The Role of the Membrane-Associated Domain of the Export Apparatus Protein, EscV (SctV), in the Activity of the Type III Secretion System

Diarrheal diseases remain a major public health concern worldwide. Many of the causative bacterial pathogens that cause these diseases have a specialized protein complex, the type III secretion system (T3SS), which delivers effector proteins directly into host cells. These effectors manipulate host cell processes for the benefit of the infecting bacteria. The T3SS structure resembles a syringe anchored within the bacterial membrane, projecting toward the host cell membrane. The entry port of the T3SS substrates, called the export apparatus, is formed by five integral membrane proteins. Among the export apparatus proteins, EscV is the largest, and as it forms a nonamer, it constitutes the largest portion of the export apparatus complex. While there are considerable data on the soluble cytoplasmic domain of EscV, our knowledge of its membrane-associated section and its transmembrane domains (TMDs) is still very limited. In this study, using an isolated genetic reporter system, we found that TMD5 and TMD6 of EscV mediate strong self-oligomerization. Substituting these TMDs within the full-length protein with a random hydrophobic sequence resulted in a complete loss of function of the T3SS, further suggesting that the EscV TMD5 and TMD6 sequences have a functional role in addition to their structural role as membrane anchors. As we observed only mild reduction in the ability of the TMD-exchanged variants to integrate into the full or intermediate T3SS complexes, we concluded that EscV TMD5 and TMD6 are not crucial for the global assembly or stability of the T3SS complex but are rather involved in promoting the necessary TMD–TMD interactions within the complex and the overall TMD orientation to allow channel opening for the entry of T3SS substrates.


INTRODUCTION
Diarrheal diseases are a major global health concern and are considered the second leading cause of death in children under the age of five. According to the World Health Organization (WHO), there are nearly 1.7 billion cases of childhood diarrheal disease per year with an estimated 500,000 deaths annually. One of the main infectious agents of pediatric diarrhea is enteropathogenic Escherichia coli (EPEC; Clarke et al., 2002). This pathogen was related to a series of outbreaks of infantile diarrhea in the 1940s and 1950s (Robins-Browne, 1987). While EPEC is no longer considered to be an important cause of acute diarrhea in many countries, there has been a recent reemergence with severe disease outcomes being associated with EPEC infections (Croxen et al., 2013).
Enteropathogenic E. coli belongs to a family of bacteria that form a distinctive histological lesion in the intestinal epithelium, collectively called attaching and effacing (A/E) pathogens (Goosney et al., 2000). In the A/E lesion, the bacteria tightly attach to the host's intestinal epithelial cells, causing a disruption of the brush border microvilli and promoting formation of actin pedestals that elevate the pathogen above the epithelial cell. This morphology is mediated by a protein transport nanomachine termed the type III secretion system (T3SS; Buttner, 2012;Deng et al., 2017;Wagner et al., 2018). The T3SS delivers virulence factors directly into host cells, and these manipulate the host cell cytoplasm rearrangement. The injected effectors also interfere with and modify critical cellular pathways to improve bacterial survival and replication (Bhavsar et al., 2007). The core architecture of the T3SS consists of a basal body embedded within the bacterial membranes, a periplasmic inner rod, a transmembrane export apparatus, and a cytosolic platform, which includes an ATPase complex and the C-ring. In addition, a distinct hollow needle is assembled on the extracellular face of the basal body, which is linked in A/E pathogens to an extracellular long filament, and a pore complex at the host membrane to create a channel for protein secretion (Buttner, 2012).
The T3SS structural genes are encoded within the bacterial chromosome on a large 35-kbp genomic pathogenicity island called the locus of enterocyte effacement (LEE). The LEE is organized into seven operons (LEE1-LEE7) that encode structural proteins, as well as regulators and several protein effectors (Elliott et al., 2000;Deng et al., 2004;Franzin and Sircili, 2015;Gaytan et al., 2016). The export apparatus, which is found at the center of the inner membrane ring and facing the cytoplasmic side, is among the most conserved substructures within the T3SS complex. This structure is essential for secretion and acts as the entry portal for the T3SS substrates. The export apparatus is assembled from five highly conserved membrane proteins, named EscR, EscS, EscT, EscU, and EscV, which were shown to form a multimeric protein complex with a stoichiometry of 5:1:4:1:9, respectively, in the homologous T3SS of Salmonella typhimurium (Kuhlen et al., 2018). The complexity of this structure is illustrated by the estimation that a total of 104 transmembrane domains (TMDs) are involved in its formation (Zilkenat et al., 2016). Among the export apparatus components, EscV, which is named SctV according to the T3SS unified nomenclature (Wagner and Diepold, 2020), is the largest protein (72 kDa), and because it forms a nonamer, it constitutes the largest portion of the export apparatus complex.
EscV is divided into two large domains: an N-terminal region with seven to eight predicted TMDs and a C-terminal cytoplasmic domain (Wagner et al., 2010;Abrusci et al., 2013). The presence of a putative N-terminal cleavable signal sequence suggests that EscV is directed to the inner membrane through the sec pathway (Garmendia et al., 2005), and it was found that its membrane localization was independent of the T3SS (Gauthier et al., 2003). EscV and its homologs in Salmonella and Shigella (InvA and MxiA, respectively) were shown to oligomerize and form a cytoplasmic homo-nonameric ring that is located directly below the secretion pore and above the ATPase complex (Abrusci et al., 2013;Bergeron et al., 2013;Majewski et al., 2020).
EscV and its homologs in both the virulent and flagellar T3SSs have been implicated in the recruitment of T3SS substrates, chaperones, and proteins from the "gatekeeper" family of proteins to the T3SS apparatus as part of the regulation process of hierarchical secretion of T3SS substrates (Diepold et al., 2012;Minamino et al., 2012;Abrusci et al., 2013;Kinoshita et al., 2013;Portaliou et al., 2017). The binding between EscV and various T3SS cargo proteins was shown to occur via EscV's cytoplasmic C-terminus (Minamino et al., 2012;Gaytan et al., 2016;Shen and Blocker, 2016;Portaliou et al., 2017). Mutations in two amino acid residues located on the surface of MxiA, the Shigella SctV homolog, were shown to lead to twofold to threefold increased secretion of the IpaH effector compared to the wild-type (WT) strain (Shen and Blocker, 2016).
Overall, these studies indicated that the SctV family of proteins is part of the export gate complex where it forms an IM pore, which is required for the assembly and proper function of the T3SS, and acts as a substrate selection checkpoint. Nevertheless, although the EscV is an integral membrane protein that contributes more than half of the TMDs of the export apparatus, most of the available information about this protein is related to its soluble domain. Therefore, in this study, we investigated the role of EscV TMDs in protein function and their involvement in global T3SS assembly and activity.

