) using polyclonal rabbit antibodies against AmyE and Protein A affinity medium (Mabselect Sule, Cytiva, Massachusetts, USA). The immunoprecipitated proteins were separated by LDS‐PAGE as described above and radio‐labelled proteins were visualized by phosphor imaging with an Amersham Typhoon Biomolecular Imager. Statistical Analyses The data obtained was analysed using GraphPad Prism 8.0. Statistical differences were assessed using one‐way ANOVA tests with a p‐value of < 0.05 indicating significance, followed by Sidak's multiple‐comparison test to compare each sample to the positive control (native SPAmyE) and negative control (No SP). Results A Modified Lipoprotein SP Can Drive AmyE Secretion As a first approach to investigate whether a substantially shortened SP can drive AmyE secretion, we explored whether a modified lipoprotein SP can fulfil this function. To this end, the native SPAmyE was substituted with the SP of the lipoprotein MntA (SPMntA), but without the cleavage site for the lipoprotein‐specific signal peptidase. Instead, the Ala‐Ser‐Ala type I signal peptidase cleavage site of AmyE was retained (Figure 1). As predicted with SignalP 6.0, AmyE provided with the SPMntA can still be secreted via the Sec pathway (Figure S3A,B). Indeed, as shown by Western blotting, the B. subtilis TS10 strain expressing SPMntA‐AmyE did secrete AmyE, but in much lower amounts than when AmyE secretion was driven by the native SPAmyE (Figure 2A). Western blotting also revealed that the levels of the quality control protease HtrA in the TS10 strain did not detectably differ, regardless of the SP that was used to drive AmyE secretion (Figure 2B). This indicated that expression of SPMntA‐AmyE did not elicit more secretion stress than expression of AmyE with its native SP, since htrA gene expression is controlled by the secretion stress‐responsive CssRS two‐component regulatory system (Hyyryläinen et al. 2001; Darmon et al. 2002). Elevated HtrA levels in the extracellular milieu would have reflected an elevated secretion stress response, which is generally observed when unfolded or misfolded proteins accumulate at the membrane‐cell wall interface (Westers et al. 2006). Furthermore, SPMntA facilitated the secretion of active AmyE as shown by the formation of halos of starch degradation around colonies on plates supplemented with 1% starch (Figure 2C) and by enzyme activity assays on culture supernatant fractions (Figure 2D). However, while SPMntA facilitated the secretion of AmyE to levels that were significantly higher than background, it directed AmyE secretion to much lower levels than the native SPAmyE, as was reflected by the halo sizes around colonies of B. subtilis TS10 expressing SPMntA‐AmyE and the enzyme activity levels in the respective culture supernatant. Wild‐type and engineered SPs for AmyE secretion. The image presents the amino acid sequences, numbers of amino acid residues (in parentheses) and the GRAVY hydrophobicity scores of the native SPAmyE and engineered SPs that were used to drive AmyE secretion. The amino acid sequences are colour‐coded based on the subdomains predicted with SignalP 6.0. For comparison, the amino acid sequence of the SP of the pectate lyase C (SPPel) is included, which represents the shortest predicted SP of B. subtilis (Tjalsma et al. 2000; Lenz et al. 2023). Underlined amino acids refer to the AXA signal peptidase recognition sites of the native SPAmyE or the native SPPel. Green, N‐domain; red, C‐domain; yellow, H‐domain. Comparison of AmyE levels in the culture supernatant of B. subtilis TS10 expressing AmyE with the native SPAmyE or SPMntA. Western blotting analysis of AmyE (A) and HtrA levels (B) in the bacterial cells and respective culture supernatant fractions of the different strains. Please note that the ‘No SP’ control strain, which expresses AmyE without a SP, also produced some AmyE due to expression of the chromosomally encoded native amyE gene. HtrA* indicates a dominant degradation product of HtrA that relates to cleavage by the AprE and AprX proteases (Öktem et al. 2024). The expected position of the full‐size HtrA is indicated with an arrow. (C) Starch plate amylase activity assay for B. subtilis TS10 producing AmyE with the native SPAmyE or SPMntA, or without a SP. Please note that the ‘No SP’ control strain also produced a halo of starch degradation due to expression of the chromosomally encoded native amyE gene. (D) Amylase activity in culture supernatants of B. subtilis TS10 expressing AmyE with SPAmyE or SPMntA, or without a SP. Each assay was conducted in three biological replicates and the data were presented as the mean ± the standard error of the mean (SEM). *: p value of ≤ 0.05; ****: p value of ≤ 0.001. Mutation of the N‐ or C‐Domains of SPAmyE Is More Impactful on AmyE Secretion Than Mutation of the H‐Domain To understand how the substitution of the native SPAmyE by SPMntA led to reduced AmyE levels in the culture supernatant, each of the three SPAmyE domains as indicated in Figure 1 was mutated. In this case, each domain mutation in SPAmyE served to generate a SP domain that was similar in length as the respective domain in the SPMntA, but shortened compared to SPAmyE. To achieve this, the domain length as well as the amino acid composition and characteristics were taken into account. Specifically, the N‐domain of SPAmyE was replaced by the two first residues of SPMntA (i.e., MR), the N‐terminal part of the H‐domain of SPAmyE was shortened to a stretch of five Leu residues since Leu is the dominant residue in the H‐domain of SPAmyE, and the C‐domain was reduced to the signal peptidase recognition site plus one Ala residue as in the SPMntA‐AmyE fusion (Figure 1). In addition, a mutant SP (SPAmyENHCdom) was generated that combined the three N‐, H‐ and C‐domain mutations (Figure 1). As shown by Western blotting, the AmyE levels in the culture supernatant of B. subtilis TS10 expressing the different SP‐AmyE fusions depended significantly on the mutated SP domains (Figure 3A). The SPAmyE with the mutated N‐domain directed lower AmyE secretion than the SPAmyE with mutated H‐ or C‐domains. Furthermore, the mutant SP that combined the three domain mutations (SPAmyENHCdom) led to the lowest AmyE levels in the culture