Introduction

Virulence-associated type III secretion systems (T3SS), whose core structure is also known as injectisome, are a highly conserved protein complex mediating translocation of a variety of effector proteins into host cells. T3SSs are large machines with a size of ∼7 MDa, consisting of about 200 subunits and spanning the bacterial inner (IM) and outer (OM) membranes as well as a host membrane (Zilkenat et al., 2016). These nanomachines consist of several substructures, including the export apparatus, the membrane-anchored base, the cytoplasmic components, and the needle and tip.

The export apparatus is a core component of the T3SS, localized within the membrane-anchored base. The export apparatus, which is also conserved in the closely related bacterial flagella, is composed of five subunits, namely SctR, SctS, SctT, SctU, and SctV, with a stoichiometry of 5:4:1:1:9 (Kuhlen et al., 2018; Zilkenat et al., 2016). SctRSTU constitute the core export apparatus, which forms a pore that T3S substrates pass through before entering the secretion channel formed by needle adapter and filament proteins (Hu et al., 2019; Torres-Vargas et al., 2019). In non-secreting conditions, the export apparatus pore is gated by the SctT plug, but contact with substrates results in opening of the gate and substrate secretion (Hüsing et al., 2021; Kuhlen et al., 2018; Miletic et al., 2021). The export apparatus has been shown to contribute to hierarchical secretion of substrates through SctU (Björnfot et al., 2009; Edqvist et al., 2003; Inoue et al., 2019; Monjarás Feria et al., 2015) and SctV (Marcos-Vilchis et al., 2025; Portaliou et al., 2017; Yuan et al., 2021). SctV directly interacts with the substrates and their chaperones through its cytoplasmic domain (Gilzer et al., 2022; Portaliou et al., 2017; Xing et al., 2018) and also couples the proton motive force with substrate translocation (Erhardt et al., 2017; Minamino et al., 2011, 2016).

Recent structural studies of the export apparatus have shed light on its unique structural features. The core export apparatus (SctR₅S₄T₁U₁) is a helical pore, resembling an inverted cone, with the tip oriented toward the cytoplasmic side. Although the subunits are predicted to be transmembrane proteins, the core export apparatus is not fully integrated into the IM but instead protrudes significantly into the periplasmic side (Kuhlen et al., 2018). The core export apparatus displays pseudo-hexameric symmetry, which results from structural similarity between SctT and the combination of SctR and SctS (Kuhlen et al., 2018). In addition, the export apparatus exhibits a right-handed helical structure, characterized by the hydrophobic alpha helices being staggered by approximately 25 Å. The staggered conformation is stabilized by inter- and intramolecular salt bridges (Kuhlen et al., 2018; N. Singh et al., 2021). This helical configuration templates the helical needle adapter and filament in the downstream T3SS assembly (Goessweiner-Mohr et al., 2019; Torres-Vargas et al., 2019).

Based on biochemical and structural data, the helical assembly of the export apparatus is believed to occur in a sequential manner (Dietsche et al., 2016; Johnson et al., 2019; Kuhlen et al., 2018; N. Singh et al., 2021). Assembly begins with the formation of pentameric SctR, followed by the integration of one SctT molecule into the SctR pentamer to form a stable SctR₅T₁ core, which serves as a pseudo-hexameric building block. Four SctS peripherally assemble onto the SctR₅T₁ core in a sequential manner, forming SctR₅S₄T₁ (Dietsche et al., 2016; Johnson et al., 2019; Kuhlen et al., 2018; N. Singh et al., 2021). The assembly of the core export apparatus is completed by recruitment of one SctU (Kuhlen et al., 2020; Wagner et al., 2010). Once the assembly of the nonameric SctV is completed, the core export apparatus is repositioned from the IM into the periplasmic space, a process that is further stabilized by interactions with the IM base substructures (Kuhlen et al., 2018). The export apparatus has been shown to be a critical starting point for the T3SS assembly process (Diepold et al., 2010; Wagner et al., 2010). It facilitates the assembly of the base as a scaffold, which in turn allows the assembly of the cytoplasmic components, enabling secretion and assembly of the needle components (Wagner et al., 2010).

As the nucleus and scaffold of T3SS assembly, the export apparatus must be efficiently and correctly assembled. While many protein complexes rely on diverse regulatory mechanisms to achieve stoichiometric and orderly assembly rather than relying on stochastic interactions, the regulation of export apparatus assembly remains poorly understood. Previous studies have shown that the core export apparatus genes exhibit high synteny (Abby & Rocha, 2012), which indicates physiological relevance, and often contain regulatory motifs. In the context of an operon, gene order dictates the assembly order of the protein complex, especially when the order is conserved (Wells et al., 2016). However, in the case of the export apparatus, gene order does not reflect assembly order. This observation led us to investigate whether the highly conserved gene order plays a role in regulating the assembly of this unique helical export apparatus.

