The protective function of an immunity protein against the cis-toxic effects of a Xanthomonas Type IV Secretion System Effector

Many bacterial species use specialized secretion systems to translocate proteinaceous toxic effectors into target bacterial cells. In most cases, effectors are encoded in bicistronic operons with their cognate immunity proteins. The current model is that immunity proteins could, in principle, provide protection in two different ways: i) by avoiding self-intoxication (suicide or cis-intoxication) or ii) by inhibiting intoxication due to “friendly-fire” translocation from neighboring sister cells (fratricide or trans-intoxication). Here, we set out to distinguish between these two protection mechanisms in the case of the bactericidal Xanthomonas citri Type IV Secretion System (X-T4SS), where killing is due to the action of a cocktail of secreted effectors (X-Tfes) that are inhibited by their cognate immunity proteins (X-Tfis). We use a set of X. citri mutants lacking multiple X-Tfe/X-Tfi pairs to show that X-Tfis are not absolutely required to protect against trans-intoxication. Our investigation then focused on the in vivo function of the lysozyme-like effector X-TfeXAC2609 and its cognate immunity protein X-TfiXAC2610. We observe the accumulation of damage in the X. citri cell envelope and inhibition of biofilm formation due to the action of X-TfeXAC2609 in the absence of X-TfiXAC2610. We show that X-TfeXAC2609 toxicity is independent of an active X-T4SS and that X-TfiXAC2610 protects the cell colony against X-TfeXAC2609-induced cis-intoxication via autolysis. In vitro assays employing X-TfiXAC2610 mutants were used to test and validate an AlphaFold2-derived model of the X-TfeXAC2609-X-TfiXAC2610 complex which presents topological similarities with the distantly related Tse1/Tsi1 complex from P. aeruginosa and the the i-type lysozyme from Meretrix lusoria (MI-iLys) in complex with PliI-Ah from Aeromonas hydrophila. While immunity proteins in other systems have been shown to protect against attacks by sister cells (trans-intoxication), this is the first description of an antibacterial secretion system in which the immunity proteins are dedicated to protecting cells against cis-intoxication.


Introduction
Bacteria are continuously competing with each other for space and nutrients. One widespread competition strategy is the secretion of effectors into adjacent bacterial cells (Russell et al. 2011;Benz and Meinhart 2014;Souza et al. 2015;Cianfanelli, Monlezun, and Coulthurst 2016) mediated by specialized secretion systems, as for example: the Type IV Secretion Systems (T4SSs) from Xanthomonas citri (Souza et al. 2015) and Stenotrophomonas maltophilia , the Type V Secretion System (T5SS) from Escherichia coli (Aoki et al. 2005), the Type VI Secretion Systems (T6SSs) from Pseudomonas aeruginosa (Russell et al. 2011), Burkholderia thailandensis (Schwarz et al. 2010), Salmonella typhimurium (Sana et al. 2016), Vibrio cholerae (MacIntyre et al. 2010) and Serratia marcescens (Murdoch et al. 2011), the Caulobacter crescentus CDZ-based Type I secretion System (García-Bayona, Guo, and Laub 2017), and the Type VII Secretion Systems (T7SSs) encoded by Gram-positive bacteria Staphylococcus aureus (Cao et al. 2016) and Bacillus subtilis (Tassinari et al. 2022;Kobayashi 2021). All these different secretion systems have at least one effector and one cognate immunity protein, the latter of which protects the donor from self-intoxication.
T4SSs are large protein complexes which traverse the cell envelope, producing a channel through which proteins or protein-DNA complexes can be secreted into animal or plant hosts or other bacterial cells (Alvarez-Martinez and Christie 2009;Ilangovan, Connery, and Waksman 2015;Y. G. Li and Christie 2018;Sheedlo et al. 2022). Canonical type-A T4SSs are usually composed of 12 proteins, VirB1-VirB11 and VirD4 (Fronzes, Christie, and Waksman 2009;Alvarez-Martinez and Christie 2009;Costa et al. 2015). Chromosome-encoded T4SSs found in the order Xanthomonadales and some other proteobacterial species (X-T4SSs) secrete antibacterial effectors (X-Tfes) that are recruited via interactions with the VirD4 ATPase coupling protein (Alegria et al. 2005;Souza et al. 2011;Oka et al. 2022;Sgro et al. 2019). All X-Tfes contain a conserved C-terminal domain termed XVIPCD that interacts directly with the all-alpha domain of VirD4 (Alegria et al. 2005;Oka et al. 2022). In the phytopathogen Xanthomonas citri, the X-Tfe XAC2609 effector is secreted in a manner that is dependent on its XVIPCD and on a functional X-T4SS (Souza et al. 2015;Oka et al. 2022). The N-terminal portion of X-Tfe XAC2609 contains a glycoside hydrolase family 19 (GH19) domain that cleaves peptidoglycan (PG) and a PG binding domain. This PG hydrolase activity is inhibited by its cognate immunity protein X-Tfi XAC2610 (Souza et al. 2015). Therefore, X-Tfe XAC2609 and X-Tfi XAC2610 form an effector/immunity protein pair associated with the X. citri X-T4SS, which provides an adaptive advantage for X. citri in co-cultures with E. coli and other bacterial species (Oka et al. 2022;Souza et al. 2015).
Secretion system-mediated bacterial killing is typically evaluated in interbacterial competition assays between attacker and prey cells that code for one or more different effector-immunity protein pairs. The rationale is that the prey is susceptible to the toxicity of the delivered effector(s) because they do not produce at least one cognate immunity protein (Aoki et al. 2005;Russell et al. 2011Russell et al. , 2012García-Bayona, Guo, and Laub 2017;Kobayashi 2021;Tassinari et al. 2022). Since most bacteria-bacteria interactions are between genetically identical cells (e.g. bacterial colonies), bacteria coding for toxic effectors also code for immunity proteins that protect themselves against intoxication via their own effectors. It is reasonable to suppose that immunity proteins should be localized in the same subcellular compartment where its cognate effector acts (Benz and Meinhart 2014;Whitney et al. 2013;Russell, Peterson, and Mougous 2014;Jurėnas and Journet 2021); for example, the X-Tfe XAC2609 toxin that targets peptidoglycan has a cognate immunity protein, X-Tfi XAC2610 , that carries an N-terminal signal peptide and lipobox that directs it to the periplasm (Souza et al. 2015;Sgro et al. 2019). In this scenario, immunity proteins could, in principle, provide protection against two different, not necessarily exclusive, toxicity mechanisms: i) intoxication due to "friendly-fire" translocation of toxic effectors from neighboring sister cells (fratricide or trans-intoxication; Figure 1A) or ii) self-intoxication (suicide or cis-intoxication; Figure 1B) that results from the action of endogenously produced toxins. Both cis-and trans-intoxication mechanisms have been observed previously for T6SS mediated effector transfer (Basler and Mekalanos 2012;Hood et al. 2010;Russell et al. 2011;Dong et al. 2013); (M. Li et al. 2012;Whitney et al. 2015); (Basler and Mekalanos 2012;Hood et al. 2010;Russell et al. 2011;Dong et al. 2013).
Here, we show that a X. citri strain lacking multiple effector-immunity protein pairs remains resistant to fratricidal X-T4SS-mediated attack by wild-type X. citri cells, thus providing evidence that the role of X-T4SS immunity proteins (X-Tfis) is not restricted to avoiding X-T4SS-mediated fratricide (trans-intoxication). We also show that a X. citri X-Tfi XAC2610 knockout strain suffers autolysis in a process that is mediated by X-Tfe XAC2609 but is independent of the X-T4SS. Cell ultra-structural aspects of the autolysis process were analyzed by fluorescence and electron microscopies. We demonstrate that X-Tfi XAC2610 is important for biofilm formation and is required to protect against the detrimental effects of X-Tfe XAC2609 , even in the absence of a functional X-T4SS. These results support the conclusion that the protective function of X-Tfis is geared mainly towards the cis-intoxication (self-intoxication) effects of the endogenous X-Tfes.

