The strains and plasmids used in this study are listed in Supplementary Table 1. If not stated otherwise, A. tumefaciens and respective mutant strains were cultivated in Luria-Bertani (LB), M9 minimal medium or AB minimal medium (Schmidt-Eisenlohr et al., 1999) at 30°C supplemented with kanamycin (50 μg/ml), spectinomycin (300 μg/ml), or streptomycin (100 μg/ml) if appropriate. For virulence induction, strains were grown to an OD₆₀₀ of 0.2 in AB minimal medium at 30°C [pH 5.5, supplemented with 1% (w/v) glucose] prior to the addition of acetosyringone (AS, Sigma-Aldrich, St. Louis, MO, United States) or DMSO to a final concentration of 0.1 mM. Cells were further incubated at 23°C for 18 h.

Recombinant DNA work was carried out according to standard methods and protocols (Sambrook and Russell, 2001). For the chromosomal integration of a 3xFLAG-tag sequence, regions upstream and downstream of the respective gene’s stop codon were amplified by PCR using the primers specified in Supplementary Table 2. The FLAG-tag coding sequence was amplified from pBO2337 (Möller et al., 2014) using the primers 3xflag_fw_bamHI and 3xflag_rv_acc65I inserting a TGA stop codon at the 3′ of the 3xFLAG-tag sequence. The PCR products (upstream, 3xFLAG sequence, downstream) were digested with the engineered restriction sites and subsequently ligated into pk19mobsacB suicide vector (Schäfer et al., 1994), resulting in plasmids pBO4701 (atu3772-3xFLAG), pBO4702 (hflC-3xFLAG), and pBO4704 (ftsH-3xFLAG). Chromosomal integration of plasmids was carried out as previously described by electroporation (Möller et al., 2014). Briefly, plasmids were transferred into A. tumefaciens via electroporation (800 Ω, 25 μF, 2 kV) and colonies screened for homologous recombination events using selective media (LB with kanamycin). Single colonies were grown for 18 h in LB without kanamycin and spread on LB plates containing 10% (w/v) sucrose. Double cross over events resulted in a sucrose tolerant and kanamycin sensitive phenotype. Putative mutants containing the 3xFLAG-tag sequence in frame with the target gene on the chromosome were verified by colony PCR. For the construction of mutagenesis plasmids to delete genes encoding for SPFH proteins, 400–500 bp fragments up- and downstream of atu3772, hflK, and hflC were amplified using the primer pairs specified in Supplementary Table 2. PCR product pairs were digested with PstI, ligated and cloned into pK19mobsacB suicide vector using the restriction sites for EcoRI and HindIII resulting in the plasmid pBO3709 (for atu3772 deletion), pBO3710 (for hflC deletion) and pBO3711 (for hflK deletion). Mutant construction was carried out as described before (Czolkoss et al., 2016). Double (hflKC) and triple (hflKC/atu3772) deletion was achieved using respective hflK single and hflKC double mutant strains as a background for mutagenesis.

Cultures were grown to early stationary phase in LB or AB minimal medium as described above and OD₆₀₀ was adjusted to 4. Cells were harvested at 4°C and supernatant was discarded. Pellets were washed and resuspended in 700 μl 50 mM Tris-HCl (pH 7.4). Next, 1 mM PMSF and 1.5 mg/ml lysozyme were added and cells were further incubated for 90 min at 4°C. Disruption was achieved by ultrasonication using a VialTweeter instrument (Hielscher, Teltow, Germany). Cell debris was removed by centrifugation at 20.000 × g at 4°C for 15 min. Membranes were pelleted via ultracentrifugation at 320.000 × g for 90 min at 4°C. Two hundred microliter of supernatant were transferred to a fresh microcentrifuge tube and stored for further analysis (cytosolic samples, “C”). Membrane pellets were resolved in 600 μl Lysis and Separation Buffer from the CelLytic^TM MEM Protein Extraction Kit (Sigma-Aldrich, St. Louis, MO, United States) containing 6 μl of the provided protease inhibitor cocktail. The lysate was centrifuged for 15 min at 20.000 × g and 4°C. Subsequently, 100 μl of supernatant were transferred to a fresh microcentrifuge tube and stored for further analysis (membrane samples, “M”). Phase-separation between DSMs and DRMs was carried out following the manufactures’ instructions. Ultimately, 100 μl of each fraction were stored for further analysis.

For sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequent Western blotting of the membrane preparation samples, proteins from equal volumes were precipitated using chloroform and methanol. Briefly, samples were mixed with 4x volumes of methanol and vortexed. 1x volume of chloroform was added and samples were again vortexed before the addition of 3x volumes milliQ grade water. Samples were centrifuged at 20.000 × g for 6 min at room temperature. Upper layer was removed and 3x volumes of methanol were added. Samples were again centrifuged at 20.000 × g for 6 min and methanol was evaporated in a SpeedVac system. Protein pellets were resolved in 1x SDS sample buffer, containing 1 M Tris (pH 6.5), 50% glycerol (vol/vol), 10% SDS (wt/vol), 0.5% bromophenol blue (wt/vol) and 5% β-mercaptoethanol (vol/vol). Equal volumes of fractions were analyzed by SDS-PAGE and Western blot analysis. FLAG-tagged proteins were detected using monoclonal ANTI-3XFLAG M2 antibody (Sigma-Aldrich, St. Louis, MO, United States) and secondary goat anti-mouse IgG (H + L)-HRP conjugate (Bio-Rad, Hercules, CA, United States). VirB5, VirB9, TssI, and TssH were detected using protein-specific antibodies (rabbit) and secondary goat anti-rabbit IgG (H + L)-HRP conjugate (Bio-Rad, Hercules, CA, United States). Protein signals were visualized using Luminata Forte Western HRP substrate (Merck-Millipore, Billerica, MA, United States). For signal detection, a ChemiImager Ready system (Alpha Innotec, Kasendorf, Germany) was used.

For immunofluorescence, cells were cultivated as described above in LB or AB minimal medium to early stationary phase, harvested and adjusted to an OD₆₀₀ of 2 in PBS buffer. Cells were fixed in 2.7% paraformaldehyde and 0.01% glutaraldehyde for 30 min at 4°C. Cells were washed in PBS and GTE buffer containing 50 mM glucose, 10 mM EDTA, and 20 mM Tris. Permeabilization of the membrane was achieved by incubation in PBS buffer containing 5 mM EDTA and 5 mg/ml lysozyme at 4°C for 5 min followed by PBS washing and incubation in PBS containing 0.1% Triton X-100 for 5 min at room temperature. Triton X-100 was removed by PBS washing and cells were blocked with PBST [PBS containing 0.1% tween 20 (vol/vol)] containing 3% BSA (wt/vol) for 1 h at room temperature. Cells were again washed in PBST and incubated with primary antibody solution in PBS, 0.1% BSA for 1 h at 30°C. After washing with PBST, cells were incubated with secondary antibody solution in PBST for 1 h at 30°C. Finally, cells were washed three times in PBST, resuspended in 200 μl PBS and mounted onto agar pads containing 0.5% low-melting agarose. Cells were imaged with an Olympus BX51 microscope using a U-UCD8 condenser and an UPlanSApo 100XO objective and images were taken with a CC12 digital color camera (all components by Olympus, Hamburg, Germany).