We demonstrate that the conserved gene order of the export apparatus harbors a secondary RNA structure within the sctS mRNA, which tightly regulates the translation of sctT in Salmonella Typhimurium. This mechanism enables translational coupling between sctS and sctT, ensuring precise stoichiometric control. This regulation is particularly crucial for SctT, as SctT is overexpressed in the absence of the regulation and self-assembles into cytotoxic, unregulated multimers that may form leaky pores in the bacterial inner membrane. Thus, the translational coupling between sctS and sctT plays a vital role in maintaining robust and efficient assembly of the export apparatus and thus vitality of the bacteria harboring T3SS.

Results

Genes of the T3SS core export apparatus show a strict synteny

It was noted that the export apparatus genes are conserved in their genomic order. This drew our interest, especially given that the assembly order of the export apparatus differs from the order of the respective genes (Fig. 1A and 1B). In order to analyze the conservation of the gene order of the export apparatus components systematically, we performed a comprehensive in silico analysis of its gene organization in many bacterial species expressing T3SS.

Figure 1.

Using the Cblaster program (Gilchrist et al., 2021), we performed BLAST searches to identify homologs of the sctRSTU gene cluster. To minimize redundancy, particularly from overrepresented or extensively sequenced strains, only hits from bacterial reference genomes available in NCBI were included. Genomes containing all four sctRSTU genes were retained for further synteny analysis, which resulted in 1,152 bacterial genomes. These hits were categorized into different groups, based on each gene’s strand (+ or -) and locus information, and the intergenic distances between the genes in a cluster. Only when all genes in a sctRSTU cluster were localized to either the positive or negative strand and had intergenic distances of less than 500 bp, the hits were considered as having the synteny. Among the 1,152 export apparatus clusters, 848 were of T3SS (injectisome) and 304 were of flagella (Fig. 1C). In both injectisome and flagellar export apparatus genes, sctRSTU genetic organization was highly conserved, being present in 87.1 % (n = 1,003) in all analyzed genomes (83.8 % and 96.1 % in export apparatus gene clusters of injectisome and flagella, respectively) (Fig. 1C and Supplementary Table 2). In 3.6 % of flagella (n = 11), the sctU gene was separated from sctRST (sctRST-U), which is the case for S. Typhimurium strain LT2 genome. In injectisomes, this was7.3 % (n = 62). Other genetic organizations were additionally observed in injectisome export apparatus clusters, including sctRS-T-U (7.2 %), sctR-STU (0.9 %), sctRS-TU (0.6 %), and sctR-S-TU (0.1 %). The sctRS-T-U was predominantly observed in the Burkholderiaceae family, whereas the sctRST-U was mainly found in the Alphaproteobacteria class (Figure 1C and 1D, and Supplementary Table 2).

Effective plasmid-based complementation of a chromosomal sctT deletion mutant requires expression of more than just sctT

We then sought to investigate the functional implications of the observed high level of gene order conservation using the export apparatus of the T3SS encoded by Salmonella pathogenicity island 1 (SPI-1) as a model. We reasoned that in trans complementation may be compromising if the synteny of export apparatus genes is important for assembly.

S. Typhimurium strains with individual deletions of export apparatus genes (ΔsctR, ΔsctS, ΔsctT, ΔsctU) were complemented in trans from rhamnose-inducible pT10 plasmids expressing a combination of export apparatus subunits. The secretion functionality of the assembled T3SS was assessed in each strain by luminescence measurements of secreted SipA-NanoLuc (NL) fusion protein (Fig. 2A) (Westerhausen et al., 2020). ΔsctR, ΔsctS, and ΔsctU strains exhibited comparable secretion levels of 50% of the wild-type when the missing export apparatus subunit was expressed alone or together with other subunits. In contrast, complementation of the ΔsctT mutant only reached secretion of less than 20% of the wild-type except when all four export apparatus genes were expressed together. These results indicate that the genetic context of sctT is critical for the assembly of the export apparatus.

Figure 2.

We further investigated the relevance of the high synteny on the accumulation levels of export apparatus components by SDS-PAGE and on export apparatus assembly by blue native (BN)-PAGE (Zilkenat et al., 2024). SctR, SctT, and SctU, respectively, were analyzed using a 3xFLAG epitope tag, Western blotting and immunodetection as previously reported (Wagner et al., 2010). In the wild-type strains, all proteins were detected robustly by SDS-PAGE at their expected molecular mass (Fig. 2B). In the BN-PAGE, a distinct band of the respective proteins as part of the assembled needle complex was detected at above 1236 kDa. Subcomplexes of assembly intermediates containing SctRT (only weakly), SctRST and SctRSTU were detected around 242 kDa as previously reported (Fig. 2B) (Singh et al., 2021).

In the complementation of the ΔsctR mutant, expression of all constructs resulted in wild-type-like accumulation levels of the respective export apparatus components. Also the respective complexes occurred like in the wild-type strain (Fig. 2B).

Complementation of the ΔsctS mutant by plasmids lacking sctR resulted in SctT^FLAG and SctU^FLAG accumulation levels reduced to about 1/3rd of the wild-type, pointing at a severe stability defect of the respective proteins (Fig. 2B). Concomitantly, SctRST(U) subcomplexes were not found by BN-PAGE while protein complexes at the size of the full needle complex could be detected. Complementation of the ΔsctS mutant by plasmids encoding SctRS, SctRST and SctRSTU, respectively, yielded stable protein accumulation levels as well as the formation of all complexes. It is noteworthy that the intensity of the SctRSTU complex band was substantially stronger in the complementation with SctRST and even more with SctRSTU compared to complementation with SctRS only.