Cultivation conditions
The oligonucleotides, plasmids and bacterial strains used in this work are described in Tables S1, S2 and S3, respectively. X. citri strains were cultivated in LB-agar or 2xTY media. The concentrations of the antibiotics used were: 70 μg/mL ampicillin, 100 μg/mL spectinomycin, 20 μg/mL gentamicin and 100 μg/mL kanamycin. Experiments involving X. citri were initiated by picking isolated colonies for inoculation in 5 mL of 2xTY medium supplemented with antibiotics and grown for 12 hours. The cells were then harvested, the optical density (OD 600nm ) adjusted to 0.05 with fresh 2xTY medium and a 2 mL volume of this culture was grown at 30°C, 200 rpm for another 12-18 hours.

Western blotting
After the cultivation of the X. citri inocula in 2xTY supplemented with spectinomycin, cells were harvested by centrifugation (5,000 rpm, 5 minutes) and resuspended in water to an OD 600nm of 40. A 5-µL aliquot of resuspended cells was lysed in SDS-PAGE loading buffer at 100ºC for 5 minutes, separated by SDS-PAGE and assayed by Western blot. Rabbit polyclonal antibodies serum at 1:1,000 dilution anti-X-Tfe XAC2609 , anti-X-Tfi XAC2610 and anti-VirB7 (Souza et al. 2015) were detected by IRDye 800 CW goat anti-rabbit IgG (LI-COR Biosciences) at 1:30,000 dilution. Secondary antibody signals were detected using an Odyssey infrared imaging system (LI-COR Biosciences).

Colony transparency assay
After the cultivation of the X. citri inocula in 2xTY medium supplemented with spectinomycin, cells were harvested by centrifugation (5,000 rpm, 5 minutes), washed in water twice and the OD 600nm was normalized to 0.05 in 2xTY medium.
Then, 5 µL of each X. citri strain culture was applied onto the surface of 1.5% LB-agar plates supplemented with 70 μg/mL ampicillin, spectinomycin and 0.1% arabinose. The plates were cultured at 30°C and photographs were recorded on a transilluminator every 24 hours.

Colony viability assay
After the cultivation of the X. citri inocula in 2xTY supplemented with spectinomycin and 0.1 % arabinose, cells were harvested by centrifugation (5,000 rpm, 5 minutes), washed in water twice and the OD 600nm was normalized to 0.05. Then, 5 µL of each X. citri strains were pipetted onto 1.5 % LB-agar plates supplemented with spectinomycin and 0.1% arabinose. Next, the plates were cultured at 30°C. To assess the cell viability, X. citri colonies were removed from the LB-agar plates and gently resuspended in 1 ml of 2xTY, every 24 hours of cultivation. Serial dilutions were made on LB-agar (1.5%) plates supplemented with ampicillin and spectinomycin to estimate viability of colony forming units per ml (CFU/mL). The results shown are the means of 3 independent experiments.