Protein and LC-MS/MS analysis was carried out as previously described (Plohnke et al., 2014, 2015). Briefly, proteins were concentrated to a single band in SDS gels by terminating electrophoresis when the migration front reached the interface between stacking and separating gel. Protein bands were cut out and digested with trypsin (Rexroth et al., 2003). Peptides were extracted from the gel using 50% (vol/vol) acetonitrile and 1% (vol/vol) formic acid prior to lyophilization. For LC-MS/MS analysis, peptides were re-solubilzed in 0.1% (vol/vol) formic acid. For reversed-phase chromatography a gradient of solvent A [0.1% (vol/vol) formic acid] and solvent B [99.9% (vol/vol) acetonitrile and 0.1% (vol/vol) formic acid] was used. MS-analysis was carried out using a Thermo LTQ Orbitrap XL mass spectrometer in duty cycle consisting of one 400–2000 m/z FT-MS and four MS/MS LTQ scans. MS and MS/MS data were acquired using Xcalibur (all by Thermo Fisher Scientific, Waltham, MA, United States). Each sample consisted of three independent biological replicates. Obtained LC-MS/MS spectra were analyzed using the Sequest algorithm (Eng et al., 1994). Peptides were matched against UniProt database of A. tumefaciens for protein identification. Proteins were quantified using spectral counting and normalized spectral abundance factors (NSAF) when identified by at least one unique peptide in two of three replicates (Plohnke et al., 2015). Proteins were classified as significantly enriched in DRMs or DSMs according to Student’s t-test on a significance level of 95% if their ratio was greater than 1.5 or lower than 0.7. Automated localization prediction of proteins found in DRM and DSM fractions was performed using the CELLO2GO algorithm (E-value set to 0.005) (Yu et al., 2014). Protein groups were identified and quantified with information from the UniProt and KEGG database and the BlastKOALA annotation tool (Kanehisa et al., 2016). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD028782 (Perez-Riverol et al., 2019).

Phospholipids of A. tumefaciens strains were isolated according to Bligh and Dyer and analyzed as previously described (Bligh and Dyer, 1959; Czolkoss et al., 2016). For two-dimensional TLC (2D-TLC), mixtures of chloroform:methanol:water (65:25:4) and chloroform:methanol:acetic acid:water (90:15:10:3.5) were used as running solvents for first and second dimension, respectively. For the visualization of the lipids, plates were charred after CuSO₄-treatment at 180°C

Biofilm formation of A. tumefaciens strains was quantified using a crystal violet staining assay as described elsewhere with slight modifications (Knoke et al., 2020). Briefly, 1 ml of M9 minimal medium in a 12-well microtiter plate (Sarstedt, Nümbrecht, Germany) was inoculated with overnight cultures to an OD₆₀₀ of 0.5. Plates were incubated for 3 days at 15°C. After 24, 48, and 72 h, 50 μL crystal violet (0.5%) were added and the plates were further incubated for 10 min at room temperature with gentle rocking. Wells were carefully washed three times with 1.5 ml water and the dye was eluted from the organic material with 1 ml 100% ethanol. Biofilm was quantified by measuring the absorbance at 570 nm.

Swimming motility of A. tumefaciens was analyzed by spotting 3 μl of stationary phase cultures on M9 minimal medium soft agar plates [0.5% agar (wt/vol)]. Plates were incubated for 24 h at 30°C and swarm diameters were measured.

Functionality of the T4SS was monitored using the AGROBEST assay as described previously (Wu et al., 2014; Groenewold et al., 2019). Briefly, Arabidopsis seeds were sterilized and germinated in MS medium for 7 days at 22°C in a 16 h/8 h light/dark period prior to infection. A. tumefaciens wild type and SPFH mutant strains were transformed with pBISN1 harboring the gusA-intron, cultivated in AB minimal medium for virulence induction as described above and resuspended to an OD₆₀₀ of 0.02 in infection medium (50 ml MS medium, 50 ml AB medium, 200 μM AS). Arabidopsis seedlings were incubated with A. tumefaciens solutions for 3 days as described above and measured for transient GUS expression by staining with 5-bromo-4-chloro-3-indolyl glucuronide (X-Gluc) for 3 h at 37°C.