Complementation of the ΔsctT mutant by plasmids encoding SctT^FLAG or SctT^FLAGU showed an up to three-fold overexpression of SctT^FLAG compared to the wild-type and a ladder of complex bands by BN-PAGE, presumably consisting of SctT multimers (Fig. 2B). Complementation of this mutant with every other combination except SctRSTU showed strongly reduced protein accumulation levels and no detectable formation of SctRST(U) subcomplexes.

Formation of the SctRSTU complex was also reduced in the ΔsctU mutant unless complemented with all four core export apparatus components (Fig. 2B).

Together, the results of the complementation analysis show that gene synteny is most critical for SctT and to a lesser extent also for SctS. Assembly of the SctRSTU subcomplex suffers if these proteins are not expressed in their genetic context. However, the surplus of proteins generated by plasmid-based complementation seems to yield a sufficient amount of functional needle complexes in most cases as a significant functional defect is limited to the complementation of the ΔsctT mutant. Expressing sctT without an upstream sctS gene results in SctT overexpression and futile multimerization of the protein.

Scrambling of gene order and disconnecting sctS and sctT result in unproductive complex assembly

To further understand how the genetic organization affects export apparatus assembly, we generated constructs with scrambled orders of export apparatus genes. pT10 plasmids containing sctSTRU, sctTRSU, and sctTSRU complemented a ΔsctRSTU S. Typhimurium strain. Scrambling the gene order resulted in about 95% reduction in SipA-NL secretion compared to the wild-type gene order (Fig. 3A). A significant reduction in the accumulation levels of the respective proteins as well as inefficient NC and SctRSTU complex assembly were observed throughout with the exception of the overexpression of SctT and its concomitant multimerization in complementations with sctT^FLAGRSU and sctT^FLAGSRU (Fig. 3B).

Figure 3.

A strict requirement for the conservation of gene order indicates that the connections between the genes may play a role in core export apparatus assembly. Towards testing this hypothesis, a 20 bp spacer (denoted as _) was introduced between SctR and SctT, and SctS and SctT, respectively. The respective native Shine-Dalgarno (SD) sequence of sctS or sctT was placed within the spacer at the native distance from the start codon. As SctU is not strictly required for NC assembly (Wagner et al., 2010) and sctU does not show high synteny relative to sctRST (Fig. 1B), the sctT-sctU gene connection was omitted from analysis. The spacer between sctR and sctS did not affect the assembly (Fig. 3B) and resulted only in a minor reduction (14%) of SipA-NL secretion (Fig. 3A). On the other hand, the spacer between sctS and sctT resulted in impaired assembly and secretion function (96% reduction) (Fig. 3A), as well as in SctT overexpression and aggregation, alongside a reduced accumulation of SctR and SctU (Fig. 3B). The effect of disconnecting sctS and sctT was even more pronounced when the spacer was introduced in the chromosome, with an increase in the SctT accumulation levels of 14-fold, highlighting the importance of the sctS-sctT gene connection (Supplementary Fig. 1). Together, these results indicate that the need for a particular connection between sctS and sctT is one critical aspect of the synteny of core export apparatus genes. Disruption of this connection leads to unregulated expression of SctT with concomitant multimerization, instability of the other core export apparatus components and a defect in the assembly of functional needle complexes.

An mRNA stem-loop structure at the end of sctS inhibits independent translation of sctT

The observation of the overexpression of sctT in the absence of sctS immediately upstream points to a negative regulation of sctT expression by sctS. We hypothesized that this regulation could come from an inhibitory mRNA structure. Prediction of the mRNA structure (J. Singh et al., 2021) identified a stem-loop structure at the 3’ end of the sctS mRNA (Fig. 4A). As the SD sequence of sctT is located within the 3’ end of sctS, the predicted sctS stem-loop structure may restrict SD recognition by a ribosome and independent translation initiation of sctT.

Figure 4.

To investigate the role of the stem-loop in regulating sctT expression, we designed constructs in which sctS and sctT were replaced by mCherry and sfGFP, respectively, and were the sctS stem-loop sequence was fused to mCherry to mimic the native gene connection between sctS and sctT (Fig. 4B). The use of fluorescent proteins ensures that results are not influenced by membrane targeting and insertion, the stability of individual proteins, or complex assembly. We then introduced silent mutations to destabilize the stem-loop, and the aforementioned 20 bp spacer between the mCherry and sfGFP genes as described above, recapitulating the disconnection of sctS and sctT genes. Analysis of the fluorescence ratio of mCherry and sfGFP from both stem-loop destabilization and spacer mutants revealed a significant increase in sfGFP fluorescence compared to mCherry (Fig. 4C). This result indicates that the sctT SD sequence is concealed by the stem-loop structure and may require melting by a ribosome that translates SctS. This mechanism would couple the two genes translationally.