Transmission electron microscopy (TEM)
After growth, liquid cultures of X. citri cells were centrifuged in microtubes (5,000 rpm, 5 minutes, room temperature) and supernatants were discarded.
Cell washes were performed twice by adding 1 mL of phosphate buffer solution (0.2 M, pH 7.4). After 15 minutes at room temperature, samples were centrifuged (6,500 rpm, 5 minutes) and the supernatant discarded. A fixation step was performed by the addition of 1 mL of modified Karnovsky's solution (2.5% glutaraldehyde, 2% paraformaldehyde, 200 mM phosphate buffer, pH 7.4) (Watanabe and Yamada 1983). Then, the samples were incubated for 90 minutes at room temperature and centrifuged (5,000 rpm, 5 minutes, room temperature), the supernatant was discarded and an extra washing step was carried out with phosphate buffer (200 mM, pH 7.4). A second fixation step was performed by the addition of 1 mL of 1% OsO 4 solution followed by incubation for 1 hour on ice. The samples were then centrifuged (5,000 rpm, 2 minutes, room temperature) and washed with water. The pellets were dissociated from the microtubes with the use of a glass rod. Thereafter, 1 mL of 0.5% aqueous solution of uranyl acetate was added, followed by another incubation period of 1 hour at room temperature.Next, uranyl acetate was removed by centrifugation and 1 mL of 60% ethanol solution was added. Successive incubations were performed with increasing concentrations of ethanol (70%, 80%, 90%, 95% and 100%). After the last dehydration step with 100% ethanol, 1 mL of propylene oxide solution (Electron Microscopy Sciences) was added, followed by incubation at room temperature for 10 minutes. Cells were collected by centrifugation as above and three more incubations and washes with propylene oxide were performed. Then, the samples were submitted to 4 successive exchanges of propylene oxide and resin solutions for microscopy (Low Viscosity embedding Kit (# 14300), Spurr's) in proportions of 1:1, 1:3, 0:1 and 0:1 (propylene oxide:resin) with incubation times of 40 minutes, 90 minutes and 12 hours and 72 hours, respectively. The first 3 incubations (1:1, 1:3 and 0:1) were performed with gentle shaking at room temperature. The fourth incubation step was performed at 60°C without agitation in order to solidify the resin. After resin solidification, histological and ultrathin sections were obtained. The slices were

Time-lapse fluorescence microscopy
Time-lapse fluorescence microscopy assays were carried out as previously described (Oka et al. 2022). Briefly, after the cultivation of the X. citri inocula in 2xTY supplemented with appropriate antibiotic, cells were harvested by centrifugation (5,000 rpm, 5 minutes), washed in water twice and the OD 600nm was normalized to 0.5. Then, 1 µL of each X. citri strain were pipetted onto a thin LB-agarose support supplemented with propidium iodide (1 µg/mL), appropriate antibiotic and observed with a Nikon Eclipse Ti microscope equipped with filters for GFP (GFP-3035B-000-ZERO, Semrock) and propidium iodide (TxRed-4040B, Semrock) and a Nikon Plan APO 100x objective. Images were collected every 10 minutes. Image processing and quantitative analysis of the number of cells having a damaged cell envelope and cell counting related to Smovies (1-5) were performed manually using Fiji software (Schindelin et al. 2012) multipoint tool. Video microscopy showing interbacterial competition between X. citri strains wild type transformed with pBBR-RFP against Δ8Δ2609-GFP transformed with pBBR-GFP, the after cells growing and washing steps, the cultures were mixed 1:1 and the microscopy was performed as described above, except by supplementing the LB-agarose support with propidium iodide.

Protein expression and purification
X-Tfi XAC2610 His-22-267 expression and purification were carried out as previously described (Oka et al. 2022), with some adaptations. Briefly, cells of E. coli Bl21 (DE3) carrying the vectors (Table S2) were grown to OD 600nm = 0.8 and the heterologous expression of recombinant proteins was induced using 0.5 mM IPTG for 16 hours at 18 o C and 180 rpm in 2xTY medium. After induction of protein expression, cell recovery and lysis, the proteins of interest were submitted to affinity chromatography using HiTrap Ni 2+ -chelating resin (Cytiva) previously equilibrated with 20 mM Tris-HCl buffer (pH 8.0), 200 mM NaCl, 20 mM imidazole, and 2% (v/v) glycerol, subsequently washed with the same buffer and eluted with an imidazole gradient (20-500 mM). X-Tfe XAC2609 (1-308) was purified by chromatography as previously described (Souza et al. 2015), using an anion exchange Q-sepharose column (GE Healthcare) and a size exclusion Superdex S200 column (GE Healthcare). Pure proteins were identified by SDS-PAGE, and the quantification was estimated using absorbance at 280 nm.

Biofilm formation assay
X. citri inocula containing pBRA-and pBBR-derived plasmids were grown for 12 hours in 2xTY supplemented with ampicillin, gentamicin, spectinomycin and 0.3% arabinose. After growth, X. citri inocula OD 600nm was normalized to 0.05 and incubated for 5 days at 30 o C in a laminated microscopy chamber (Nu155411; Lab-Tek, NUNC). Biofilm images were acquired using a Nikon Eclipse Ti microscope equipped with a 100x magnification objective (CFI Plan Apo Lambda 100XH) and a fluorescence filter for GFP (GFP-3035B, Semrock).
Images were collected from the base to the top over 20 µm in the Z axis at 0.5-µm intervals, and stacked using the FIJI software (Schindelin et al. 2012).
The results shown are representative images of 3 independent experiments.
For the 24 wells plate-based biofilm assay, X. citri inocula were grown for 24 hours at 30 o C, 200 rpm in 2xTY medium supplemented with ampicillin, and then allowed to grow at room temperature (22°C) for seven days without shaking. To quantify X. citri biofilm using a 24 well plate, the protocol described in (Dunger et al. 2014) with some modifications was followed. After seven days of cultivation, the 2xTY medium was carefully removed and replaced with 1 mL of 0.1M NaCl solution. The cells were then resuspended by pipetting and transferred into 1.5 mL microtubes. After vortexing and centrifuging (6000 rpm, 5 minutes at 4°C) the supernatant was discarded, this step was repeated two more times. Next, 1 mL of 0.1% crystal violet solution was added, mixed by vortexing, and incubated for 30 minutes at room temperature. The tubes were then centrifuged (6000 rpm, 5 minutes at 4°C) , the supernatant was discarded, and the cells were washed two more times with 1 mL of 0.1M NaCl followed by centrifugation steps. Finally, the pellets were resuspended with 100% ethanol and the absorbances were measured at 570 nm.