Bacterial Strains
WT EPEC O127:H6 strain E2348/69 (streptomycin-resistant strain) and EPEC-null mutants ( escN and escV; Gauthier et al., 2003) were used to evaluate the T3SS and translocation activities ( Table 1). E. coli BL21 strain (λDE3) and E. coli DH10B were used for protein expression and for plasmid handling, respectively (Table 1). E. coli FHK12, which encodes β-galactosidase under the control of the ctx promoter, and E. coli PD28 (a malE-deficient E. coli strain) were used for assessment of TMD oligomerization ( Table 1). The strains were grown at 37 • C in a Luria-Bertani (LB) broth or Dulbecco's Modified Eagle Medium (DMEM) with or without IPTG in the presence of the appropriate antibiotics. Antibiotics were used at the following concentrations: carbenicillin (100 µg/ml), streptomycin (50 µg/ml), and chloramphenicol (30 µg/ml). Bacterial growth was recorded at 600 nm every 30 min on the Infinite 200 PRO multimode plate reader (Tecan Group Ltd., Switzerland).

Bioinformatics
FASTA sequence for the EscV protein was downloaded from GenBank on the NCBI and UniProt database. To predict the TMD sequences, we used the TMHMM software 1 . ClustalW multisequence alignment algorithm was used to identify highly conserved protein domains by comparing sequences of EscV orthologs 2 (no changes were made in the default parameters of the server; Larkin et al., 2007).