To test how far a ribosome needs to melt this stem-loop structure to translate sctS, we introduced ochre stop codons along the sequence of sctS (Fig.4A) and measured the luminescence output of NanoLuc luciferase (NL) that replaced the sctT gene (Fig. 4D). Introducing stop codons upstream of the stem-loop resulted in a strongly diminished luminescence compared to the wild-type. On the other hand, introducing stop codons within the stem-loop led to NL expression. Starting from position L72, which base-pairs with the sctT SD sequence, a substantial increase in the NL expression was observed upon stop codon introduction (Fig. 4E). These results indicate the presence of a strict translational coupling between sctS and sctT: sctT is only translated when the ribosome translating sctS melts the stem-loop structure and allows access to the sctT SD. Otherwise, the sctS stem-loop inhibits sctT SD recognition by a ribosome and independent sctT translation.

Previous studies on translational coupling have shown that the distance between the stop codon of the upstream gene and the start codon of the downstream gene affects translational coupling efficiency (van de Guchte et al., 1991). To investigate if this distance is fine-tuned for the translational coupling of sctS and sctT, we made use once more of the mCherry-sctS_stem-loop-sfGFP plasmid (Fig. 4B). The distance between sctS stop and sctT start codons was either increased or decreased by 3 nt (denoted as +3 and -3). Additionally, overlapping stop and start codons (AUGA) were generated (denoted as -6). In order to maintain the stem-loop structure upon changing the distance, complementary mutations (denoted as ‘c’) were introduced to the opposite strand of the stem-loop. The sole effect of the complementary mutations was tested as a control (denoted as ‘oc’). The ratio between mCherry and sfGFP fluorescence did not change upon varying the distance between mCherry stop and sfGFP start codons (Fig. 4C). Together, these results suggest that the distance between the sctS stop and sctT start codons is not optimized towards a particular translational coupling, eventually leading to the required 4:1 stoichiometry of the two complex components.

Regulation of sctT translation by the stem-loop structure is independent of the gene in front of sctT

Proper folding and assembly of protein complexes often depend on co-translational processes (Shiber et al., 2018). Since export apparatus subunits require each other for stability (Wagner et al., 2010), co-translation may help circumvent an inherently unstable monomeric state, allowing the complex to form more efficiently.

As SctS and SctT proteins interact and are translationally coupled, we wondered if SctS and SctT require each other for co-translational assembly. To this end, mCherry was inserted upstream of sctT, coupling sctT and mCherry in a manner analogous to the native sctS-sctT connection and at the same time separating sctS from sctT. The sctS stem-loop sequence was fused to the 3’ end of the mCherry gene either in the presence or absence of sctS (Fig. 5A).

Figure 5.

Then, we assessed the SipA-NL secretion after complementation of the ΔsctRSTU mutant with the pT10-sctRSmcherryTU construct. Surprisingly, translational coupling by mCherry-sctT instead of sctS-sctT did not impact the secretion function, indicating that co-translational interaction between SctS and SctT is not the critical point of export apparatus assembly and the reason for translational coupling of these two proteins (Fig. 5B).

In addition, we assessed the assembly of the SctR₅T₁ subcomplex with mCherry-sctT translational coupling and without SctS via in vivo photo-crosslinking. This method relies on the genetic encoding of the artificial UV-reactive amino acid para-benzoyl-phenylalanine (pBpa) by suppressed amber stop codons (UAG) that can be introduced by site-directed mutagenesis at any protein position of interest (Farrell et al., 2005; N. Singh et al., 2021). Upon UV irradiation, pBpa forms free radicals and crosslinks to closely interacting proteins. The observation of a pBpa crosslink, e.g. by Western blotting of crosslinked proteins, ascertains that a specific protein-protein interaction has occurred. To assess SctR-SctT assembly, we replaced V170 of SctR by pBpa; SctR_V170X was reported previously to crosslink to SctT (Fig. 5C) (Dietsche et al., 2016). The absence of SctS did not affect the crosslinking efficiency of SctR_V170X to SctT (Fig. 5D), supporting the notion that co-translational interaction of SctS and SctT is not relevant for assembly of the core export apparatus as long as SctT expression is regulated by translational coupling via the sctS mRNA stem-loop.

While the assembly of the SctR₅T₁ subcomplex proves to be independent of the SctS protein, SctS requires SctR₅T₁ for its assembly. In order to evaluate how critical translational coupling of sctS and sctT is for SctS assembly onto SctR₅T₁ complexes, we designed constructs in which sctS was placed upstream or downstream of the sctRT genes (sctSRTU or sctRTSU, respectively). In the designs of the sctRT genes, we either maintained sctT translational coupling through a remnant sctS stem-loop (denoted as SL) or connected the sctR and sctT genes just by the original TAA AAA ATG sequence without the regulating stem-loop but providing the sctT SD within the sctR gene (denoted as C for directly Connected) (Fig. 5E). The A223K and T224G mutations of SctR to accommodate sctT’s SD within sctR (denoted R*STU in Figs. 5E-G) did not have a substantial impact on protein accumulation levels, complex assembly and type III secretion function. However, SipA-NL secretion was strongly reduced when sctS was upstream of sctRT (sctSRTU) in both C and SL connections (Fig. 5F). When expressing SctS behind SctRT, only the construct that included the sctS stem-loop between sctR and sctT (sctR^SLTSU) successfully complemented type III secretion function while the direct connection of sctR and sctT failed to do so (Fig. 5F). Together, this suggests that the direct availability of the SctR₅T₁ subcomplex is important for the assembly of SctS.