T6SS-dependent secretion of TssD (Hcp) was evaluated as described before with minor modifications (Wu et al., 2008). Briefly, A. tumefaciens cultures grown in AB minimal medium at 30°C for 6 h were harvested by centrifugation and total protein samples were prepared using 1x SDS sample buffer. Supernatants harboring the secreted proteins were filtered through a 0.45 μm membrane and concentrated by TCA precipitation. Protein pellets were resuspended in 10 μl of 1 M Tris buffer and 10 μl 2x SDS sample buffer. Equal volumes of total protein samples (T) and supernatant-derived samples (S) were analyzed by SDS-PAGE and Western detection of TssD using a protein-specific antibody as described above.

To explore the presence and protein composition of membrane microdomains in A. tumefaciens, we established a DRM isolation protocol using the CelLytic^TM MEM Protein Extraction Kit (Sigma-Aldrich, St. Louis, MO, United States). This kit has successfully been applied to bacterial model organisms like B. subtilis, B. anthracis, and Staphylococcus aureus (López and Kolter, 2010; Somani et al., 2016) and enables rapid purification of DRMs without tedious sucrose density centrifugation (Wermser and López, 2016). Our protocol is based on an initial membrane preparation followed by selective solubilization (by a mixture of non-ionic detergents) and subsequent low-speed centrifugation to achieve phase separation between DSMs and DRMs (Figure 1A). This protocol is suitable for the isolation of DRMs from the inner membrane (IM) and outer membrane (OM), which is important since microdomains are found in both membranes (Toledo et al., 2018). SDS-PAGE analysis of the DRM and DSM fractions from cells grown in LB medium revealed a distinct and reproducible protein pattern with some proteins – especially in the higher molecular range – enriched in DRMs, while others were depleted (Figure 1B).

Proteins from the SPFH (stomatin/prohibitin/flotillin/HflKC) family are commonly co-purified with DRMs (Browman et al., 2007) making them suitable markers for the validation of our membrane fractionation protocol. We searched for putative SPFH homologs in A. tumefaciens and identified three candidates. The genes atu2044 (hflC) and atu2045 (hflK) encode putative modulators of the FtsH protease, and atu3772 codes for a protein with high homology to E. coli QmcA, another FtsH interaction partner (Supplementary Figure 1A) (Chiba et al., 2006). The hflC and hflK genes form a bicistronic operon on the circular chromosome. Atu3772 is encoded on the linear chromosome immediately upstream of atu3773, coding for an NfeD (nodulation formation efficiency D)-like protein. SPFH domains presumably are important for interaction with the membrane. In HflC, HflK and Atu3772 they are followed by predicted coiled-coil regions involved in protein oligomerization (Rivera-Milla et al., 2006). Importantly, the three Agrobacterium SPFH proteins lack the C-terminal flotillin domain (InterPro: IPR031905) found in some prokaryotic flotillin homologs, such as B. subtilis FloT (Supplementary Figure 1B).

We generated strains expressing atu3772, hflC, and ftsH as FLAG-tagged variants from their native gene locus and validated protein production by Western blot analysis using FLAG-tag specific antibodies (Supplementary Figure 1C). The signals corresponded to the calculated molecular masses (Atu3772^FLAG: 40.5 kDa; HflC^FLAG: 37.5 kDa; FtsH^FLAG: 73.5 kDa). Protein levels of HflC and Atu3772 were irrespective of growth whereas the FtsH signal diminished in late stationary phase. This might be explained by the loss of the C-terminal FLAG-tag due to the self-processing activity of FtsH in late growth phases (Akiyama, 1999).

All three FLAG-tagged proteins in the engineered strains were exclusively found in the membrane, and not in the cytoplasmic fraction (Figure 1B) demonstrating that the tag does not interfere with proper membrane localization of these proteins. Moreover, HflC^FLAG, Atu3772^FLAG, and FtsH^FLAG showed a substantial enrichment in the DRM fraction, which is in good agreement with data from B. subtilis, S. aureus, and B. burgdorferi, and therefore validates the practicability of our purification protocol (López and Kolter, 2010; Yepes et al., 2012; Toledo et al., 2015).