Fluorescence spectroscopy
The protein X-Tfi XAC2610 (55-267) was expressed and purified as previously described (Souza et al. 2015). Fluorescence assays of the purified protein were measured employing an ATF-105 spectrofluorometer (Aviv Biomedical).
Thermal denaturation experiments were conducted with X-Tfi XAC2610 (55-267) at 0.5 mM in 20 mM Tris-HCl (pH 7.5) and 50 mM NaCl. Where indicated, EGTA, salts (MgCl 2 and CaCl 2 ) were added with final concentrations of 0.5 mM and 0.75 mM, respectively. The temperature was increased between 20°C and 90°C, with steps of 2°C and 6 minutes equilibration per step. Intrinsic tryptophan fluorescence was excited at 295 nm (bandwidth of 2 nm) and emission was detected at 337 nm (bandwidth of 5 nm). The fraction of folded protein was calculated as described in (Pace and Scholtz 1997)

Bioinformatics
Prediction of the structure of the X-Tfi XAC2610 (54-267)/X-Tfe XAC2609 (1-194) complex was performed using the UCSF ChimeraX (v1.4, 2022-06-03) software that includes an integrated link to the ColabFold-AlphaFold2 suite (Mirdita et al. 2022;Jumper et al. 2021;Varadi et al. 2022). The search for homologous proteins was performed using the Blast algorithm (Dooley 2004) with X-Tfi XAC2610 (54-267) as a query against refseq NCBI protein data bank using an expect values threshold of 1x10 -7 , and outliers and sequences with 99% redundancy were manually discarded. Next, the sequences were realigned with Muscle (Edgar 2004) using the X-Tfi XAC2610 sequence as a reference. The realigned sequence file (S1 Supplementary File) was used as input for the Weblogo software (Crooks et al. 2004) to create the conservation profile.

Citrus canker assay
Citrus canker assays were performed as previously described ). Briefly, X. citri inocula were grown for 12 hours in 2xTY supplemented with ampicillin. After growth, X. citri cells diluted in PBS buffer (137 mM NaCl, 2.7 mM KCl, 8 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 ) to an OD 600nm of 0.1, and 100 µl used to infiltrate sweet orange leaves (Citrus sinensis (L.) Osbeck) using a syringe with a needle. Plants were maintained at 28 o C with a 12 hour photoperiod, and symptom development was regularly observed and recorded.

X. citri vs X. citri competition assays
For the bacterial competition assay based on viability, X. citri harboring pBBR5GFP or pBBR2RFP were grown at 30 o C for 12 hours in 2xTY supplemented with gentamicin (20 µg/mL) or kanamycin (50 µg/mL), respectively. After three washing steps with fresh 2xTY, the X. citri cultures at OD 600nm of 2.0 were mixed 1:1, and five microliters were spotted in 1% LB-agar plates supplemented with ampicillin (100 µg/mL) for 40 hours at 30°C. Finally, the colonies were retrieved and resuspended in 2xTY medium, and the cellular viability per colony was estimated by serial dilution using selective medium supplemented with kanamycin or gentamicin in LB agar plates.

Results
The X-T4SS immunity proteins are not the primary defense against trans-intoxication (fratricide) Figure S1 shows that X. citri is able to kill E. coli in an X-T4SS dependent manner, as has been previously shown (Souza et al. 2015;Oliveira et al. 2016;Oka et al. 2022). Figure 2A shows CPRG-based colorimetric assays to quantitatively monitor real-time killing of E. coli cells in X. citri/E. coli co-cultures.
As previously shown, the ΔVirB7 and X. citri Δ8Δ2609-GFP strains do not kill E.
coli Oka et al. 2022) demonstrating that the X-T4SS presents antibacterial activity and that this activity is dependent on the presence of a cohort of secreted effectors (X-Tfes). On the other hand, the deletion of X-Tfi XAC2610 or the X-Tfe XAC2609 /X-Tfi XAC2610 pair does not significantly impair the antibacterial function of the X-T4SS under the conditions tested. In order to test whether the trans-intoxication (fratricide) hypothesis ( Fig 1A) is valid for the X-T4SS, we performed intraspecies bacterial competition assays using wild-type X. citri against its derivative mutants. Fig 2B shows that the X-T4SS fails to confer a competitive advantage against the X. citri ΔvirB7 strain that does not produce a functional X-T4SS due to the absence of the VirB7 subunit (Souza et al. 2015;Oliveira et al. 2016;Sgro et al. 2018). This result itself is not inconsistent with the trans-intoxication hypothesis since the target X. citri ΔvirB7 strain still carries a full set of X-Tfis. However, wild-type X. citri cells were also unable to kill the ∆X-Tfe XAC2609 /∆X-Tfi XAC2610 double-mutant X. citri strain which lacks the X-Tfe XAC2609 -X-Tfi XAC2610 toxin-antitoxin pair ( Figure 2B) nor do they kill the Δ8Δ2609-GFP X. citri strain in which eight other X-Tfe/X-Tfi effector/immunity protein pairs were deleted (XAC2885/XAC2884, (Oka et al. 2022)). The observation that wild-type X. citri is unable to kill the ∆X-Tfe XAC2609 ∆X-Tfi XAC2610 or X. citri Δ8Δ2609-GFP strains indicate that the primary role of X-T4SS immunity proteins is not to neutralize exogenous effectors that were injected by neighboring bacteria (trans-intoxication) as exemplified in Fig 1A. X-Tfi XAC2610 provides immunity against in vivo intracellular autolytic activity of X-Tfe XAC2609 As the above results are inconsistent with the proposition that X-Tfis provide protection against trans-intoxication, we performed experiments to test the hypothesis that they instead provide protection against self-intoxication (cis-intoxication), as illustrated in Fig 1B. In the case of effectors that normally act in the periplasm of target cells (for example, lysozyme-like effectors), cis-intoxication may arise if the effector is transferred across the inner membrane and into the periplasm of the producing cell by one or more routes, either dependent or independent of the X-T4SS. To test this hypothesis, we created X. citri strains with single or multiple in-frame deletions of genes coding for the X-Tfe XAC2609 lysozyme-like effector, its cognate immunity protein X-Tfi XAC2610 and the VirB7 and VirD4 subunits that are essential for X-T4SS function.