In BN-PAGE, the constructs expressing the directly connected sctR-sctT genes without a sctT-controlling stem-loop produced a strongly altered banding pattern of subcomplexes with much reduced amounts of assembled needle complexes (Fig. 5G). A mild SctT overexpression and multimerization phenotype was observed when expressing sctR^CTSU, while, with the exception of SctR in sctR^SLTSU, all tested export apparatus components showed reduced accumulation levels in the different mutants pointing at compromised assembly and stability. Concomitantly, only expression of the sctR^SLTSU construct resulted in wild type-like band intensities of SctR^FLAG-containing subcomplexes.

In summary, our results reveal two important aspects in the core export apparatus assembly: First, sctT requires translational coupling through the stem-loop for appropriate expression levels, but the nature of the coupled gene in front is not important. Second, it seems to be critical for sctS to be translated after sctR in order to form SctRST complexes efficiently.

Reducing sctT translation with inefficient start codons alleviates effect of uncoupled translation

Since translational coupling through the stem-loop prevented SctT overexpression and multimerization, we investigated whether the requirement for translational coupling can be waived by reducing sctT translation levels. It is known that translation levels correlate with the start codon usage (O’Donnell & Janssen, 2001), and that the alternative start codons GTG and TTG are used in 14% and 3% of E. coli genes, respectively, whereas ATG is used in 82% (Blattner et al., 1997; Hecht et al., 2017). Therefore, in the constructs that led to SctT overexpression, namely, sctTRSU and sctTSRU, we replaced the ATG start codon of sctT with two alternative start codons, GTG and TTG. The use of the TTG start codon significantly reduced SctT levels compared to the ATG start codon (Fig. 6A). In the BN-PAGE, high and low molecular weight SctT-containing complexes were reduced when the alternative start codons were used, with the TTG start codon having a more pronounced effect than GTG, reflecting their translation efficiency (Fig. 6A). Having SctT overexpression reduced also alleviated the secretion defect by 2-fold with TTG and up to 4-fold with GTG compared to the ATG start codon (Fig. 6B). In the case of the sctTSRU construct, having a GTG start codon restored the secretion function up to 82% of the wild-type gene order, compared to18% with the ATG start codon (Fig. 6B). We also observed that SctT overexpression caused growth attenuation of the expressing bacteria (Fig. 6C). Reducing SctT expression with alternative start codons restored growth to wild-type levels (Fig. 6C). Taken together, these results suggest that reducing SctT overexpression can compensate for the lack of translational coupling between sctS and sctT, and the associated problems in assembly, secretion and growth.

Figure 6.

Strict regulation of sctT translation by the mRNA stem-loop prevents futile polymerization of SctT

Why is SctT overexpression so detrimental to export apparatus assembly and even results in affecting bacterial growth? To address this, we had a closer look at the structure of the export apparatus. SctT resembles a unit of SctR and SctS, which makes the export apparatus a pseudo-hexameric pore (Kuhlen et al., 2018). Hence, it is well conceivable that SctT assembles into SctT-only multimers in the absence of stem-loop regulation (Fig. 7A). To validate self-assembly of SctT, we performed in vivo photo-crosslinking of SctT with a pBpa mutant (SctT_F132X) that reveals SctT-SctS crosslinks in the wild-type (Dietsche et al., 2016) (Fig. 7AB). When SctT was overexpressed due to the lack of control by the stem-loop, these crosslinks were absent. Instead, SctT was crosslinked with SctT, forming homo-oligomers (Fig. 7B). This result supports that the ladder observed by BN-PAGE upon SctT overexpression corresponds to SctT multimers (Fig. 6A). This self-assembly likely results in the formation of ungated leaky pores that impair membrane integrity and consequently bacterial growth. Alternative to the assumed helical self-assembly of SctT, also a planar penta- or hexameric pore formation is conceivable as predicted by AlphaFold multimer (Supplementary Fig. 2). Due to the propensity of SctT for futile self-assembly, a tight control by the stem-loop may become crucial to ensure that only a few copies of SctT are translated, which then are swiftly assembled onto SctR₅ to form SctR₅T₁.

Figure 7.

Conservation of sctS stem-loop structure in other T3SS

Our study has identified translational coupling of sctST as a crucial regulatory mechanism for export apparatus assembly in SPI-1 T3SS in S. Typhimurium. Given the conserved synteny in the genes encoding the export apparatus subunits among pathogens with T3SS (Fig. 1A and 1C), we hypothesized that control of sctT translation through a sctS mRNA stem-loop may be a more universal mechanism to facilitate faithful assembly of the core export apparatus. Indeed, we found similar sctS mRNA stem-loop structures that are predicted to be present around sctT SD sequences of other T3SS (Supplementary Fig. 3), supporting our notion. As the sctS mRNA structures are machine learning-based predictions, the presence of sctS stem-loop structures and their regulatory effects on sctT translation and export apparatus assembly awaits further experimental validation for other pathogens.