Given that A. tumefaciens SPFH proteins were co-purified with DRMs, which are thought to originate from a spatial membrane segregation into distinct domains, one would expect a patchy localization of these proteins in vivo. To test this assumption, we used an immunofluorescence-based approach to visualize Atu3772, HflC, and FtsH thanks to their FLAG-tag. Additionally, strains producing Hfq or Atu8019 as FLAG-tagged variants were analyzed as controls for a cytosolic or uniform membrane localization, respectively (Figure 1C; Knoke et al., 2020). The two SPFH proteins Atu3772 and HflC, as well as FtsH were detected in multiple and distinct fluorescent foci after treatment with FLAG-antibodies (mouse) followed by incubation with AlexaFluor 488 goat-anti-mouse secondary antibodies. A wild-type control strain did not exhibit any fluorescence signals under the same conditions (data not shown).

The findings described above prompted us to identify the protein composition of A. tumefaciens DRMs. Proteins residing in DRMs from cells grown in LB medium were identified by an LC-MS/MS-based approach, and peptides were matched against the A. tumefaciens UniProt database. For statistical parameters, see the materials and methods section. In total, 73 proteins were classified as “DRMs only” (n = 53) or “strongly enriched in DRMs” (n = 20). Eighty nine proteins localized to DSMs, and 49 thereof were classified as “DSMs only” (see Supplementary Table 3 for a full list). A list with selected proteins discussed in more detail below is included in Table 1. Localization prediction of the proteins identified in DRMs suggests a substantial enrichment of inner and outer membrane proteins (Supplementary Figure 2A). A number of annotated “cytoplasmic” proteins resided in the DSM fractions, which might be the result of proteins dynamically interacting with the membrane, thus being included in the initial membrane fraction prior to detergent treatment. This assumption is corroborated by the identification of components of the division machinery (FtsA), the respiratory chain (NuoE, NuoF, NuoG), and substrate binding proteins of ABC transporter systems in DSMs (Table 1).

To attribute biological functions to the identified proteins, we used information from the UniProt and KEGG databases and screened for selected functional groups using the BlastKOALA webservice (Kanehisa et al., 2016). The SPFH proteins HflK, HflC, and Atu3772 were manually assigned as an individual group (SPFH) with all three members being most abundant in DRMs (Figure 2A and Table 1). Other strongly overrepresented proteins in DRMs were involved in cell envelope biogenesis (e.g., MsbA, LptDE, ExoP), transport and secretion (e.g., TssL, AvhB9-11), or assigned to motility- and chemotaxis-related processes (e.g., MotB, Atu0027). Interestingly, several FtsH substrates and interaction partners like PpiD, SecY, SecD, and YidC were also found in DRMs (Table 1). PpiD is a membrane-associated chaperone that interacts with the Sec translocon and is degraded by FtsH, at least in E. coli (Sachelaru et al., 2014; Bittner et al., 2015). SecY and SecD are additional FtsH substrates in E. coli (Akiyama et al., 1996; Arends et al., 2016; Bittner et al., 2017), and a direct interaction between SecY and a bacterial SPFH protein was demonstrated in B. subtilis (Bach and Bramkamp, 2013). The FtsH interaction partner YidC mediates the insertion of multimeric protein complexes into the membrane (van Bloois et al., 2008). In contrast, proteins enriched with DSMs were involved in various metabolic processes, e.g., energy, amino acid, or carbohydrate metabolism and translation (Figure 2A).

The strong overrepresentation of proteins involved in transport or secretion processes, as well as the identification of FtsH along with its potential modulators (HflK, HflC, and Atu3772) and subtrates (PpiD, SecY, SecD) in the DRM fractions suggests a functional partitioning between detergent-soluble and non-soluble membrane domains within the plasma membranes of A. tumefaciens.