Figure S2
show that colonies of X. citri wild-type, ΔX-Tfi XAC2610 , that produces a cytoplasmic version of X-Tfi XAC2610 . Importantly, no reduction in colony opacity was observed after 72 hours of growth for the double mutant ΔX-Tfe XAC2609 ΔX-Tfi XAC2610 , indicating that X-Tfe XAC2609 must be present for the development of the observed phenotype (Fig. S2A). Also of interest, the fact that a cytossolic version of X-Tfi XAC2610 His-22-267 protects against X-Tfe XAC2609 toxicity suggest that complex formation in the cytoplasm may impede leakage into the periplasm. Finally, transparent colonies are observed even in the absence of VirB7 (ΔvirB7), demonstrating that this phenotype does not depend on a functional X-T4SS, but rather results from the absence of the immunity gene. The colonies phenotypes shown were further validated using a convolutional neural network (CNN) analysis (Fig S2).
All genes studied in this manuscript are located in a single genomic locus, and knock-outs could potentially interfere with protein expression levels of nearby genes. To discard this possibility, Western blot assays were performed. These experiments showed that only the expression levels of the targeted genes are affected by the genetic manipulations (Fig. S2B).

Figure 2C
shows that after 48 hours of growth on LB agar, no difference in cell counting or viability between wild-type and ΔX-Tfi XAC2610 strains was observed. However, after 72 and 96 hours of growth, the ΔX-Tfi XAC2610 strain shows lower viability compared to the wild-type strain. The viability of the ΔX-Tfi XAC2610 strain was restored when complemented with an extrachromosomal plasmid expressing X-Tfi XAC2610 .
In conclusion, these results indicate that i) the transparent colony phenotype is due to the activity of X-Tfe XAC2609 , ii) it can be inhibited by the presence of X-Tfi XAC2610 and iii) the phenotype does not require a functional X-T4SS. These data are consistent with a cis-intoxication mechanism by X-Tfe XAC2609 that is inhibited by X-Tfi XAC2610 .

X-Tfi XAC2610 is important for maintaining the integrity of the X. citri cell envelope
The transparent colony phenotype described above suggests that, in the absence of its cognate immunity protein, the X-Tfe XAC2609 lysozyme-like effector induces cell autolysis. To test this hypothesis, we performed a set of time-lapse microscopy assays (Fig 3 and Movies S1-5) using wild-type and different X.
citri mutant strains growing in media supplemented with propidium iodide (PI), a fluorescent dye that only stains nucleic acids when cell envelope integrity is compromised. We observed a significant increase in PI permeability in ΔX-Tfi XAC2610 cells (Fig 3, Movie S2, Table S4) compared to wild-type cells (Fig   3, Movie S1. Table S4). Permeability is reduced to wild-type levels in the ΔX-Tfe XAC2609 ΔX-Tfi XAC2610 double mutant strain (Fig 3, Movie S3, Table S4).
Transformation of ΔX-Tfe XAC2609 ΔX-Tfi XAC2610 with a plasmid over-expressing the catalytic N-terminal domain of X-Tfe XAC2609 (1-306), which lacks the XVIPCD X-T4SS secretion signal, greatly increased the amount of PI-permeable cells (Fig 3, Movie S4, Table S4). Moreover, multiple cell autolysis events were observed for the ΔX-Tfi XAC2610 ΔvirD4 X. citri strain (Fig 3, Movie S5, Table S4), confirming that toxicity caused by X-Tfe XAC2609 does not require the XVIPCD secretion signal and is not mediated by the X-T4SS. Close inspection of Movies S2, S4, S5 and S6 and Figure 4A shows that PI-permeability of X. citri cells coincides with a rapid change in cell morphology, from natural rod to spherical, consistent with X-Tfe XAC2609 -induced weakening of the cell wall. The onset of PI permeability was observed to occur both in isolated cells and in cells in contact with neighbors (Movie S6 and Figure 4A). Another interesting observation is that the onset of PI-permeability was frequently observed in cells that were undergoing cell-division (Movie S6), perhaps coinciding with a phase in the cell cycle where peptidoglycan integrity is more susceptible to the deleterious hydrolytic activity of X-Tfe XAC2609 .
We then used transmission electron microscopy (TEM) in order to obtain a more detailed picture of the differences in the cell envelope ultrastructure of X.
citri wild-type and ΔX-Tfi XAC2610 cells (Fig. 4B). TEM analyses of thin sections of previously fixed X. citri cells embedded in resin were used to assess structural details of the plasma membrane, cell wall and intracellular content. TEM micrographs of X citri ΔX-Tfi XAC2610 cells showed that they were frequently broken open with leakage of filamentous materials or were devoid of cellular contents ( Fig. 4B and S3, S4). In contrast, wild-type cells typically present an intact cell envelope and a high-density intracellular environment ( Fig. 4B and   Fig. S3, S4). Micrographs of the ΔX-Tfi XAC2610 ΔVirB7 strain presented a phenotype similar to that observed for ΔX-Tfi XAC2610 while micrographs of ΔVirB7 and ΔX-Tfi XAC2610 c lineages showed mostly intact cells as observed for the wild-type strain ( Fig. 4B and Fig. S4). Analysis of the number of intact versus damaged cells in TEM micrographs indicates significant statistical differences in the percentage of lysed cells in strains producing both X-Tfe XAC2609 and X-Tfi XAC2610 (wild-type, ΔVirB7 and ΔX-Tfi XAC2610 c) compared with cells expressing the former but not the latter (ΔX-Tfi XAC2610 and ΔX-Tfi XAC2610 ΔVirB;  Table). Together, fluorescence microscopy and TEM analyses indicate that, in the absence of X-Tfi XAC2610 , the activity of X-Tfe XAC2609 promotes damage of the X. citri cell envelope, regardless of the presence of a functional X-T4SS.