supernatant. Interestingly, the HtrA levels in culture supernatant fractions of the strains expressing AmyE with mutated SPs did not show major differences and, importantly, they did not exceed the HtrA levels observed for the strain expressing AmyE with the native SPAmyE (Figure 3B). This shows that none of the tested SP‐AmyE fusion constructs elicited a CssRS‐dependent protein secretion stress response. However, expression of AmyE with SPs carrying mutations in the H‐domain (especially SPAmyENHCdom), or with SPMntA led to elevated extracellular levels of the cytoplasmic marker protein thioredoxin A (TrxA) than expression of AmyE with the native SPAmyE or without a SP (Figure S4A). The presence of TrxA in the culture supernatant fraction is indicative of cell envelope perturbation in cells of B. subtilis . This implies that SP engineering may elicit the lysis of some bacteria in the respective cultures. AmyE production and secretion by B. subtilis TS10 strains expressing AmyE with SPAmyE variants carrying mutations in the N‐, H‐ and/or C‐domains. Western blotting was performed to visualize the AmyE (A), or HtrA (B) levels in the bacterial cells and culture supernatants. Please note that the ‘No SP’ control strain, which expresses AmyE without a SP, secreted some AmyE due to expression of the chromosomally encoded native amyE gene. HtrA indicates the expected position of the full size HtrA protein while HtrA* indicates a dominant degradation product of HtrA. (C) Starch plate assay for B. subtilis TS10 expressing AmyE with different SP domain mutations. Please note that the ‘No SP’ control strain also produced a halo of starch degradation due to expression of the chromosomally encoded native amyE gene. (D) Amylase activity assays for strains expressing AmyE with different SP domain mutations. Each assay was conducted in three biological replicates, and the data was presented as the mean ± the standard error of the mean (SEM). ****, p value of ≤ 0.001; ns, no significance. The results obtained by Western blotting were verified by examining halos of starch degradation on plates containing 1% starch. Indeed, a similar pattern of starch degradation was observed in the plate assay, showing that the bacteria expressing AmyE with the SPAmyENHCdom yielded the smallest halo (Figure 3C). Amylase activity assays on culture supernatant fractions also showed a similar pattern, albeit that the strain with a C‐domain mutation in the SPAmyE presented significantly lower AmyE activity levels compared to the strain producing AmyE with the native SPAmyE (Figure 3D). Together, these results show that especially mutation of the N‐ or C‐domains of SPAmyE led to significantly decreased secretion of active AmyE. Of note, the combination of the three domain mutations in SPAmyENHCdom led to comparable levels of active AmyE secretion as were directed by SPMntA. Furthermore, the data suggested that shortening of the SP in itself was not the only cause of decreased AmyE secretion but, rather, that particular subdomain properties are critical for efficient AmyE secretion. Structural Model for Interaction of the B. subtilis Ffh Protein With SPAmyE Since shortening of the H‐domain of SPAmyE had relatively little negative impact on active AmyE secretion, if any, we raised the question whether reduction of the H‐domain might influence the possible interaction of SPAmyE with the Ffh protein. To evaluate this, a molecular docking method was employed. To this end, we first predicted the structure of B. subtilis Ffh with AlphaFold3 and compared it with the structure of the E. coli Ffh. This showed that both Ffh proteins are structurally highly similar (Figure 4A). Importantly, the M‐domains of both Ffh proteins have a comparable structure (Figure 4B and Figure S1A,D,E). The predicted structure of B. subtilis Ffh also matched with crystal structures of the separate M‐ and N‐domains of B. subtilis Ffh (Figure S1A,B). Furthermore, B. subtilis Ffh requires the 4.5S RNA equivalent for stability and function (Nakamura et al. 1992). Therefore, we performed a molecular docking analysis, which indicated that the 4.5S RNA binds to the B. subtilis Ffh protein at the site that was expected based on comparison with a structure of the B. subtilis SRP molecule including Ffh and the 4.5S RNA that was determined by cryo‐electron microscopy (cryo‐EM; Figure 4C and Figure S1C). Indeed, this specific interaction was predicted to promote Ffh stability and to support the C‐terminal M‐domain of Ffh in forming a hydrophobic pocket for SP binding (Figure 4D). In particular, this hydrophobic pocket has the potential to interact with the H‐domain of SPs, as was verified by performing a docking analysis for E. coli Ffh and the SP of E. coli DsbA (SPDsbA; Figure S1E), which is Ffh‐dependently exported to the periplasm (Schierle et al. 2003; Shimohata et al. 2005). Next, using SPAmyE as a SP model, the H‐domain of SPAmyE was shown to interact with the hydrophobic pocket of Ffh (Figure 4D,E). Interestingly, residues of the N‐ and C‐domains of SPAmyE were also predicted to make contact with Ffh, albeit not in the hydrophobic pocket. To test whether additional amino acids from the mature AmyE protein would alter the predicted binding of a SP to the B. subtilis Ffh protein, an additional stretch of 20 amino acids from the mature AmyE protein was added to the SPAmyE. The predicted model shows that the Ffh protein still bound in the same way to the H‐domain of SPAmyE irrespective of the 20 additional amino acids (Figure S5A,B). Altogether, the obtained SP‐Ffh interaction model suggests that not only the H‐domain of a SP can interact with Ffh, but that also residues from the other SP sub‐domains can influence the SP‐Ffh interaction. Predicted structure of Bacillus subtilis Ffh. The structure of the B. subtilis Ffh protein was predicted with AlphaFold3. Overlays are shown for the predicted structures of the B. subtilis and E. coli Ffh as full proteins (A) or focused on the M‐domains only (B). Blue: N‐domain of B. subtilis Ffh; purple: N‐domain of E. coli Ffh; orange: M‐domain of B. subtilis Ffh; red: M‐domain of E. coli Ffh. (C) Cryo‐EM structure of the B. subtilis SRP complex (grey) zoomed‐in on the Ffh protein (orange) with the bound MifM‐polypeptide (blue), and the 4.5S RNA (purple). (D) Complex of Ffh‐SPAmyE with hydrophobicity colouring based on the molecular hydrophobicity potential, where a higher value of the molecular hydrophobicity potential indicates a relatively more hydrophobic surface. (E) Docking of SPAmyE and Ffh as predicted with HADDOCK. Predicted contacts between amino acids in SPAmyE and Ffh are visualized with thin dashed purple lines; blue: N‐domain of B. subtilis Ffh; orange: M‐domain of B. subtilis Ffh; green: SPAmyE N‐domain; yellow: SPAmyE H‐domain; red: SPAmyE C‐domain. Minimum Number of Hydrophobic Amino Acids in the H‐Domain of SPAmyE for Efficient AmyE Secretion Since SPAmyE was predicted to interact with the Ffh‐RNA complex mostly with its H‐domain, and since SRP is known to interact preferentially with extended hydrophobic transmembrane protein domains (Schaffitzel et al. 2006; Janda et al. 2010), it was conceivable that a shortened H‐domain might interfere with the SP‐Ffh interaction. Therefore, SPAmyEHdom was also evaluated for its possible interaction with Ffh. Molecular docking prediction showed that SPAmyEHdom can bind to the Ffh protein (Figure 5A). Residues of the N‐ and H‐domains of SPAmyEHdom were shown to interact with Ffh, possibly reinforcing the complex binding affinity. Interestingly, the predicted ∆G and Kd values for SPAmyEHdom binding to Ffh were lower compared to the values predicted for native SPAmyE binding to Ffh (Table 2), indicating a higher binding affinity and increased complex stability, respectively. To further evaluate the possible effects of shortened H‐domains on the interaction with Ffh, two additional SPAmyE variants with even shorter H‐domains, namely SPAmyEHdom2 and SPAmyEHdom3, were constructed (Figure 1). In these variants, the H‐domain was shortened to seven or five Leu residues, respectively. The predicted SPAmyEHdom2 interaction with Ffh was similar to that of SPAmyEHdom, as residues in the N‐ and H‐domains of SPAmyEHdom2 were predicted to form close contacts with the hydrophobic pocket of Ffh (Figure 5B). Although the predicted ∆G value was similar to that of SPAmyEHdom, a higher Kd value was predicted for the SPAmyEHdom2‐Ffh complex (Table 2). Meanwhile, SPAmyEHdom3 was also predicted to form a complex with Ffh, albeit that most of the contacts involved amino acids of the N‐domain rather than the H‐domain (Figure 5C). As expected, the predicted Kd value for the SPAmyEHdom3‐Ffh interaction was increased compared to the Kd values predicted for the interactions of SPAmyEHdom and SPAmyEHdom2 with Ffh (Table 2). However, the predicted Kd value of SPAmyEHdom3‐Ffh interaction was still lower than the one predicted for the SPAmyE‐Ffh interaction (Table 2). These predictions suggest that the three SPs with shortened H‐domains may form more stable interactions with Ffh than the native SPAmyE. Models of the interactions between Ffh and SPAmyE variants with shortened H‐domains as predicted with HADDOCK. (A) Model of SPAmyEHdom‐Ffh, (B) SPAmyEHdom2‐Ffh, or (C) SPAmyEHdom3‐Ffh. The side chains of SP amino acids forming contact with Ffh are visualized. Dashed thin purple lines indicate predicted contacts: blue, N‐domain of Ffh; green, SPAmyE N‐domain; grey, amino acid residues of AmyE; orange, M‐domain of Ffh; red, SPAmyE C‐domain; yellow, SPAmyE H‐domain. TABLE 2Predicted ∆G and Kd values of predicted SP‐Ffh complexes.SPAmyE SPAmyEHdomSPAmyEHdom2SPAmyEHdom3SPAmyE7SPAmyE8SPPel ∆G (kCal/Mol) a −9.3−10.9−10.2−9.9−9.7−11.1−10Kd (nM) a 30019641001501594 a All ∆G and Kd values were predicted using the PRODIGY webtool. Predicted ∆G and Kd values of predicted SP‐Ffh complexes. All ∆G and Kd values were predicted using the PRODIGY webtool. To evaluate whether H‐domain shortening in SPAmyE would affect the overall AmyE secretion, quantitative Western blotting and amylase activity assays were conducted (Figure 6A–C). Beforehand, the chromosomal amyE gene was removed from the B. subtilis TS10 strain by replacement with a tetracycline resistance marker to eliminate background AmyE secretion. The results show that SPAmyEHdom2 drives efficient secretion of active AmyE, comparable to SPAmyEHdom and the native SPAmyE (Figure 6B). Nonetheless, the levels of secreted AmyE protein as obtained with SPAmyEHdom2 or SPAmyEHdom were somewhat higher compared to SPAmyE (Figure 6A,C), indicating a somewhat lowered specific activity of AmyE. These findings suggest that SPAmyEHdom and SPAmyEHdom2 may be more efficient in directing the secretion of the AmyE protein than its native SP, but that some of the secreted AmyE does not fold properly into its active conformation. In turn, this suggests that protein folding catalysts that are required for proper AmyE folding, such as the peptidyl‐prolyl cis/trans isomerase and chaperone PrsA (Hyyryläinen et al. 2001), may become saturated once the rate of AmyE secretion is substantially increased. In contrast, the strain expressing SPAmyEHdom3 fused to AmyE presented strongly reduced secretion of AmyE into the culture supernatant (Figure 6A–C) and a substantial fraction of the secreted AmyE had a higher mobility on LDS‐PAGE (Figure 6A), which is indicative of aberrant AmyE cleavage and degradation. Interestingly, the respective bacterial cells also contained lower AmyE levels and the AmyE in the cells had a slightly lower mobility on LDS‐PAGE, suggesting that proteolytic removal of the SPAmyEHdom3 by signal peptidase was affected. Importantly, these findings combined with the molecular docking results suggest that the inability of SPAmyEHdom3 to direct efficient secretion of AmyE could relate to steps in the secretion process that follow after the binding of this SP to Ffh, such as correct engagement with the Sec machinery or cleavage by signal peptidase. Furthermore, the evaluation of secretion stress, as judged by the HtrA levels, showed that bacteria secreting AmyE with the aid of SPAmyEHdom2 presented a relatively lower level of the dominant degradation product of HtrA (Figure 6D). This implies that SP shortening has the potential to reduce secretion stress sensed by the CssRS system, possibly by promoting more efficient AmyE translocation and subsequent folding. On the other hand, TrxA protein levels in supernatant fractions of strains