Discussion

The T3SS is an intricate protein complex machinery that requires highly coordinated assembly of its subunits and subcomplexes, which involves a variety of regulatory measures. In this study, we investigated the assembly regulation of the export apparatus. Our findings reveal that high synteny in export apparatus genes relates to successful export apparatus assembly. A key regulatory motif in the conserved genetic organization of the export apparatus is the translational coupling between the sctS and sctT genes. This translational coupling occurs through an mRNA stem-loop structure, which conceals the Shine-Dalgarno (SD) sequence of the sctT gene. The stem-loop inhibits de novo translation of SctT by a new ribosome, allowing SD recognition only through translational coupling with sctS. In absence of translational coupling via the stem-loop, SctT is overexpressed and assembles into futile multimers (Fig. 8).

Figure 8.

Our results stand in contrast to a study reporting that complete recoding of the SPI-1 operon, eliminating all internal regulatory elements for assembly, still permitted formation of needle complexes (Song et al., 2017). However, this study only provides a qualitative assessment of complex formation with no quantitative information on the efficiency of T3SS assembly from the synthetic T3SS operon, in which all components were overexpressed and thus available in surplus. In physiological context, cellular resources are limited, and the T3SS, a massive and energy-costly complex, requires its components to be produced in strict stoichiometric ratios. Supporting this, ribosome profiling in E. coli has shown that subunits of protein complexes are typically synthesized in precise proportion to their final stoichiometric ratios (Li et al., 2014). In line with this, our results clearly demonstrate that the conserved gene order sctRSTU is important for stoichiometric expression of the export apparatus subunits. This is especially important for SctT, whose overexpression is detrimental to pathogen fitness, and it ultimately ensures efficient and robust assembly of the export apparatus.

The propensity of SctT to self-interact can be inferred from the structure of the export apparatus. SctT alone resembles a structural combination of SctR and SctS and constitutes a unit within the pseudo-hexameric SctR₅S₄T₁ pore (Kuhlen et al., 2018). It is thus conceivable that SctT forms pore-like assemblies similar to the export apparatus. However, in the absence of the gating mechanism in the export apparatus (Hüsing et al., 2021; Kuhlen et al., 2018; Miletic et al., 2021), such aberrant SctT pores may lead to membrane leakage and growth defects, as observed under SctT overexpression conditions (Fig. 6). Given that the export apparatus is a highly conserved component of the T3SS and shares high structural similarity with its homologs across closely related bacterial species as well as in the flagellar system (Johnson et al., 2019; Lunelli et al., 2020), it is plausible that SctT homologs likewise possess a propensity for self-assembly and hence require regulations for proper complex formation.

The fate of SctS molecules during formation of SctR₅T₁ subcomplexes remains an intriguing question. A previous study by our lab demonstrated that SctT can stably assemble into the export apparatus only when salt bridges are established within the SctR₅T₁ subcomplex (N. Singh et al., 2021). Since SctS is translated before SctT, it is conceivable that SctS molecules diffuse away from the site of export apparatus assembly. A gene order matching the assembly order would therefore seem more logical unless there is another reason for sctS to be positioned upstream of sctT. Studies on translational coupling have shown that proteins from downstream genes are often chaperoned by proteins from upstream genes, aiding in correct folding (Basu et al., 2004; Nakatogawa et al., 2004; Praszkier & Pittard, 2002). However, our crosslinking data showed that SctS is dispensable for SctR₅T₁ subcomplex formation, arguing against the possibility that SctS functions as an intramembrane chaperon for SctT (Fig. 5D). On the other hand, we showed that expressing the export apparatus genes in the assembly order, with wild-type level SctT expression, still fails to recapitulate the robustness of the wild-type gene order (Fig. 6A). This indicates that the conserved gene order indeed fine-tunes the assembly and suggests a possible additional role of SctS in this process. It is possible that SctS transiently interact with the preformed SctR pentamer and stabilize its assembly by forming salt bridges with SctT. This hypothesis requires further investigation.

Given this inherent risk of SctT’s self-assembly, it is conceivable that maintaining the sctS-sctT synteny serves to preserve the sctS stem-loop, which in turn tightly regulates SctT translation and ensures proper assembly of the export apparatus. While our study shows that the sctS stem-loop regulates SctT translation via translational coupling in Salmonella, it may well be that the sctT stem-loop has different functions and mechanisms of action in other bacteria. For instance, in Yersinia, the presence of an sctS stem-loop has been reported, which harbors the SD sequence of sctT, like in Salmonella (Pienkoß et al., 2022). However, the Yersinia sctS stem-loop acts as an RNA thermometer, enabling translation of SctT upon reaching the host temperature (Pienkoß et al., 2022). An RNA thermometer is a regulatory motif frequently observed in T3SS regulation in Yersinia, as it aligns with Yersinia’s lifestyle and infection process (Hoe & Goguen, 1993; Pienkoß et al., 2021, 2022).