The hydrolytic activity of X-Tfe XAC2609 inhibits X. citri biofilm formation in the absence of X-Tfi XAC2610
The results so far show that the absence of X-Tfi XAC2610 compromises the integrity of the X. citri cell envelope of only a small number of actively dividing cells during the exponential growth phase (Fig. 3) and eventually leads to the lysis of a significant fraction of the cell population during stationary phase ( Fig  S2). Since bacterial biofilms are a complex array of cells and extracellular polymers that develop over extended periods of time (many hours to days) with a progressive reduction in the rate of cell division (Dunger et al. 2014(Dunger et al. , 2016Sena-Vélez et al. 2016), we decided to investigate the possible physiological effects associated with the deletion of X-Tfi XAC2610 on biofilm formation. Figure   5A shows that, unlike the X. citri wild-type strain, ΔX-Tfi XAC2610 and  (Souza et al. 2015) was able to form a normal biofilm. These results confirm that, in the absence of its cognate inhibitory protein, the deleterious effects of X-Tfe XAC2609 is dependent on its glycohydrolase activity.
We then asked if specific individual components of the X-T4SS are necessary for the translocation of X-Tfe XAC2609 to the periplasm (Pathway (1) in Figure 1B). This is a a relevant hypothesis considering that cytoplasmic substrates of the T4SS are known to interact with various T4SS subcomplexes along the secretion pathway (Cascales and Christie 2004a;Atmakuri, Cascales, and Christie 2004;Cascales and Christie 2004b). To address this question, we deleted the genes coding for several X-T4SS subunits that are associated with the bacterial inner membrane (VirB4, VirB6, VirB8 and VirD4) (Macé et al. 2022), the outer membrane (VirB7, VirB9 and VirB10) (Fronzes, Christie, and Waksman 2009;Chandran et al. 2009;Souza et al. 2011;Oliveira et al. 2016;Sgro et al. 2018) and the VirB5 subunit believed to form part of the extracellular pilus (Alvarez-Martinez and Christie 2009; Christie, Whitaker, and González-Rivera 2014; Sheedlo et al. 2022). These deletions were introduced in both the X. citri wild-type and ΔX-Tfi XAC2610 genetic backgrounds. Growth of these strains in 24-well plates for 24 hours with agitation followed by five days without agitation revealed that all X. citri strains lacking X-Tfi XAC2610 cannot form biofilm, independent of the presence or absence of any X-T4SS structural component (Fig. 5B, Fig. S5). Taken together, these results show that, in the absence of X-Tfi XAC2610 , cis-intoxication by X-Tfe XAC2609 inhibits biofilm formation in a manner that is independent of any subunit or subassembly of the X-T4SS apparatus.
Finally, as X. citri is the causal agent of citrus canker, we asked whether the absence of X-Tfi XAC2610 could affect the ability of this phytopathogen in causing disease. Figure S6 shows that X. citri ΔX-Tfi XAC2610 can induce the appearance of citrus canker lesions in sweet orange leaves in a manner very similar to those caused by the wild-type strain. This phenotype was previously shown to be dependent on a type III secretion system (Cappelletti et al. 2011) but independent of the X-T4SS ). Figure S7 shows the sequence conservation profile derived by the multiple sequence alignment of 429 non-redundant X-Tfi XAC2610 homologs (Supplementary File 1) and reveals several conserved motifs. One conserved motif is a region with several negatively charged amino acids corresponding to X-Tfi XAC2610 residues 151-159 that form a Ca 2+ -binding loop in the crystal structure of X-Tfi XAC2610 (Souza et al. 2015). Figure S8 shows that the presence of Ca 2+ ions significantly increases the thermal stability of X-Tfi XAC2610 . This result suggests that the family of X-Tfi XAC2610 homologs could all be stabilized by divalent cation binding at this site.