expressing AmyE fused to SPAmyEHdom, SPAmyEHdom2 or SPAmyEHdom3 were relatively high compared to the strain expressing AmyE with the native SPAmyE (Figure S4B). This suggests that SPs with engineered H‐domains may elicit more cell envelope perturbation than the native SPAmyE. Effects of H‐domain shortening in SPAmyE on AmyE secretion by B. subtilis TS10 amyE::Tet. (A) Western blots of AmyE in the bacterial cells and the respective culture supernatants. Particular samples were diluted between 3‐ and 150‐fold as indicated in the legend above the blot to allow quantification of the respective AmyE amounts. Please note that, for unknown reasons, the levels of AmyE secretion driven by SPAmyEHdom were generally higher in B. subtilis TS10 amyE::Tet than in B. subtilis TS10 (see Figure 2). ‘No SP’ marks samples of bacteria expressing AmyE without a SP. (B) α‐Amylase activity of secreted AmyE. (C) AmyE protein concentrations in the cells and supernatant fractions. (D) Western blot of HtrA in the culture supernatant. HtrA indicates the expected position of the full size HtrA protein, while HtrA* indicates a dominant degradation product of HtrA. Each assay was conducted in three biological replicates and the data are presented as the mean ± the standard error of the mean (SEM). ***, p value of ≤ 0.005; **, p value of ≤ 0.001; *, p value of ≤ 0.05; ns, no significance; SN, supernatant. The models presented in Figures 4 and 5 indicated that amino acid residues in the N‐domain of SPAmyE could also form contacts with Ffh. Indeed, in the native SPAmyE three hydrophobic amino acids were identified, namely Phe2, Ala3 and Phe6, which may also form contacts with Ffh and which could help to stabilize the complex. It was thus hypothesized that hydrophobic residues in the N‐domain may compensate for the potentially detrimental effects of a shortened H‐domain to maintain sufficient functional length and hydrophobicity of the SP. To evaluate this idea, two additional shortened SPAmyE variants were constructed, namely SPAmyE7 and SPAmyE8 (Figure 1). SPAmyE7 includes seven hydrophobic amino acids in the H‐domain and lost three hydrophobic amino acids from the N‐domain. SPAmyE8 was constructed by ‘shifting’ the hydrophobic amino acids from the N‐domain to the H‐domain. Accordingly, both SPs only included one Met, two Lys and one Arg residues in their N‐domains. Additionally, the hydrophobic amino acids in both SPs were mixed, rather than introducing seven Leu residues as in the SPAmyEHdom2. Our protein docking approach showed that SPAmyE7 and SPAmyE8 have the potential to bind to Ffh (Figure 7A,B). At a glance, SPAmyE8 presented more contacts to Ffh compared to SPAmyE7, which may be relevant for complex stability. Prediction of the binding affinities showed that the predicted ∆Gs for SPAmyE7 and SPAmyE8 were in the same range as the ∆G predicted for SPAmyEHdom2, although SPAmyE7 had a higher predicted ∆G for Ffh binding compared to SPAmyE8 (Table 2). However, the predicted Kd value for the interaction of SPAmyE7 with Ffh was substantially higher than the predicted Kd values for SPAmyEHdom2 or SPAmyE8 (Table 2), which would suggest the formation of less stable interaction between SPAmyE7 and Ffh. On the other hand, the predicted Kd value of SPAmyE8 binding to Ffh was lower compared to that of SPAmyEHdom2 or any of the other investigated SPs, suggesting a very high affinity of the mutated SPAmyE8 for Ffh. These findings suggest that hydrophobic residues in the N‐domain of a SP may impact the SP‐Ffh complex stability in B. subtilis . Evaluation of AmyE secretion efficiency with the aid of SPAmyE7 or SPAmyE8 in B. subtilis TS10 amyE::Tet. Models of SPAmyE7‐Ffh (A) and SPAmyE8‐Ffh (B) complexes with contacts as predicted with HADDOCK. (C) Western blotting analysis of AmyE in the bacterial cells and respective culture supernatants. Particular samples were diluted between 30‐ and 150‐fold as indicated in the legend to allow quantification of the respective AmyE amounts in this particular blot. ‘No SP’ marks samples of bacteria expressing AmyE without a SP. (D) Amylase activity of secreted AmyE. (E) AmyE protein concentration in the cells and culture supernatant fractions as determined by quantitative Western blotting. Each assay was conducted in three biological replicates and the data is presented as the mean ± standard error of the mean (SEM). ****: p value of ≤ 0.0001; **: p value of ≤ 0.001; *: p value of ≤ 0.05; ns: No significance. (F) Western blotting analysis of HtrA in the culture supernatant fractions. HtrA indicates the expected position of the full size HtrA protein, while HtrA* indicates a dominant degradation product of HtrA. (G) Model of SPPel binding to B. subtilis Ffh as predicted with HADDOCK. In panels A, B and G the thin dashed purple lines indicate predicted contacts. Side chains of SP amino acids forming contacts with Ffh are visualized. Blue, N‐domain of Ffh; green, SPAmyE or SPPel N‐domain; grey, amino acid residues of AmyE or Pel; orange, M‐domain of Ffh; red, SPAmyE or SPPel C‐domain; yellow, SPAmyE or SPPel H‐domain. Western blotting results showed that strains expressing the SPAmyE7‐AmyE fusion did not secrete intact active AmyE (Figure 7C–E). Nonetheless, a degradation product of AmyE was detectable in the culture supernatant (Figure 7C). Furthermore, AmyE was barely detectable in the cellular fraction of this strain. Meanwhile, intact and active AmyE was detected in the supernatant of the strain expressing AmyE fused to SPAmyE8 (Figure 7C–E). The amount of AmyE in the supernatant of this strain was comparable to that secreted by strains expressing AmyE fusions to SPAmyEHdom2 or the native SPAmyE (Figure 7C). Furthermore, amylase activity assays showed that strains in which AmyE secretion was facilitated by SPAmyE8 presented high AmyE activity, comparable to strains secreting AmyE with the aid of the native AmyE (Figure 7D). Consistent with the Western blotting results, the culture supernatant of the strain expressing AmyE fused to SPAmyE7 did not present amylase activity (Figure 7D). Interestingly, the GRAVY value of SPAmyE7 was similar to that of the native SPAmyE (Figure 1). This suggests that the shortening of this SP rather than its hydrophobicity was associated with the