Together, we conclude that the synteny of the export apparatus genes provides a critical regulation for helical assembly of the pseudo-hexameric complex. Due to its unique structural symmetry, the export apparatus subunit SctT is prone to self-assembly and the sctS stem-loop provides a strict expression control via translational coupling between sctS and sctT.

Materials and methods

Materials

Most chemicals were purchased from Sigma-Aldrich, Novagen, New England Biolabs, Merck, Carl Roth, Becton Dickinson, or Applichem, unless otherwise specified. Para-benzophenylalanine (pBpa) was purchased from Bachem. Lauryl maltose neopentyl glycol (LMNG) was from Anatrace Products LLC. SERVAGel™ TG PRiME™ 8-16% precast gels were purchased from SERVA. The monoclonal M2 anti-FLAG primary antibody was purchased from Sigma-Aldrich. The goat anti-mouse IgG secondary antibody DyLight^TM 800 conjugate was from Thermo-Fisher (SA5-35521). Plasmids and primers used in this study are listed in Supplementary Table 1. Primers were synthesized by Eurofins Genomics.

Bacterial strains, plasmids and growth condition

Bacterial strains and plasmids used in this study are listed in Supplementary Table 1. Unless otherwise specified, bacteria were grown at 37°C in liquid medium shaken at 180 rpm under low aeration. Cultures were supplemented with 1 mM arabinose and 1 mM rhamnose to induce expression of pBAD24-hilA and pT10 plasmid derivatives, respectively (Wagner et al., 2010).

In silico analysis of export apparatus synteny

The Cblaster program was used to identify homologs of export apparatus subunits and to visualize the genetic organization of the corresponding genes (Gilchrist et al., 2021). The search module was run with the following parameters: a minimum identity of 15.0%, minimum coverage of 25.0%, maximum E-value of the BLAST hit of 0.01, and a maximum distance of 20,000 bp between any two hits in a cluster. The BLAST search was performed against the NCBI refseq_protein database using the following query sequences from S. Typhimurium strain LT2: NP_461808.1, NP_461809.1, NP_461810.1, and NP_461811.1. Search results were filtered using the list of reference bacterial genomes from NCBI, and the resulting data were further processed to visualize the distribution of genetic organization patterns. Taxonomic information of each hit was retrieved from NCBI Taxonomy database and visualized as a taxonomic tree using Interactive Tree of Life (iTOL) v7.2.1 (Letunic & Bork, 2024). A comprehensive list of BLAST hits, including BLAST scores, genome annotations, and taxonomic assignments, is provided in Supplementary Table 2.

Secretion assay

Secretion assay was performed as previously described by Westerhausen et al. (Westerhausen et al., 2020). S. Typhimurium strains with the SipA-NL background were grown at 37°C for 5 h in LB medium supplemented with 0.3 M NaCl and appropriate antibiotics for plasmid selection under low aeration. Culture supernatants were collected by centrifugation at 10,000 × g for 2 min. Nano-Glo® luciferase substrate and Nano-Glo® luciferase assay buffer (Promega) were mixed at 1:50 ratio and then combined 1:1 with the culture supernatant. After a brief incubation, luminescence was measured using a microplate reader (Tecan Spark).

In vivo photocrosslinking

S. Typhimurium strains were grown at 37°C for 5 h in LB medium supplemented with 0.3 M NaCl and appropriate antibiotics for plasmid selection under low aeration. Cultures were supplemented with the artificial amino acid para-benzoyl phenyl alanine (pBpa) to a final concentration of 1 mM. 2 OD units of bacterial cells were harvested by centrifugation at 4,000 × g for 5 min and washed once with 1 mL 1x PBS. Cells were resuspended in 1 mL 1x PBS and transferred into 6-well cell culture dishes. UV irradiation (λ = 365 nm) were performed for 30 min at 4°C on a UV transilluminator table. Cells were then pelleted by centrifugation for 2 min at 10,000 × g and 4°C. The cell pellets were either immediately subjected to SDS-PAGE or stored at -20°C until further use.

SDS-PAGE and Immunoblotting

For protein detection, samples were subjected to SDS-PAGE using SERVAGel™ TG PRiME™ 8–16% 12-sample-well precast gels (SERVA Electrophoresis) and transferred onto an Immun-Blot® polyvinylidene difluoride (PVDF) membrane (0.2 µm, Bio-Rad). Membranes were blocked with 1x Blue Block PF (SERVA electrophoresis), then probed with Monoclonal M2 anti-FLAG antibody (1:5,000 in TBS-T, Sigma-Aldrich), followed by goat anti-mouse IgG DyLight 800 conjugate (1:10,000 in TBS-T, Thermo Pierce). Scanning and image analysis was performed with a Li-Cor Odyssey scanner and Image Studio software (Li-Cor, v5.2).