Agrobacterium tumefaciens deploys two bacterial secretion systems. The expression of genes coding for a T4SS, which delivers T-DNA and effector proteins into the plant cell, is induced in response to environmental signals such as phenolic compounds released from wounded plant cells and acidic milieus (Chen and Winans, 1991). The acid-induced T6SS provides a competitive advantage against other bacteria during plant colonization (Wu et al., 2012; Ma et al., 2014). Since the T6SS component TssL and proteins of the T4SS-like AvhB system (AvhB9-11) were identified in the DRM fractions from LB-grown cultures (Table 1), we hypothesized that further structural components of both the T4SS and the T6SS might reside in DRMs isolated from virulence-induced A. tumefaciens cells. We thus cultivated A. tumefaciens in AB minimal medium (pH 5.5) supplemented with acetosyringone (AS) to induce the virulence cascade. Membranes of virulence-induced (+vir) and non-induced (–vir) cultures were purified, separated into DRMs and DSMs, and subjected to LC-MS/MS. Using the criteria defined above for quality control of the raw data, 278 proteins were found to be enriched in the DRM fractions from non-induced cultures. 130 of them were classified as “DRM only” (see Supplementary Table 3 for a complete list). Thirty-one of the 142 proteins identified in the DSM fraction were absent in the DRM fraction and therefore categorized as “DSM only.” Under virulence-induced conditions, 453 proteins were overrepresented in DRMs (235 “DRM only”) and 156 in the DSM fractions (50 “DSM only”). It has been noted previously that different cellular activities under different physiological conditions can account for differences in the DRM proteome (Vijaya Kumar et al., 2020). This might explain the higher number of proteins identified from cells grown in AB compared to LB medium. Medium-dependent differences are supported by a large overlap of 128 proteins in the DRM proteome of A. tumefaciens grown in AB medium with and without acetosyringone (Supplementary Figure 2B). In total, 27 proteins were found in the DRM fractions under all three conditions (Supplementary Figure 2B and Table 2).

The DRM enrichment of proteins belonging to functional groups like “cell envelope biogenesis” or “transport and secretion” in AB medium was consistent with the results from LB-grown cultures (Figure 2B). As expected, HflC, HflK, Atu3772, and FtsH showed a clear preference for the DRM fraction and did not respond to AS treatment (Figure 2B and Table 3). The absolute number of proteins involved in transport/secretion and motility/chemotaxis increased in both DRMs and DSMs upon virulence induction while their proportional distribution between both fractions remained largely unaffected. The percentage of many proteins involved in “housekeeping” metabolic processes (energy, amino acid, carbohydrate metabolism, and translation) increased after AS treatment in DRMs. This might be explained by profound VirA/G-induced metabolic restructuring in anticipation of opines, that are produced and secreted by transfected plants. Presumably due to their low abundance, we were unable to identify many of the proteins (derived from the noc-region of the Ti-plasmid) responsible for the uptake and metabolism of nopaline in A. tumefaciens C58 (Subramoni et al., 2014).

Predictions of cellular localization of proteins in DRMs and DSMs were similar to those of LB medium, indicating the robustness of our protocol for the isolation of membrane proteins along with DRMs (Supplementary Figure 2A).

Strikingly, many of the T4SS-associated proteins were found to be highly enriched (e.g., VirB4, VirB8-10) or identified exclusively in the DRMs fractions (e.g., VirB7) of virulence-induced cultures (Table 3). Transcription of the vir operon is controlled by the two-component system VirA/VirG, which senses phenolic compounds and activates transcription of vir genes encoded on the Ti-plasmid. Notably, both proteins were either exclusively found (the sensor kinase VirA) or strongly enriched (the response regulator VirG) within DRM fractions of AS-treated cultures (Table 3). As expected, many of the VirA/VirG-regulated proteins were identified only after virulence induction (e.g., VirB2-3. VirB5, VirB7, VirC1-2, VirD1-2).