inability of a strain to secrete AmyE. Interestingly, the dominant degradation product of HtrA was observed at comparable levels in the culture supernatants of bacteria expressing AmyE fused to SPAmyEHdom2, SPAmyE7, SPAmyE8 or SPAmyE (Figure 7F). This suggests that the respective SP‐AmyE fusions elicited secretion stress to comparable levels, which is remarkable because, especially in the case of SPAmyE7 fused to AmyE, only a relatively low level of an AmyE degradation product was observed in the culture supernatant fraction (Figure 7A). This shows that at least some AmyE secretion was driven by SPAmyE7 and that, on balance, the effect of this low‐level AmyE secretion as sensed by the CssRS system was comparable to that of AmyE secretion with the aid of the native SPAmyE or the other shortened SP derivatives. On the other hand, the TrxA levels in culture supernatant fractions of bacteria expressing AmyE fused to SPAmyEHdom2, SPAmyE7 or SPAmyE8 were higher than in cultures expressing AmyE with the native SPAmyE (Figure S4C). This suggests that SPAmyEHdom2, SPAmyE7 and SPAmyE8 caused more cell envelope perturbation than the native SPAmyE. One of the naturally shortest SPs from B. subtilis is the SPPel, which drives secretion of the pectate lyase C (Pel) and which consists of 21 amino acid residues (Figure 1; Tjalsma et al. 2000; Lenz et al. 2023). The predicted model for interaction with Ffh showed that the SPPel with additional amino acids from the mature Pel protein has the potential for binding to Ffh (Figure 7G). The predicted ∆G value for SPPel binding to Ffh was similar compared to the ∆G values predicted for all presently evaluated SPs (Table 2). Yet, the predicted Kd value of the SPPel‐Ffh interaction was lower than that of the native SPAmyE‐Ffh interaction, but higher than the Kd values predicted for the SPAmyEHdom‐Ffh, SPAmyEHdom2‐Ffh and SPAmyE8‐Ffh interactions. This prediction may indicate that the shortest SPAmyE derivative that was designed in our present study, which is SPAmyE8, may still be shortened further without losing the ability to direct AmyE secretion. Influence of SP Mutation on the amyE mRNA Levels Mutation of a SP‐encoding DNA sequence may not only impact secretion of the respective protein, but also gene transcription and the stability of the respective mRNA (Kudla et al. 2009). Hence, altered protein secretion levels could potentially be related to changes in transcript levels. To investigate whether SPAmyE shortening resulted in different amyE mRNA levels, we performed RT‐qPCR experiments. In particular, we tested the amyE mRNA levels related to the expression of AmyE fused to SPAmyEHdom, SPAmyEHdom2, SPAmyEHdom3, SPAmyE7, SPAmyE8 or SPAmyE. The results show that bacteria expressing AmyE fused to SPAmyEHdom, SPAmyEHdom2, SPAmyEHdom3 or SPAmyE8 contained lower levels of amyE mRNA compared to the strain expressing AmyE fused to the native SPAmyE (Figure 8). Of these engineered SPs, SPAmyEHdom, SPAmyEHdom2 and SPAmyE8 directed relatively high levels of AmyE secretion (Figures 6 and 7). Together, these findings imply that the mutations in the latter three SPs had resulted in improved SP efficiency in terms of translation and/or translocation across the membrane, rather than enhanced transcription and/or mRNA stability. On the other hand, for SPAmyEHdom3 we observed relatively low amyE mRNA levels and low‐level AmyE secretion (Figures 6 and 8). This could mean that the low SPAmyEHdom3 efficiency in driving AmyE secretion resulted from relatively low transcript levels and/or lowered translation or translocation. Furthermore, in the bacteria expressing AmyE fused to SPAmyE7 we observed an elevated amyE mRNA transcript level compared to bacteria expressing AmyE fused to the native SPAmyE. Yet, this strain secreted negligible AmyE amounts into the growth medium (Figure 7), indicating that SPAmyE7 is inefficient in directing AmyE translation and/or secretion. Altogether, these findings show that the relatively high levels of AmyE secretion driven by SPAmyEHdom, SPAmyEHdom2 and SPAmyE8 cannot be attributed to elevated amyE transcript levels. RT‐qPCR analysis of amyE mRNA levels in B. subtilis TS10 amyE::Tet strains expressing AmyE with shortened SPs. The bars show the fold changes observed for the amyE mRNA levels in the different B. subtilis strains expressing AmyE fused to shortened SPs compared to the amyE mRNA level in the strain expressing AmyE with the native SPAmyE. Each assay was conducted in four biological replicates and the data is presented as the mean ± the standard error of the mean (SEM). ****, p value of ≤ 0.0001; **, p value of ≤ 0.001; *, p value of ≤ 0.05. Influence of SP Shortening on AmyE Translation and Secretion To investigate whether the relatively high or low efficiencies of our shortened SPs in driving AmyE secretion related to translation, translocation or both, we conducted pulse labelling experiments. To this end, the bacteria were labelled with [35S]‐methionine/cysteine for 2 min. Subsequently, the cells and culture supernatants were separated by centrifugation and radiolabeled AmyE in the two fractions was visualized by LDS‐PAGE and phosphor imaging. The results are shown in Figure 9. Within the 2 min of labelling, strains expressing AmyE fused to SPAmyEHdom or SPAmyEHdom2 released the highest amounts of mature AmyE into the growth medium. Slightly lower amounts of AmyE were secreted by strains expressing AmyE fused to SPAmyE8 or the native SPAmyE, whereas strains expressing AmyE fused to SPAmyEHdom3 or SPAmyE7 did not detectably secrete any AmyE within the 2 min of labelling. As expected, cells expressing AmyE without a SP accumulated the mature AmyE in the cell fraction (Figure 9). Interestingly, strains expressing AmyE fused to SPAmyE7 showed two dominant AmyE degradation products in the cellular fraction, which were not observed when AmyE was expressed without a SP. This implies that SPAmyE7‐AmyE is rejected by the secretion pathway and subsequently discarded by proteolytic cleavage. In contrast, the observation that AmyE that had been fused to SPAmyEHdom3 was exclusively detected in the bacterial cells within the 2 min timeframe of pulse labelling indicates that translocation of the SPAmyEHdom3‐AmyE fusion is