Construction of Plasmids Expressing Labeled EscV
EscV-expressing plasmids were constructed using the primers presented in Table 2. The pSA10 plasmid was amplified using the primer pair pSA10_F/pSA10_R ( Table 2). The escV gene was amplified from EPEC genomic DNA using the primer pairs EscV_SA10_F/EscV_His_R1 and then EscV_SA10_F/EscV_His_R2, which fused a His tag to the coding region of EscV ( Table 2). The polymerase chain reaction (PCR) products were subjected to digestion with DpnI, purified, 1 https://www.expasy.org/resources/detail/TMHMM 2 https://www.ebi.ac.uk/Tools/msa/clustalo/ and assembled by the Gibson assembly method (Gibson et al., 2008(Gibson et al., , 2009). An HA-labeled version of EscV was similarly cloned into pSA10; here, the double-HA tag was fused to the coding region of escV by amplification from EPEC genomic DNA using the primer pairs EscV_SA10_F/EscV_HA_R1 and then EscV_SA10_F/EscV_HA_R2 ( Table 2). To clone EscV labeled with the V5 tag, we amplified the pEscD-V5 (pSA10) vector (Tseytin et al., 2018a) using the primer pair pSA10_V5_F/pSA10_R and the EPEC genomic DNA with EscV_SA10_F/EscV_V5_R ( Table 2). The PCR products were subjected to digestion with DpnI, purified, and assembled by the Gibson assembly. To clone EscV-V5 into a low-copynumber plasmid (pACYC184), we amplified the pEscP-V5 (pACYC184) vector (Shaulov et al., 2017) using the primer pair pSA10_V5_F/pACYC_R and the EPEC genomic DNA with EscV-V5_184_F/EscV-V5_184_R ( Table 2). The PCR products were subjected to digestion with DpnI, purified, and cloned using the Gibson assembly. The resulting constructs, pEscV wt -V5 (in pSA10 and pACYC184), pEscV wt -HA, and pEscV wt -His expressed a full-length EscV protein with V5, HA, or His tag fused to its C-terminal, respectively. The TMD5-exchanged and TMD6-exchanged escV in pSA10 were generated by using the template of pEscV wt -His (pSA10). To replace the TMD5 of EscV by a TMD backbone sequence of 7-leucine-9-alanine (7L9A), the EscV 222-675 amino acid sequence was amplified by using the primer pair EscV_Fc5_F/EscV_His_R2 (Table 2) from the pEscV wt -His vector. The TMD 7L9A backbone was generated by annealing the primer pair 7L9A-F/7L9A-R ( Table 2) by heating the sample to 95 • C for 5 min and then decreasing the temperature to 20 • C at a rate of 5 • C/min. Then the 7L9A backbone was amplified by the primer pair EscV_TMD5-7L9A_F/EscV_TMD5-7L9A_R ( Table 2). The EscV 222−675 PCR fragment and the 7L9A backbone were then ligated by using overlapping sequences and amplified by using the primer pair EscV_TMD5-7L9A_F/EscV_His_R2 ( Table 2). The Gibson assembly was conducted by amplifying the pEscV wt -His pSA10 vector with the primer pair pSA10_F/EscV_7L9A_Gib5_R ( Table 2), followed by treating the reaction with DpnI and subjecting the amplified vector and the 7L9A-EscV 222−675 -fused PCR fragment to ligation. The resulting construct, pEscV-TMD5 ex -His (pSA10), expressed a TMD5-exchanged EscV with a His tag at its C-terminus. The TMD6-exchanged escV in pSA10 was generated by a similar manner. The TMD 7L9A backbone was amplified by the primer pair EscV_TMD6-7L9A_F/EscV_TMD6-7L9A_R and fused to the EscV 259−675 PCR fragment, which was amplified by using the primer pair EscV_Fc6_F/EscV_His_R2 ( Table 2). The Gibson assembly was conducted by amplifying the pEscV wt -His pSA10 vector with the primer pair pSA10_F/EscV_7L9A_Gib6_R (Table 2), followed by treating the reaction with DpnI and subjecting the amplified vector and the 7L9A-EscV 259−675 -fused PCR fragment to ligation. The resulting construct, pEscV-TMD6 ex -His (pSA10), expressed a TMD6-exchanged EscV with a His tag at its C-terminus. To move the TMD5-exchanged and TMD6-exchanged EscV into pSA10 that fuses the proteins with the V5 tag, pEscV-TMD5 ex -His and pEscV-TMD6 ex -His were amplified using the primer pair EscV_SA10_F/EscV_V5_R, and the pEscV wt -V5 vector was amplified by the primer pair pSA10_V5_F/pSA10_R ( Table 2). The PCR products were treated with DpnI and subjected to Gibson assembly.

Detection of the Homo-Oligomerization of TMDs Within the Membrane
The ToxR transcription activator can be successfully used to detect TMD-TMD interactions (Brosig and Langosch, 1998). DNA cassettes, encoding single TMDs of EscV (TMD1-TMD7.2), glycophorin A (GpA) TMD, E. coli aspartate receptor N-terminal TMD (Tar-1), 16-alanine backbone (A16), and no TMD ( TMD), were grafted between the cytoplasmic domain of the ToxR transcription activator protein (an oligomerizationdependent transcriptional activator) and the periplasmic domain of the maltose-binding protein (MBP). This was performed by aligning the oligonucleotides pairs encoding a NheI-BamHI TMD-DNA cassette of 16 core residues of the EscV TMD1-TMD7 ( Table 2). The double-stranded fragments were oligophosphorylated and ligated between the toxR transcription activator and the malE (encodes E. coli MBP) within a NheI-BamHI digested ToxR-MBP plasmid ( Table 2).
The MBP domain directs the chimera protein to the periplasm (Salema and Fernandez, 2013) where the TMD becomes embedded within the inner membrane. In the assay, we transform ToxR-TMD-MBP plasmids containing different TMDs into E. coli FHK12 cells, which contain a reporter gene, coding for β-galactosidase, under the control of the ctx promoter (Kolmar et al., 1995). Oligomerization of the investigated TMD derives ToxR oligomerization, which then can bind (in its oligomeric form) the ctx promoter and transcribe the reporter gene lacZ (Ottemann et al., 1992;Brosig and Langosch, 1998). Determination of the oligomerization level is calculated from the activity of β-galactosidase that is measured by the levels of a yellow color (OD 405 ) produced from the cleavage of the o-nitrophenylgalactose (ONPG), β-galactosidase substrate (Langosch et al., 1996;Fink et al., 2012). We monitored the activity of β-galactosidase for 20 min, at intervals of 30 s and calculated the V max of the reaction. These were normalized to the original cell content (measured at OD 600 ) and are presented in Miller units. We used the GpA TMD sequence as a positive control for strong homo-oligomerization (Russ and Engelman, 2000), the N-terminal TMD of the E. coli aspartate receptor (Tar-1) as a reference for moderate oligomerization (Sal-Man et al., 2004), and the A16 sequence as a control for non-oligomerizing sequences (Bronnimann et al., 2013).