Crude membrane preparation

Crude membranes were prepared as reported previously (Zilkenat et al., 2024). 8 OD units of S. Typhimurium cultures were resuspended in 750 μl lysis buffer consisting of buffer K (50 mM triethanolamine, 250 mM sucrose, 1 mM EDTA, pH 7.5) supplemented with 1 mM MgCl₂, 10 μg/ml DNAse I, 2 mg/ml lysozyme, 1:100 protease inhibitor cocktail (P8849, Sigma-Aldrich) and 2 mM EDTA. After incubation at 4°C for 30 min, samples were mixed with glass beads (150-212 µm) and lysed by bead milling for 2 minutes (SpeedMill Plus, Jena Analytics). Cell lysates were centrifuged at 1,000 × g for 1 min at 4°C, and the supernatant was collected. The glass beads were resuspended with 1 ml buffer K, centrifuged at 1,000 × g for 1 min at 4°C, and the supernatant was pooled. Cellular debris were removed by centrifugation at 10,000 × g for 10 min at 4°C. Crude membranes were pelleted by centrifugation at 186,007 × g for 45 min at 4°C using a Beckman TLA55 rotor. The resulting membrane pellets were directly processed for blue native PAGE.

Blue native PAGE

Blue native PAGE of the crude membrane samples were performed as previously described (Zilkenat et al., 2024). Crude membrane pellets were resuspended in 1% LMNG in 1x PBS and solubilized for 1 h at 4°C with gentle shaking. The concentration of extracted membrane protein was determined using a BCA assay (Pierce™ BCA Protein Assay Kits, Thermo Fisher Scientific) following the manufacturer’s instructions. Solubilized membrane samples were mixed with the loading buffer (10×, 5% Serva Blue G in 250 mM aminocaprioic acid, 50% (w/v) glycerol) and subjected to blue native PAGE using 3-12% Bis-Tris gel (Invitrogen™ NativePAGE™). After electrophoresis, the gel was processed for immunobloting.

Fluorescence measurements

S. Typhimurium strains were grown at 37°C for 5 h in LB medium with low aeration. 1 OD unit of bacterial cells was harvested by centrifugation at 5,000 × g for 2 min at 4°C. Pelleted cells were resuspended in 100 µl PBS and transferred to a black 96-well plate (clear bottom). Fluorescence was measured using a microplate reader (Tecan Spark). For sfGFP, excitation was set to 485 ± 10 nm and emission to 510 ± 10 nm. For mCherry, excitation was set to 587 ± 10 nm and emission to 610 ± 10 nm. Measurements were acquired in top-read mode with 100 gain, 30 flashes, and orbital shaking (2.5 mm, 216 rpm, 5 seconds) before each read. Fluorescence intensities were blank subtracted (PBS only) and the ratio between sfGFP and mCherry were obtained.

Growth assay

S. Typhimurium strains were grown for 9 h at 37°C in LB medium supplemented with 0.3 M NaCl in a 96-well plate inside a humidity cassette. Bacterial growth was monitored by measuring OD₆₀₀ every 5 min following brief shaking, using a microplate reader (Tecan Spark).

mRNA structure predictions

All mRNA structure predictions were performed using SPOT-RNA2 with default parameters (J. Singh et al., 2021). The sequences for mRNA predictions are listed in the Supplementary Table 3, and the predicted structures were visualized using VARNA (v3-93) (Darty et al., 2009).

AlphaFold2 predictions

Alphafold2 was used for structure predictions of SctT oligomers (Bryant et al., 2022). The list of the input sequences are listed in Supplementary Table 3.

Statistical analysis

Statistical analysis for all graphs was performed in GraphPad Prism v10.1.1. The statistical test used in each graph is indicated in the figure legend.

Data availability

Source data for all figures are in the file "Raw_data_Kim_T3SS assembly_2026.xlsx" and in further supplementary tables and figures.

Acknowledgements

We thank Libera Lo Presti for critical review of the manuscript. Bioinformatic analysis was supported by the de.NBI Cloud at the University of Tübingen within the German Network for Bioinformatics Infrastructure (de.NBI).

Additional information

Funding

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, grant WA3299/5-1), by MD stipends of the German Center for Infection Research (DZIF, grants TI07.003_Forberger and TI07.003_Weichel) and by infrastructural support of the cluster of excellence EXC2124 Controlling Microbes to Fight Infections (CMFI, project ID 390838134).

Author contributions

Eunjin Kim: Methodology, Investigation, Data curation, Visualization, Writing - original draft, Writing - review and editing;

Mirjam Forberger, Felix Weichel: Investigation, Funding acquisition; Claudia Paroll, Jialin Zhou, Iwan Grin: Investigation;

Samuel Wagner: Conceptualization, Methodology, Data curation, Visualization, Supervision, Writing - original draft, Writing - review and editing, Project administration, Funding acquisition.

Funding

Deutsche Forschungsgemeinschaft (DFG) (WA3299/5-1)

• Samuel Wagner

Deutsches Zentrum für Infektionsforschung (DZIF) (TI07.003_Forberger)

• Mirjam Forberger

Deutsches Zentrum für Infektionsforschung (DZIF) (TI07.003_Weichel)

• Felix Weichel

Deutsche Forschungsgemeinschaft (DFG) (EXC2124 Project ID 390838134)

• Samuel Wagner

Additional files

Supplementary Table 1. List of primers, plasmids and strains.

Supplementary Table 2. List of hits from in silico synteny analysis.

Supplementary Table 3. List of sequences for mRNA and protein structure predictions.

Source data 1