inefficient. This, together with the reduced amyE transcript levels (Figure 8), explains why this shortened SP drives the secretion of only limited amounts of AmyE (Figure 6). Altogether, the results obtained by pulse labelling are consistent with the AmyE secretion levels observed by Western blotting (Figures 6 and 7). Importantly, they demonstrate that the shortened SPAmyEHdom, SPAmyEHdom2 and SPAmyE8 direct efficient membrane translocation and secretion of AmyE. Of note, slightly higher AmyE levels were observed upon pulse labelling when SPAmyEHdom was fused to AmyE, compared to AmyE fusions to SPAmyEHdom2, SPAmyE8 or the native SPAmyE. This could suggest a somewhat more efficient translation of the SPAmyEHdom‐AmyE‐encoding mRNA. Pulse labelling analysis of translation and secretion of AmyE fused to different shortened SPs. B. subtilis TS10 amyE::Tet bacteria expressing the SP‐AmyE constructs were pulse labelled for 2 min with [35S]‐ methionine/cysteine. Subsequently, bacterial cells and culture supernatants were separated by centrifugation and AmyE in both fractions was collected by immunoprecipitation with polyclonal antibodies raised against AmyE. The immunoprecipitated proteins were separated by LDS‐PAGE and radiolabeled proteins were visualized by phosphor imaging using an Amersham Typhoon Biomolecular Imager. AmyE indicates the position of the full size AmyE protein, while AmyE* indicates the position of two AmyE degradation products. Pulse labelling was conducted in two biological replicates. Discussion Optimized SPs are indispensable for enhancement of protein secretion in B. subtilis . In industrial settings, an optimized SP may boost the secretion of proteins of interest, both in the case of homologous and heterologous proteins (Brockmeier, Caspers, et al. 2006; Brockmeier, Wendorff, and Eggert 2006; Caspers et al. 2010; Heinrich et al. 2019; Grasso et al. 2023). However, finding the best SP for universal protein secretion remains a serious challenge, which relates to the high variability of naturally occurring SPs and to an insufficient understanding of the diverse interactions between SPs, proteins of interest and the secretion machinery. To make a step forward in this area, the present study was aimed at investigating relationships between the length and hydrophobicity of a SP and secretion of the respective mature protein by B. subtilis . This led to the construction of a synthetic, substantially shortened SP that directed secretion of the α‐amylase AmyE of B. subtilis equally efficiently as the native AmyE SP. The design of this shortened SP was assisted by an in silico model of SP interactions with the Ffh protein, which is part of the B. subtilis SRP complex. This implies that modelling based on the Ffh structure may assist in the design of efficient SPs for secretion of proteins of interest in B. subtilis . The results of this study show that a modified lipoprotein SP (i.e., SPMntA) can drive AmyE secretion, albeit with relatively low efficiency. In this case, the SP length was reduced to almost half the size of the native SPAmyE, that is, from 33 to 18 amino acids. Inspection of the SPMntA amino acid sequence as fused to AmyE shows that, besides being shorter, the N‐ and C‐domains of SPMntA have very different characteristics compared to SPAmyE. For instance, the N‐domain of SPMntA includes only single methionine, arginine and glutamine residues, while the C‐domain includes only one alanine residue plus the Ala‐X‐Ala motif from SPAmyE. In particular, the decrease of three positively charged amino acids in the N‐domain of SPAmyE to one positively charged amino acid in the N‐domain of SPMntA may explain the decrease of AmyE secretion when fused to SPMntA. Positively charged amino acids in the N‐domain of SPs were previously shown to be crucial for SP binding to the membrane during the initiation of secretion and loss of these amino acids was shown to reduce bacterial protein secretion (Inouye et al. 1982; Lammertyn et al. 1998; Freudl 2018). Nevertheless, a study from Caspers et al. (2010) showed that a charge reduction from +3 to +2 in SPAmyE led to increased Amy secretion. It thus seems plausible that SPAmyE needs +2 or + 3 charges to function properly as indicated by the low AmyE secretion levels directed by SPMntA and SPAmyENdom. However, also the short C‐domain of SPMntA as fused to AmyE may be suboptimal. This could relate to the lack of a helix‐breaking glycine or proline residue near the start of the C‐domain (von Heijne 1986; Tjalsma et al. 2004; Grasso et al. 2023). Such residues that can break the α‐helical conformation of the H‐domain create a hinge between the H‐ and C‐domains that will assist SP cleavage by signal peptidase which, in turn, is needed for release of a secretory protein into the extracellular milieu (van Dijl et al. 1992; Tjalsma et al. 2004; Dalbey et al. 2012). Thus, ineffective AmyE secretion driven by SPMntA may not necessarily be due to the reduced SP length in itself, but it could also relate to an inappropriate amino acid composition of the N‐ and C‐domains. To investigate which domain in the SP influences the secretion of AmyE, mutations were created in each of the SPAmyE domains and, subsequently, the secretion of AmyE was tested. The results show that the H‐domain can be shortened substantially without significant reduction in the AmyE secretion. Previous studies have shown that the Pel protein has one of the shortest SPs of B. subtilis with a length of 21 amino acids (Tjalsma et al. 2000; Lenz et al. 2023). The H‐domain of SPPel was shown to include only 12 amino acids, while SPAmyEHdom includes an H‐domain of 11 residues and SPAmyEHdom2 has an H‐domain of 7 residues, suggesting that the H‐domain of B. subtilis SPs can be shortened to at least 7 residues without a loss of performance. On the other hand, mutation of the N‐domain led to reduced AmyE secretion, while the presently tested mutations in the C‐domain did not influence secretion of the AmyE protein in a major way. However, amylase activity assays showed that the strain expressing AmyE with a mutation in the C‐domain of SPAmyE presented a lowered amylase activity in the growth medium, suggesting that the secreted AmyE had a lowered specific activity. Possibly, this