Maltose Complementation Assay
Membrane localization and correct orientation of the chimera proteins were examined as described previously (Langosch et al., 1996). Briefly, we transformed PD28 cells, E. coli strain deficient for malE (Duplay and Szmelcman, 1987), with the different ToxR-TMD-MBP constructs described above. The bacteria were grown overnight, washed, and then inoculated into M9 minimal medium supplemented with 0.4% maltose. Bacterial growth (OD 600 ) was measured over time. Since PD28 cells are unable to utilize maltose, only strains that translocate the chimera protein across the inner membrane were expected to grow in this medium. A construct without TMD ( TMD) served as a negative control. This chimera protein is expected to reside in the bacterial cytoplasm and, therefore, cannot grow in minimal medium.

In vitro Type 3 Secretion Assay
Type 3 secretion assays were performed as previously described (Tseytin et al., 2018a(Tseytin et al., ,b, 2019. Briefly, overnight EPEC strains grown in LB were diluted 1:40 into preheated DMEM and grown for 6 h at 37 • C in an atmosphere of 5% CO 2 , statically, to an optical density of 0.7 (OD 600 ). Protein expression was induced by 0.1 mM IPTG. To separate bacterial cells from bacterial supernatants, we centrifuged the cultures at 20,000 × g for 5 min; the bacterial pellets were dissolved in an sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer, and the supernatants, containing secreted proteins, were collected and filtered through a 0.22-µm filter (Millipore). The supernatants were normalized according to the bacterial OD 600 and precipitated with 10% (v/v) trichloroacetic acid (TCA) overnight at 4 • C. To concentrate the secreted proteins, the supernatant samples were centrifuged at 20,000 × g for 30 min at 4 • C and resuspended in the SDS-PAGE sample buffer. To neutralize the residual TCA, we added saturated Tris.

Translocation Activity
Translocation assays were performed as previously described (Baruch et al., 2011). Briefly, EPEC strains were pre-induced for 3 h for T3SS activity (preheated DMEM, statically, in a CO 2 tissue culture incubator) and then were used to infect HeLa cells (8 × 10 5 cells per well) at a multiplicity of infection (MOI) of 1:300 for 3 h. Cells were then washed and lysed with RIPA buffer [10 mM Tris-Cl pH 8.0, 1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 140 mM NaCl; before use, add 1 mM phenylmethylsulfonyl fluoride (PMSF)]. Lysates were centrifuged at maximum speed for 5 min to remove non-lysed cells and supernatants containing cellular proteins, were collected, mixed with the SDS-PAGE sample buffer, and analyzed by western blot analysis with anti-JNK and anti-actin antibodies (loading control). Uninfected HeLa cell samples, and HeLa cells infected by EPEC escN and EPEC escV mutant strains were used as negative controls.

Bacterial Fractionation
Bacterial cell fractionation was performed as previously described (Gauthier et al., 2003). Briefly, overnight EPEC cultures were subcultured at 1:50 in DMEM and grown for 6 h, at 37 • C, in a CO 2 tissue culture incubator. Bacteria were collected, washed, and resuspended in buffer A [50 mM Tris (pH 7.5), 20% (w/v) sucrose, 5 mM EDTA, protease inhibitor cocktail (Roche Applied Science), and lysozyme (100 µg/mL)] to generate spheroplasts. The samples were incubated for 15 min, at room temperature, and while rotating, MgCl 2 (20 mM) was then added. The samples were spun for 10 min at 5,000 × g, and the supernatants, containing the periplasmic fractions, were collected. The pellets were resuspended in lysis buffer (20 mM Tris/HCl pH 7.5, 150 mM NaCl, 3 mM MgCl 2 , 1 mM CaCl 2 , and 2 mM β-mercaptoethanol with protease inhibitors) and kept at 4 • C from this step on. The samples were sonicated (Thermo Fisher Scientific, 3 × 10 s) after addition of RNase A and DNase I (10 µg/ml) and centrifuged (2,300 × g for 15 min) to remove intact bacteria, and the supernatants, containing cytoplasmic and membrane proteins, were collected. To separate the cytoplasmic and membrane fractions, the samples were centrifuged (in a Beckman Optima XE-90 Ultracentrifuge with a SW60 Ti rotor) for 30 min at 100,000 × g. The supernatants, containing the cytoplasmic fraction, were collected, and the pellets, containing the membrane fractions, were washed with lysis buffer, and the final pellets were resuspended in lysis buffer with 0.1% SDS. We evaluated the protein content by Coomassie Plus protein assay. We used various proteins as membrane (Intimin), periplasm (MBP), and cytoplasm (DnaK) markers.