X-Tfi XAC2610 complex
To investigate the molecular mechanism of the immunity provided by X-Tfi XAC2610 against X-Tfe XAC2609 activity, we predicted the structure of the X-Tfe XAC2609 /X-Tfi XAC2610 complex by AlphaFold2 (Mirdita et al. 2022;Varadi et al. 2022;Jumper et al. 2021) using the sequences of X-Tfi XAC2610 lacking the N-terminal signal peptide and the N-terminal GH10 glycohydrolase domain of X-Tfe XAC2609 . Figure 6A shows that the best model of the X-Tfi XAC2610 (54-267)/X-Tfe XAC2609 (1-194) complex superposes with the previously determined crystal structure of X-Tfi XAC2610 (Souza et al. 2015). In this model, the interface between the two proteins involves another well-conserved motif in the X-Tfi XAC2610 family that includes a loop made up of residues 165-173 ( Figure   6A-B and Figure S9). In the predicted X-Tfe XAC2609 /X-Tfi XAC2610 complex, a highly conserved tyrosine (Y170) in this loop of X-Tfi XAC2610 directly interacts with the catalytic aspartate (E48) in the active site of X-Tfe XAC2609 (Fig. 6C, Fig. S9 and S10A). To test the hypothesis that Y170 is involved in the mechanism of X-Tfi XAC2610 -mediated inhibition of X-Tfe XAC2609 , we carried out in vitro peptidoglycan hydrolysis assays using X-Tfe XAC2609 in the absence or presence of wild type X-Tfi XAC2610 or the X-Tfi XAC2610 Y170A mutant. Figure 6D shows that the wild-type version of X-Tfi XAC2610 inhibits the hydrolytic activity of X-Tfe XAC2609 , while the Y170A mutation strongly decreases peptidoglycan degradation. The lysozyme-like effector X-Tfe XAC2609 is a cytoplasmic protein that is transported in a X-T4SS-dependent manner into other bacterial cells.

Discussion
X-Tfi XAC2610 , its cognate inhibitor, has a lipoprotein box motif for localization to the cell periplasm (Souza et al. 2015). We observed that cells expressing a functional X-Tfe XAC2609 in the absence of X-Tfi XAC2610 completely abolished bacterial biofilm formation and that biofilms are still not formed by cells with this genetic background when essential X-T4SS components are knocked out.
These results show for the first time that inhibition of PG hydrolases by immunity proteins can be required for bacterial biofilm formation. Accordingly, another example of the relationship of PG stability and biofilm formation was Two general, not necessarily exclusive, hypotheses were proposed to account for X-Tfe XAC2609 -dependent toxicity in the absence of X-Tfi XAC2610 . The first, which we call trans-intoxication or fratricide, hypothesizes that X. citri cells inject a cocktail of X-Tfes, including X-Tfe XAC2609 , into neighboring X. citri cells and, in the absence of at least one cognate immunity protein, could cause cellular injury and consequent growth suppression or cell death.
Trans-intoxication has been shown to occur in P. aeruginosa cells via the H1-T6SS-dependent secretion of the PG hydrolases Tse1 and Tse3 (Russell et al., 2011). Also, in Vibrio cholerae, T6SS-associated immunity proteins have been shown to be important in preventing trans-intoxication (Dong et al. 2013).
However, in the case of the X-Tfe XAC2609 -dependent toxicity observed in the absence of X-Tfi XAC2610 , a trans-intoxication mechanism is not consistent with the observation that both the detrimental effect of X-Tfe XAC2609 and protection conferred by X-Tfi XAC2610 are independent of a functional X-T4SS and that wild-type X. citri does not outgrow X. citri strains lacking X-Tfi XAC2610 or several other immunity proteins in competition assays ( Figure 2B and movie S7).
Furthermore, X-Tfe XAC2609 -dependent autolysis in the absence of X-Tfi XAC2610 does not require the former's C-terminal XVIPCD X-T4SS secretion signal.
Thus, we can discard the trans-intoxication mechanism or at least propose that this pathway contributes only a small fraction of the X-Tfe XAC2609 -dependent toxic effects observed in the absence of X-Tfi XAC2610 . Nevertheless, the fact that wild-type X. citri is unable to kill strains lacking immunity proteins is intriguing.
That cells in some way avoid trans-intoxication is revealed by the fact that X.
citri wild-type cells carrying an X-T4SS and full cohort of X-Tfes do not kill the X.
citri Δ8Δ2609-GFP, the X. citri ∆X-Tfe XAC2609 ∆X-Tfi XAC2610 , or any other X-T4SS-deficient strain tested points to a still-to-be-characterized mechanism of protection against trans-intoxication (fratricide) that will be addressed in future studies by our group.
We are thus left to consider the cis-intoxication hypothesis to explain the toxicity of X-Tfe XAC2609 and the protection conferred by X-Tfi XAC2610 . Here, the effector exerts its toxic effects within the cell in which it was synthesized. Two cytoplasmic P. aeruginosa T6SS effectors, Tse2 and Tse6, have been shown to cause cis-intoxication in a T6SS null strain when immunity is depleted (M. Li et al. 2012;Whitney et al. 2015). In the case of X-Tfe XAC2609 , the toxin somehow makes its way into the cell periplasm where, in the absence of X-Tfi XAC2610 , it degrades the peptidoglycan layer. Analysis of the X-Tfe XAC2609 sequence by the SignalP 6.0 (Teufel et al. 2022) and other algorithms failed to detect any putative N-terminal signal peptide. Although the mechanism responsible for X-Tfe XAC2609 transfer into the periplasm is at the moment unknown, we have shown that it is independent of a functional X-T4SS and of the XVIPCD secretion signal. Other bacterial proteins have been shown to transfer into the periplasm without any obvious secretion signal, for example VgrG3 from Vibrio cholerae (Ho et al. 2017), CI2 and in HdeA expresssed in E.coli (Barnes and Pielak 2011).
We have previously pointed out that the structure of X-Tfi XAC2610 adopts a similar β-propeller fold topology to two other peptidoglycan hydrolase inhibitors: the P. aeruginosa Type VI immunity protein Tsi1 and the Aeromonas hydrophila periplasmic i-type lysozyme inhibitor PliI-Ah (Souza et al. 2015). Interestingly, both of these inhibitors block the activity of their respective targets by inserting an exposed loop into the enzyme's active site, in a manner similar to that predicted for the X-Tfe XAC2609 -X-Tfi XAC2610 complex (Fig. S10); this in spite of the fact that they share very low sequence identity with X-Tfi XAC2610 (7% for Tsi1 and 12% for PliI-Ah; (Souza et al. 2015). In the case of PliI-Ah, its crystal structure in complex with the i-type lysozyme from Meretrix lusoria (Ml-iLys) revealed a complementary key-lock interface through the interaction of an exposed loop of PliI-Ah into the substrate-binding groove of Ml-iLys (Herreweghe et al. 2015).
Comparison of the topology diagrams of X-Tfi XAC2610 , PliI-Ah and Tsi1 shows that the loops they use to interact with their cognate targets are topologically equivalent (Fig. S11). While X-Tfi XAC2610 is predicted to use Y170 to interact directly with E48 in the active site of X-Tfe XAC2609 , in Tsi1 a serine residue (S109) of its inhibitory insertion loop directly interacts with H91 in the active site of the target Tse1 PG amidase enzyme (Benz et al. 2012) (Fig. S10). In a similar manner, in PliI-Ah, a serine residue (S104) directly recognizes the active site residue E18 of the inhibited lysozyme (Herreweghe et al. 2015) (Fig. S10).
The structural similarities in the inhibitory mechanisms of X-Tfi XAC2610 , Tsi1 and PliI-Ah are intriguing. While X-Tfi XAC2610 and PliI-Ah inhibit PG glycohydrolases, Tsi1 inhibits the PG amidase activity of Tse1. It is also worthy to note that whereas X-Tfi XAC2610 inhibits the cis-intoxication activity of its cognate effector, the biological function of Tsi1 was described to protect P.
aeruginosa cells from the deleterious effects of Tse1 molecules transferred from neighboring cells via the H1-T6SS, thus avoiding fratricide or trans-intoxication (Russell et al. 2011   Trans-intoxication. In this mechanism, intoxication is due to contact-and T4SS-dependent transfer of X-Tfes (effectors) from one cell to another. Left: G enetically identical cells with equivalent repertoires of X-Tfes and cognate X-Tfis (immunity proteins) would be protected. In the scheme shown here, two wild-type cells that produce two different effector-cognate immunity protein pairs (orange and blue; e: effectors and i: immunity proteins) are immune against the toxic effects of the X-T4SS-mediated trans-intoxication due to the protective role of cognate immunity proteins. Right: Encounters between cells with non-equivalent repertoires would lead to killing. In the scheme shown here, a wild-type cell that produces two different effector-cognate immunity protein pairs (blue and orange) is in contact with a mutant cell that produces only one effector-immunity protein pair (blue). The hypothesis is that the two cells transfer effectors into each other's periplasm and since the prey cell lacks the immunity protein that inhibits the orange effector, its cell wall is susceptible to degradation. (B) Cis-intoxication. In this mechanism, instead of being transported outside of the cell by the X-T4SS, an effector is translocated into the periplasm.
Translocation could be T4SS-dependent (1) or T4SS-independent (2). Left: A wild type cell carrying a complete set of cognate immunity proteins is protected against self intoxication. Right: A bacterial strain lacking the immunity protein (ΔImmunity protein), may be susceptible to the cumulative activity of an effector that leaks into the periplasm. Cytoplasm (C); Inner membrane (IM); outer membrane (OM); Periplasm (P).   Table S4. Scale bar 10 µm.  Figures 3 and 4. X. citri strain used for the assay are described in the figure legends.
Means with a double asterisk are significantly different at P < 0.05 compared with the mean of the wild type, Tukey post hoc one-factor ANOVA test. Further details regarding quantitative and statistical analysis are shown in Table S5.   Forward primer to amplify the upstream region of virB9 (Sgro et al., 2018) AGTCACTAGTAGCTCTAAGGGCTGCCGTC