can be attributed to an aberrant secretion of AmyE that was not balanced with proper protein folding and quality control (Antelmann et al. 2003; Tjalsma et al. 2004; Zweers et al. 2009; Öktem et al. 2023). Nonetheless, secretion of AmyE driven by SPAmyECdom led to comparable extracellular HtrA levels as AmyE secretion driven by the native SPAmyE. This shows that misfolding of AmyE upon fusion to SPAmyECdom, if any, was not sensed by the CssRS two‐component regulatory system (Hyyryläinen et al. 2001; Darmon et al. 2002; Öktem et al. 2023). Also, a reduction of SPAmyE to mimic the length and characteristics of SPMntA resulted in significantly reduced AmyE activity in the growth medium fraction. This reduced SP performance could possibly be due to a combination of the deleterious effects from the altered N‐ and C‐domains. Intriguingly, the strain in which AmyE secretion was directed by SPMntA secreted slightly more AmyE compared to the strain in which AmyE secretion was directed by the SPAmyENHCdomain, albeit that both SPs shared a similar length. This suggests that the overall characteristics of the SP domains might be more relevant for protein secretion than the absolute length of the SP. However, the combined results imply that both the length and the domain features are important for optimal SP function in B. subtilis . Reducing the number of amino acid residues in the H‐domain of a SP can affect the overall hydrophobicity of the SP. However, the hydrophobic nature of a SP is crucial in the interaction between the SP and SRP, especially the Ffh protein in B. subtilis as was previously observed for the SP of the α‐amylase AmyQ (Zanen et al. 2005; Janda et al. 2010). Thus, it could be anticipated that shortening of the H‐domain of SPAmyE would affect the interaction between this SP and Ffh. To evaluate this possibility, a model of the B. subtilis Ffh protein was constructed with AlphaFold3 and subsequently compared with the known structure of E. coli Ffh. The results show that both Ffh proteins have a similar structure with a hydrophobic groove in the M‐domain for interaction with hydrophobic segments in target proteins, including uncleaved SPs of membrane proteins and SPs of particular secretory proteins. The model of Ffh with the 4.5S RNA also showed that 4.5S RNA will bind to a specific part of Ffh's M‐domain near the ‘hinge’ that clamps the SP into Ffh. This interaction has been confirmed by a cryo‐EM structure of B. subtilis SRP in which the 4.5S rRNA was shown to reside in close proximity of the Ffh protein, and it is believed to stabilize Ffh‐SP‐ribosome‐nascent‐complexes (Zopf et al. 1990; Zheng and Gierasch 1997; Keenan et al. 1998; Czech et al. 2022). The Ffh model that was thus constructed, allows inspection of the possible interactions and contact sites between the different residues within SPAmyE and Ffh using a docking approach. Interestingly, residues in all three domains of SPAmyE form potential contacts with Ffh, indicating that not only the hydrophobic amino acids of a SP will interact with Ffh. Conceivably, these non‐hydrophobic amino acids interact with residues in Ffh, thereby facilitating the binding of SPAmyE to Ffh. The predicted model of the SPAmyE‐Ffh interaction can explain the different contributions of SP length and hydrophobicity in protein secretion. This turned out helpful for understanding the effects of downsizing the H‐domain of SPAmyE, which may affect the binding affinity for Ffh. To evaluate this possibility, SPAmyEHdom2 and SPAmyEHdom3 were constructed, with 7 or 5 leucine residues in the H‐domain, respectively. Of note, the interactions of both SPAmyEHdom and SPAmyEHdom2 with Ffh were predicted to mostly involve the H‐domain of these SPs. Interestingly, the AmyE protein was produced and secreted more efficiently when fused to these SPs than when AmyE was fused to the native SPAmyE, whereas the amyE mRNA transcripts relating to SPAmyEHdom‐AmyE and SPAmyEHdom2‐AmyE were less abundant than the amyE transcript for SPAmyE‐AmyE. Clearly, the lowered transcript abundance relates to the mutations that were introduced to shorten the H‐domain of SPAmyE, but the exact underlying mechanism leading to the altered mRNA levels is currently unknown. In any case, the combined findings imply that both SPs with shortened H‐domains drive AmyE secretion with higher efficiency than the native SPAmyE. The latter view is fully supported by our pulse labelling studies. Meanwhile, interactions of SPAmyEHdom3 with Ffh were predicted mostly in the N‐domain, rather than the H‐domain. This coincides with the finding that SPAmyEHdom3 did not drive effective AmyE secretion as evidenced by Western blotting, amylase activity assays and pulse labelling. Both Western blotting and the pulse labelling results show that SPAmyEHdom3‐AmyE was expressed, but that it was mostly rejected from the secretion pathway and degraded. This could, for instance, be due to inefficient interactions with the SecYEG machinery or signal peptidases. In this context it is noteworthy that the predicted model of SPAmyEHdom3 did not reveal a typical ‘hinge’ at the start of the C‐domain, which is deemed necessary for efficient SP recognition and cleavage by signal peptidase. Thus, it is conceivable that signal peptidase cannot recognize the Ala‐X‐Ala motif in SPAmyEHdom3, resulting in ineffective AmyE secretion. It is presently difficult to say why no flexible hinge was predicted for SPAmyEHdom3 despite the presence of a glycine residue at the start of the C‐domain. Furthermore, it is noteworthy that in bacteria expressing SPAmyEHdom3‐AmyE, we noticed that the amyE RNA transcript level was reduced compared to bacteria expressing AmyE with the native SPAmyE. This shows that the mutation to generate SPAmyEHdom3 had a similar effect on the amyE transcript level as the mutations for generating SPAmyEHdom and SPAmyEHdom2. To further reduce the total length of the SP, two short synthetic SPs were created, namely SPAmyE7 and SPAmyE8. Western blotting, activity assays and pulse labelling show that SPAmyE8 can drive AmyE secretion with about the same efficiency as the native SPAmyE. Also, in bacteria expressing SPAmyE8‐AmyE, we observed reduced amyE transcript levels. On balance, this implies