Crude Membrane
We isolated the bacterial membrane fraction using the bacterial fractionation protocol described above. To extract membrane proteins from the crude membranes, samples were resuspended in lysis buffer containing 1% n-dodecyl-β-D-maltoside (DDM) and incubated for 60 min on a rotary wheel, at 4 • C. The samples were then centrifuged (20,000 × g, for 15 min, at 4 • C) to remove non-solubilized material, and the supernatants, containing membrane proteins, were collected and analyzed by blue-native (BN)-PAGE.

Blue-Native-PAGE
We incubated the extracted membrane proteins for 15 min in a BN sample buffer (30% glycerol with 0.05% Coomassie Brilliant Blue G250), and then loaded the samples onto a Criterion XT Tris-Acetate 3-8% precast gradient native gel (Bio-Rad). Electrophoresis was carried out on ice using a cathode buffer (15 mM Bis-Tris and 50 mM bicine, pH 7) and anode buffer (50 mM Bis-Tris, pH 7) until full separation (5-6 h). The gel was then incubated in a transfer buffer and subjected to western immunoblotting.

EscV TMD5 and TMD6 Support TMD-TMD Interactions
As TMDs are known to be involved in protein-protein interactions, we examined the ability of isolated EscV TMDs to support self-interaction. For that purpose, we employed the ToxR assembly system (Figure 2A), which monitors the strength of TMD-TMD interactions within the bacterial inner membrane (Langosch et al., 1996;Joce et al., 2011). We compared the oligomerization level of EscV TMDs with that of GpA's TMD sequence, which contains a GxxxG motif and is used as a reference for strong homo-oligomerization (Lemmon et al., 1992;Adair and Engelman, 1994;Russ and Engelman, 2000). We also compared the EscV TMD oligomerization levels with the N-terminal TMD of the E. coli aspartate receptor (Tar-1), which has moderate oligomerization (Sal-Man et al., 2004), and polyalanine's (A16) sequence as a non-oligomerizing sequence (Langosch et al., 1996;Sal-Man et al., 2005). Since the amino acid sequence of TMD7 was significantly longer than that of the other TMDs and the ToxR system responded differently to various TMD lengths (Langosch et al., 1996), we decided to test two different forms of this TMD, TMD7.1, and TMD7.2, which are of a similar length as the other examined TMDs. The sequences of the TMDs studied are presented in Figure 2A. We observed strong TMD self-oligomerization activity for EscV's TMD5, TMD6, and TMD7.2 compared to the activities of the GpA and Tar-1 TMDs, whereas EscV's TMD1, TMD2, TMD3, TMD4, and TMD7.1 showed reduced oligomerization activities compared to GpA ( Figure 2B). As expected, the oligomerization of the A16 background control was low ( Figure 2B). These findings suggested that TMD5, TMD6, and TMD7.2 of EscV might be involved in the oligomerization of the full-length protein EscV, through TMD-TMD interactions. To exclude the possibility that the high self-oligomerization activity of EscV's TMD5, TMD6, and TMD7.2 resulted from higher expression levels of these chimera proteins, we subjected the bacterial samples to SDS-PAGE and western immunoblotting analysis with an anti-MBP antibody. All samples showed comparable expression levels ( Figure 2B). To verify that the ToxR-TMD-MBP chimera proteins correctly integrated into the inner membrane, we employed the maltose complementation assay. For that purpose, we used an E. coli strain with a malE gene (PD28) deletion, which cannot produce endogenous MBP and therefore cannot support bacterial growth in minimal medium with maltose as the sole carbon source (Langosch et al., 1996). Only strains that express the chimera protein ToxR-TMD-MBP and orient it across the inner membrane, with MBP facing the periplasm, will support bacterial growth. We observed that all examined strains demonstrated bacterial growth, which indicated proper membrane integration, while the negative control that did not contain a TMD ( TMD) showed no growth, as expected ( Figure 2C). Overall, these results suggest that TMD5, TMD6, and TMD7.2 of EscV are involved in EscV self-oligomerization through TMD-TMD interactions. However, due to the high conservation of TMD6 and the GxxxG motif within TMD5, on one hand, and the unclear boundaries of TMD7, on the other, we decided to focus on EscV's TMD5 and TMD6.

Replacement of EscV TMDs With a Non-oligomerizing Sequence (7L9A) Affects Bacterial Fitness
To examine whether EscV's TMD5 and TMD6 serve solely as membrane anchors or have a functional role within the fulllength protein, we constructed EscV mutant proteins lacking TMD5 or TMD6 sequences. Since EscV lacking its TMD5 or TMD6 will likely adopt alternate protein folding compared to the native protein or have impaired localization, we constructed TMD5-and TMD6-exchanged EscV proteins, where the native core TMD5 and TMD6 sequences (16 amino acids in length) were replaced by a hydrophobic sequence. We chose a hydrophobic sequence of seven consecutive leucine residues followed by nine alanine residues (7L9A), which was previously shown to be sufficiently hydrophobic to support protein integration into the membrane yet cannot support TMD-TMD interactions (Sal-Man et al., 2005). To determine the biological effect of this replacement, we transformed the TMD5-and TMD6exchanged EscV (EscV-TMD5 ex -His and EscV-TMD6 ex -His), as well as EscV wt -His, into the escV-null strain ( escV) and examined their ability to restore T3SS activity. However, when EscV overexpression was induced by addition of IPTG to a concentration of 0.25 mM, growth rate was reduced in all strains (Figure 3). To determine the conditions that allow EscV expression without severe reduction of bacterial fitness, we grew WT EPEC, EPEC escV, and EPEC escV carrying either pEscV wt -His, pEscV-TMD5 ex -His, or pEscV-TMD6 ex -His in LB or in DMEM (which is used for T3SS-inducing conditions), in the presence (0.1 or 0.25 mM) or the absence of IPTG. Optical density at 600 nm was measured over time (Figure 3). We observed that expressions of EscV WT and TMD-exchanged versions have reduced fitness when induced with IPTG at a concentration higher than 0.1 mM (Figure 3). These results suggest that overexpression of EscV is toxic to bacteria and therefore negatively affects bacterial growth. Based on these results, we used 0.1 mM IPTG for our experiments.