R_DOWN_VirB10
Reverse primer to amplify the downstream region of virB10 (Sgro et al.,  Suicide vector for delection of virB4 in the X. citri ∆VirB4 , Km r This study pNPTS138-VirB5 Suicide vector for delection of virB5 in the X. citri ∆VirB5 , Km r This study pNPTS138-VirB6 Suicide vector for delection of virB6 in the X. citri ∆VirB6, Km r This study pNPTS138-VirB7 Suicide vector for delection of virB7 in the X. citri in ∆VirB7 ∆/ Km r

Figure S3. Transmission electron microscopy (TEM) of X. citri ∆X-Tfi XAC2610 cells.
Insets highlight cells having an impaired enveloped cellular (arrows).      the crystal structure of Tsi1-Tse1 complex from (PDB 3VPJ) (Ding et al. 2012). In A, B and C, the inhibitors are colored according to the degree of conservation at each residue position in the corresponding family of homologs (lowest, cyan; highest, purple) and the cognate enzymes are colored in gray. Right panels: Zoom of the interaction interfaces showing the insertion of the inhibitory loops into the active sites of the enzymes. Stick models highlight the interactions between residues in the inhibitory loops (X-Tfi XAC2610 Y170 , PliI-Ah S104 and Tsi1 S109) and the enzyme active sites (X-Tfe XAC2609 E48, MI-iLys E18, Tse1 H91). D: Sequences in the inhibitory loops of X-Tfi XAC2610 , Plil-Ah and Tsi1 that interact directly with the catalytic site of the corresponding toxins. Underlined are X-Tfi XAC2610 Y170 , PliI-Ah S104 and Tsi1 S109. ) and (C) Tsi1 (PDB 3VPJ). Diagrams were generated using the PDBsum server (http://www.ebi.ac.uk/thornton-srv/databases/cgi-bin/pdbsum/GetPage.pl?pdbcode=ind ex.html). Secondary-structure elements are indicated as red cylinders (α-helices) and pink arrows (β-strands). The dotted-squares highlight the common β-sheet found in the three immunity proteins containing a loop between the second and third β-strands that inserts into the active site of the cognate enzyme.