that the shortened SPAmyE8 is somewhat more efficient than the native SPAmyE. Intriguingly, the model of SPAmyE8 binding to Ffh indicated that this SP is bound in the inverse orientation compared to the other mutated SPs investigated in our study. Whether this inversed binding of SPAmyE8 to Ffh really occurs in bacteria expressing SPAmyE8‐AmyE remains to be demonstrated but, judged by our modelling results, this could be possible. In this respect, it is noteworthy that it is presently unknown whether SPs need to bind to Ffh in one particular orientation for effective protein targeting to the membrane‐embedded Sec translocon. Importantly, SPAmyE8 has a total length of 22 amino acids, including an H‐domain of 10 hydrophobic residues. This is currently the shortest SPAmyE variant that can drive efficient AmyE secretion. Compared with the SPPel, the H‐domain of SPAmyE8 contains two hydrophobic residues less. On the other hand, SPAmyE8 is still one amino acid longer than SPPel. It is thus conceivable that it will be possible to further shorten SPAmyE8 in a future study. In contrast to SPAmyE8, the shortened SPAmyE7 turned out to be ineffective in driving AmyE secretion. In fact, cells expressing the SPAmyE7‐AmyE fusion presented only cleavage products of AmyE, suggesting that this precursor was recognized by the bacterial machinery for protein quality control and degraded by one or more of the membrane‐associated or cytoplasmic proteases of B. subtilis (Dalbey et al. 2012; Harwood and Kikuchi 2021). Our pulse labelling analysis suggests that this degradation happened predominantly in the bacterial cells, albeit that Western blotting showed an AmyE degradation product in the culture supernatant. Since the HtrA levels detected for bacteria producing SPAmyE7‐AmyE were not significantly elevated compared to those of bacteria producing AmyE with the native SPAmyE, it seems unlikely that the SPAmyE7‐AmyE precursor elicits a CssRS‐dependent secretion stress response. Lastly, while our present study has provided novel insights into the requirements concerning SP length and hydrophobicity for protein secretion in B. subtilis , it was focused on the secretion of one model reporter protein, namely AmyE. Accordingly, it is our objective for future studies to determine how well our approach for SP shortening and modelling of the SP‐Ffh interaction will work with secreted proteins other than AmyE. It is furthermore important to bear in mind that, although interaction of SPAmyE or the shortened derivatives of SPAmyE with Ffh is possible, as judged by the results from our docking analyses, this interaction has not been demonstrated experimentally. Accordingly, it will be interesting for future studies to investigate these interactions by in vitro measurements and to determine how well the predictions of the respective parameters, for instance the Kd values, fit with the actual parameters. Nonetheless, based on our present observations, we consider our approach to use the B. subtilis Ffh protein as a ‘molecular SP mold’ of value for predicting whether mutant SP designs could be able to drive effective protein secretion. Another observation that deserves further investigation concerns the impact of modified SPs on the bacterial physiology. While the here‐described engineered SPs fused to AmyE did not elicit strong CssRS‐dependent secretion stress responses, they did lead to some perturbation of the bacterial cell envelope as evidenced by elevated extracellular levels of the cytoplasmic marker protein TrxA. At present, we can only speculate about the underlying mechanism(s). Conceivably, the shortened SPs engage to some extent directly with the phospholipid bilayer of the membrane, or they cause to some degree interference with the Sec pathway that leads to some leakage of cytoplasmic content. Importantly, the shortened SPs did not detectably impair the overall growth of the bacterial cultures, so the membrane‐perturbing effects were relatively minor. Nonetheless, it will be interesting to dissect the possible stresses that different engineered SPs can impose on the bacterial cells. Conclusion Altogether, our present study has shown that the SP of B. subtilis AmyE can be shortened to at least 60% of its original length while still retaining its full capacity to drive AmyE secretion. Furthermore, the use of an in silico model to predict the interaction between SPAmyE and Ffh proved to be beneficial for evaluating the synthetic SP functionality. In this regard, in silico modelling is likely to become a promising tool to evaluate the interactions of SP‐mature protein combinations with the protein secretion machinery of B. subtilis and other microbial cell factories, potentially accelerating the design of high‐fidelity SPs to maximize secreted protein yields. Author Contributions Lorenzo de la Parra Soto: investigation. Girbe Buist: writing – review and editing. Dicky A. Pranoto: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, funding acquisition. Danilo Milo: investigation. Jan Maarten van Dijl: conceptualization, investigation, funding acquisition, writing – original draft, methodology, writing – review and editing, project administration, supervision, resources. Jolanda Neef: investigation, writing – review and editing, methodology. Funding This work was supported by theGraduate School of Medical Sciences of the University of Groningen, the Netherlands. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Structural analysis of the B. subtilis Ffh protein. Figure S2: Melting curves of the amyE and gyrA qPCR products. Figure S3: SignalP 6.0 secretory pathway prediction for (A) SPAmyE‐AmyE and (B) SPMntA‐AmyE with an Ala‐Ser‐Ala signal peptidase I recognition motif. Figure S4: Western blotting analysis of TrxA protein levels in cells or culture supernatant fractions of strains expressing AmyE fused to different mutated SPs, or the native SPAmyE. Figure S5: Predicted complexes of the B. subtilis Ffh protein with (A) the native SPAmyE alone or (B) with SPAmyE plus 20 additional amino acids from the mature AmyE protein. Table S1: List of primers used in this study. Table S2: Predicted template modelling (pTM) scores from AlphaFold3 for the SPs and Ffh proteins used for molecular docking studies.