TMD5 and TMD6 Are Critical for EscV Activity
To examine whether EscV TMD5 and TMD6 sequences are critical for the activity of the full-length protein, we examined whether EscV-TMD5 ex -His and EscV-TMD6 ex -His can restore the T3SS activity of the EPEC escV strain. Only functional EscV can restore the T3SS of the escV strain, which is measured by the ability of EPEC strains to secrete three T3SS translocators (EspA, EspB, and EspD) into the culture supernatant, when grown under T3SS-inducing conditions. First, we evaluated the ability of WT EscV to restore the T3SS activity of escV. We observed that the expression of EscV wt -His within the escV strain restored secretion of translocators but also resulted in hypersecretion of effectors (Tir and NleA; Figure 4A). To evaluate whether this phenotype occurs due to the labeling of EscV or to its expression from a plasmid, we examined the T3SS activity of escV-carrying plasmids with unlabeled EscV or EscV labeled with various tags and expressed from low-and high-copy-number plasmids. We observed that transformation of unlabeled EscV wt resulted in a milder phenotype and that only a slight elevation in effector secretion was observed. In contrast, expression of labeled EscV, regardless of the tag type, resulted in hypersecretion of effectors (Supplementary Figure 2). Interestingly, expressions of both EscV-TMD5 ex -His and EscV-TMD6 ex -His failed to complement the T3SS activity of the escV strain and demonstrated a secretion profile similar to that of escV and escN (Figure 4A). Comparable protein expression of the WT and the exchanged versions was observed by analyzing whole-cell lysates by western blot analysis using anti-His antibody (Figure 4A).
To analyze whether the unregulated secretion of escV transformed with pEscV wt -His affected the ability of the bacteria to infect host cells, we examined the ability of the strain to infect and translocate effectors into the HeLa cells. For this FIGURE 4 | Replacement of EscV TMD5 and TMD6 by an alternative hydrophobic sequence abolishes T3SS activity. (A) Protein secretion profiles of EPEC WT, escV, escN and EPEC escV strain carrying the pEscV wt -His, pEscV-TMD5 ex -His, or pEscV-TMD6 ex -His plasmids grown under T3SS-inducing conditions with 0.1 mM IPTG. The secreted fractions were filtered and protein content was concentrated from the supernatants of bacterial cultures and analyzed by SDS-PAGE and Coomassie blue staining. The T3SS-secreted translocators and effectors EspA, EspB, EspD, NleA, and Tir are marked on the right of the gel. EspC, which is not secreted via the T3SS, is also indicated. EscV expression was examined by analyzing bacterial pellets by SDS-PAGE and western blot analysis with an anti-His antibody. Numbers on the left are molecular masses in kilodaltons. (B) Proteins extracted from HeLa cells infected with WT, escN, escV, or escV carrying the pEscV wt -His, pEscV-TMD5 ex -His, or pEscS-TMD6 ex -His, were subjected to western blot analysis using anti-JNK antibody and anti-actin (loading control). JNK and its degradation fragments are indicated.
purpose, we infected HeLa cells with WT, escN, escV, and escV transformed with pEscV wt -His and examined the cleavage pattern of JNK, a cellular protein that is cleaved by NleD, a translocated EPEC effector (Baruch et al., 2011). WT EPEC induced extensive degradation of JNK, as expected, relative to the uninfected sample and to the samples infected with escN or escV mutant strains ( Figure 4B). EPEC escV transformed with the plasmid encoding EscV wt -His showed a JNK degradation profile, indicating functional complementation by His-labeled EscV (Figure 4B). In addition, the escV strain transformed with EscV TMD-exchanged versions (pEscV-TMD5 ex -His or pEscV-TMD6 ex -His) showed no degradation of JNK, as observed for the uninfected sample ( Figure 4B). Overall, our results suggest that His-labeled EscV functionally complements the T3SS activity; however, replacing the native TMD5 or TMD6 sequences of EscV with an alternative hydrophobic sequence (7L9A) impairs the function of the T3SS (Figure 4B).