The T6SS machinery is encoded in two different operons (imp and hcp), which are both activated under acidic conditions via transcriptional regulation by ChvG/ChvI (Wu et al., 2012). Interestingly, T6SS functionality decreases in the presence of phenolic compounds, which presumably results from reduced amounts of tube forming TssD (Wu et al., 2012; Lin et al., 2013). Consistent with these reports, we observed that the relative amounts of TssD were lowered when AS was added to the medium (Table 3). Importantly, several of the T6SS proteins from both operons were strongly overrepresented in DRM fractions (e.g., TagFH, TssH, TssM TagE, TssC₄₀). These data clearly demonstrated the preference of both T4SS and T6SS for a detergent-resistant membrane environment, suggesting that lateral membrane organization into domains of distinct physicochemical parameters might be beneficial for infection and interspecies competition.

To validate the localization of the T4SS and T6SS to restricted membrane domains by an independent approach, we tracked the localization of selected components of these secretion systems by specific antibodies (Figure 3A). Following SDS-PAGE and Western blotting, antisera against VirB5 and VirB9 as components of the T4SS and against the T6SS components TssH (ClpV) and TssD (Hcp) were used (Figure 3B). Consistent with the MS analysis, both Vir proteins were detected at the calculated sizes (VirB9: 32 kDa; VirB5: 23 kDa) in total membrane fractions (M) and in DRMs of virulence-induced cultures. Only small amounts of VirB9 were found in DSMs from AS-treated cultures, and VirB5 was not at all detectable in DSM fractions. TssH and TssD were detected at the calculated molecular masses of 96 and 17 kDa, respectively, and were clearly enriched in DRMs.

We further investigated the spatial distribution of the T4SS and T6SS in vivo using an immunofluorescence-based approach (Figure 3C). The T4SS was detected via the minor pilus core component VirB5 and the T6SS via the ATPase subunit TssH using monoclonal primary antibodies followed by AlexaFluor secondary antibody treatment. As expected, VirB5 was not detected under non-induced conditions (–vir). After virulence induction (+vir), several fluorescence foci appeared at the periphery of the cell. VirB5 has a role during pilus assembly and primarily localizes to the tip of the pilus (Aly and Baron, 2007). The T6SS component TssH was also found in multiple spots along the perimeter of the cell when grown in AB minimal medium (–vir). In summary, these findings provide evidence that the delivery of DNA and effector proteins via both secretion systems is restricted to specific sites in the bacterial membrane that confer detergent resistance to the embedded proteins.

Because SPFH proteins are hallmarks of both prokaryotic and eukaryotic DRMs, we asked whether the localization of A. tumefaciens T4SS and T6SS in DRMs is dependent on these proteins. Corresponding single-, double-, and triple-mutant strains were constructed and analyzed for various phenotypes that might be associated with an altered membrane organization. All mutants showed wild type-like growth, biofilm formation and motility (Supplementary Figure 3A). Lipid profiles from SPFH mutant strains were indistinguishable from the wild type featuring phosphatidylglycerol (PG), phosphatidylethanolamine (PE), monomethyl-PE (MMPE), phosphatidylcholine (PC), ornithine lipids (OL), and cardiolipin (CL) in comparable amounts (Supplementary Figure 3B). Importantly, T4SS-mediated transformation of Arabidopsis seedlings and T6SS-mediated secretion of TssD were not impaired in SPFH mutants compared with the wild type (Supplementary Figures 3C,D). Accordingly, the enrichment of VirB5 and VirB9 (T4SS; Figure 4A) as well as TssH and TssD (T6SS; Figure 4B) in DRM fractions purified from the different mutant strains was not altered. Cumulatively, these results suggest that Atu3772, HflK, and HflC are dispensable for DRM localization and functionality of both the T4SS and the T6SS.