Transmembrane Domain Replacement Does Not Affect EscV Localization to the Bacterial Membrane
To exclude the possibility that EscV-TMD5 ex -His and EscV-TMD6 ex -His failed to complement the escV T3SS activity due to impaired subcellular localization, we grew the strains under T3SS-inducing conditions and fractionated them into periplasmic, cytoplasmic, and membrane fractions. Our results showed that EscV-TMD5 ex -His and EscV-TMD6 ex -His localized mostly to the membrane fraction, as was seen for EscV wt -His ( Figure 5). Correct bacterial fractionation was confirmed by analyzing the samples with anti-MBP (periplasmic marker), anti-DnaK (cytoplasmic marker), and anti-intimin (membrane marker) antibodies. Overall, our results indicated that replacement of TMD5 and TMD6 did not disrupt EscV localization to the bacterial membrane.

EscV TMD6 Is Involved in Complex Formation
To investigate whether the EscV TMD-exchanged variants fail to complement the T3SS activity of the escV-null strain due to their inability to properly integrate into the T3SS complex, we prepared crude membrane samples of EPEC escV and EPEC escV-null strains transformed with EscV wt -His, EscV-TMD5 ex -His, and EscV-TMD6 ex -His grown under T3SSinducing conditions. The samples were then analyzed by BN-PAGE and immunoblotting. BN-PAGE analysis revealed that EscV wt -His and EscV-TMD5 ex -His preserved the ability to integrate into the T3SS complex, as they migrated primarily as a large complex (>1 MDa) to the top of the gel. However, EscV-TMD6 ex -His integration into the complex appeared to be impaired (Figure 6). To verify that the modified running pattern of the EscV TMD6-exchanged version was not due to reduced protein expression, we analyzed the crude membrane extracts by SDS-PAGE and western blotting using anti-His antibody. Similar expression levels were observed for all EscV variants (Figure 6). These results suggest that TMD5 and TMD6 are not critical for the integration of EscV into the T3SS complex, as EscVexchanged versions enabled the formation of high-molecularweight complexes. EscV-TMD5 ex -His fully preserved the ability to integrate into the full or intermediate T3SS complexes, while integration of EscV-TMD6 ex -His was impaired.

A Single Mutation Within the GxxxG Motif of TMD5 Abolished EPEC T3SS Activity and Complex Formation
To examine whether the GxxxG motif identified within TMD5 is critical for protein activity, we mutated the glycine residues at positions 213 and 217 to either alanine or leucine (G213A, G217A, G213L, and G217L). Due to expression challenges of the mutated proteins tagged with His-tag, we labeled EscV WT and single mutants with the V5 tag, which resulted in a similar secretion profile to EscV wt -His (Supplementary Figure 2). The single mutants were transformed into escV, and their T3SS activity was examined. We observed that mutations G213A and G217A had similar secretion profiles to the escV strain transformed with EscV wt -V5, while the single mutation G213L completely abolished T3SS activity ( Figure 7A). The effect was much milder when the escV strain was transformed with the EscV G217L mutant (Figure 7A). To confirm proper expression of the EscV point mutation variants, whole-cell lysates were submitted to western blot analysis using anti-V5 antibody. Comparable protein expression was detected for the WT and the single mutants ( Figure 7A). Our results suggest that replacement of the glycine residues of the GxxxG motif found in TMD5 by a large residue (leucine) disrupts the activity of the protein while replacement by a small residue (alanine) does not.
To investigate the effect of the single mutation G213L on the assembly of the T3SS complex, we examined the ability of mutant EscV proteins to properly integrate into the T3SS complex. For this purpose, we grew the EPEC WT and EPEC escV strain transformed with EscV wt -V5, EscV G213A -V5, and EscV G213L -V5 under T3SS-inducing conditions. We prepared crude membranes and analyzed them by BN-PAGE and immunoblotting. BN-PAGE analysis showed that the escV mutant strain transformed with EscV wt -V5 and EscV G213A -V5 migrated mainly as a large complex at the top of the gel, while the EscV G213L -V5 integration into the complex appeared to be impaired ( Figure 7B). To confirm that the altered running pattern of the EscV G213L -V5 mutant form was not due to reduced protein expression, the crude membrane extracts were analyzed by SDS-PAGE and immunoblotting using the anti-V5 antibody. We detected a lower expression level of EscV G213L -V5 relative to EscV wt -V5 and EscV G213A -V5, but not to a level that explains the significant reduction in complex formation ( Figure 7B). Overall, our results indicate that the GxxxG motif and more specifically the glycine at position 213 are critical for the proper EscV integration into the T3SS complex.

DISCUSSION
The high conservation of the sequence of EscV TMD6 and the conserved GxxxG motif within TMD5 (Figure 1B), together with the numerous studies regarding TMD-derived oligomerization of membrane complexes (Fink et al., 2012), urged us to examine whether EscV TMDs are involved in protein oligomerization. Results using the isolated ToxR system demonstrated that TMD5 and TMD6 exhibited strong self-oligomerization activities, with activities similar to that of the well-characterized GpA TMD sequence (Figure 2A).
To investigate whether TMD5 and TMD6 sequences are critical for the activity of the full-length protein, we replaced each of these TMDs with an alternative hydrophobic sequence (7L9A). The plasmids encoding TMD5-or TMD6-exchanged EscV versions were transformed into the escV-null strain, and their T3SS activity was examined. We found that expression of either EscV-TMD5 ex -His or EscV-TMD6 ex -His failed to complement the T3SS activity of the EPEC escV strain while the expression of EscV wt -His restored T3SS (Figure 4A). Infection of HeLa cells with bacterial strains that express either TMD5-or TMD6exchanged EscV versions was non-virulent and demonstrated JNK degradation profiles comparable to those of uninfected cells ( Figure 4B). Since we observed that the membrane localization of both WT and TMD-exchanged EscVs was not disrupted (Figure 5), we concluded that EscV's TMD5 and TMD6 are critical not only for proper membrane anchoring but also for T3SS activity and EPEC's ability to infect host cells as they cannot be replaced by an alternative hydrophobic sequence. Based on the ToxR results, we presume that TMD5 and TMD6 are involved in protein oligomerization although we did not detect complete complex dissociation for T3SS with TMDexchanged variants (Figure 6). These results suggest that EscV's TMD5 and TMD6 are not crucial for the global assembly or stability of the T3SS complex but rather that they are involved in promoting the proper TMD-TMD interactions within the complex and their overall orientation to allow passage of T3SS substrates.
To examine the role of the GxxxG motif found within TMD5 on the overall activity of the T3SS, we mutated single glycine residues within the motif, replacing them with either a nonpolar small amino acid (alanine) or a non-polar large amino acid (leucine). We found that the original glycine residues could be replaced by alanine residues with no effect on T3SS activity ( Figure 7A). These results are in agreement with previous reports suggesting that the GxxxG motif is equivalent to the Small-xxx-Small motif (Lock et al., 2014;Curnow et al., 2020;Wang et al., 2020). In contrast, substitution of leucine for glycine at position 213, but not at 217, abolished T3SS activity ( Figure 7A). These results suggested that the two glycine positions do not contribute equally to the activity of the protein and that position 213 is more critical for EscV function within the T3SS complex. Interestingly, while we observed reduced complex formation with the G213L mutation ( Figure 7B), we did not observe a similar reduction for EscV TMD5-exchanged (Figure 6), although both had glycine converted to leucine at position 213. These results suggest that TMD-TMD packing is context-dependent and is not dependent on a single residue or motif. Our results are in agreement with previous reports that demonstrated that the GxxxG motif supports TMD interactions within the context of oligo-methionine and oligo-valine sequences, but not within randomized TMDs (Brosig and Langosch, 1998;Unterreitmeier et al., 2007;Langosch and Arkin, 2009).
Expression of EscV wt -His within the escV null strain unexpectedly resulted in hypersecretion of effectors compared to that seen with WT EPEC (Figure 4A). Interestingly, HA-and V5-tagged EscV expressed from a high copy-number plasmid (pSA10) presented a similar secretion profile, as did expression of HSV-tagged EscV from a low copy-number plasmid (pACYC184, Supplementary Figure 2). A milder phenotype was observed for expression of unlabeled EscV (Supplementary Figure 2). Overall, these results suggested that overexpressing, and to a larger extent, labeling EscV at its C-terminus, regardless of the nature of the tag, interferes with substrate secretion regulation. Our results correlate well with previous studies that indicated that the EscV is involved in substrate secretion regulation through interaction with the "gate-keeper" SepL and several T3SS chaperons (Portaliou et al., 2017;Gaytan et al., 2018). The observation that EscV interacts with SepL via its C-terminal (Portaliou et al., 2017) suggests that labeling EscV at this critical domain disrupts EscV-SepL interaction and therefore induces uncontrolled T3S. This conclusion was further supported by our inability to recapitulate EscV-SepL interaction when EscV was labeled on its C-terminal (data not shown).
Examination of the ability of EPEC escV expressing EscV wt -His to infect HeLa cells revealed a similar infection capacity as the WT EPEC strain (Figure 4B). This result was unexpected as previous studies revealed that strains with dysregulated T3 substrate secretion ( sepL, sepD, and escP) showed reduced infectivity and effector translocation abilities (Deng et al., 2004(Deng et al., , 2015Shaulov et al., 2017). To our knowledge, this is the first example of an EPEC strain that lacks hierarchical substrate secretion regulation but shows similar virulence capabilities to the WT strain. We assume that in contrast to previous strains, the amount of secreted translocators of escV that expresses EscV wt -His was not reduced, and therefore robust infection was allowed.
In summary, in this work we have shown that TMD5 and TMD6 of EscV are critical for T3SS activity, likely due to their role in TMD-TMD packing within the complex. Further investigation will be required to determine the structural organization within the bacterial inner membrane and to depict the direct interaction partners of EscV within the T3SS complex.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
BM and NS-M designed the research. BM and SL performed the research. All authors analyzed the data and contributed to the writing and editing of the manuscript.