The genome sequences of the three citrus-infecting CaL species [Las (CP001677.5), CLaf (CP004021.1) and CLam (CP006604)] are available at the National Center for Biotechnology Information (NCBI) database. Protein sequences, based on the homology to Las1345, were downloaded from NCBI under the following accession numbers: serralysin-like proteins from CLas strain psy62 (WP_015452346), from CLaf strain PTSAPSY (AKK19938.1) and from CLso strain ZC1 haplotype B (WP_013461860); PrtA from S. marcescens (CAA39139.1), Ser1 from S. liquefaciens FK01 (BAK39731) and PrtC from Dicheya chrysanthemii B374 (WP_038909783.1). The sequences were aligned using Clustal Omega and MultiAlin (Corpet, 1988; Sievers and Higgins, 2018), following a manual adjustment in the C-terminal sequence, considering the sequence pattern associated with the ABC exporter motif. Las1345 was homology modeled using Swiss Model server (Waterhouse et al., 2018).

Bacterial strains and plasmids are described in Table 1. Rhizobium leguminosarum bv. viciae A34 (A34), S. marcescens mutant in prtA (prtA), Xanthomonas campestris pv. campestris (Xcc) wild type and the gumB mutant strains were grown in tryptone-yeast (TY), Miller’s Luria-Bertani (LB) and peptone-yeast-malt (PYM) medium, respectively (Russo et al., 2006; Torres et al., 2007; Bruna et al., 2018). Plasmids were mobilized into R. leguminosarum bv. viciae A34 and S. marcescens by biparental mating using Escherichia coli S17 cultured LB, at 37°C (Vozza et al., 2016; Bruna et al., 2018). Conjugation was performed at 28°C and 37°C for R. leguminosarum and S. marcescens, respectively. Xcc was transformed by electroporation (Rigano et al., 2007). Bacterial growth was monitored at an optical density of 600 nm (OD600).

CLIBASIA_01345 was amplified from CLas-infected plant tissue via polymerase chain reaction (PCR) and subsequently cloned into p15TV-L (Pagliai et al., 2010) to obtain p15::HisLas1345. This vector was used as template to obtain pBBR2::HisLas1345 by restriction sites cut with NcoI and HindIII. The plasmid p15::HisLas1345 was also used as template to amplify tagless CLIBASIA_013145 to generate pBBR2::Las1345 using the following primers: 5-GTCGGTACCATGCATAATATAAAACCGG-3/5-GTCAAGCTTTCAGGAAAAATCATGATTTA-3. To express Las1345 in plants, CLIBASIA_01345 was amplified from p15:HisLas1345, adding KpnIATG/HindIII sites with the following primers 5-GTCGGTACCATGCATAATATAAAACCGG-3 and 5-GTCAAGCTTTCAGGAAAAATCATTTA-3 and cloned into pENTR3c-Las1345 to generate pENTR3c::Las1345. This vector was amplified with AHL primers (5-TAGTTAGTTACTTAAGCTCGGGC-3/5-CAGAGCTGCAGCTGGATGGC-3) and the PCR product was transferred into binary vectors (pMDC83 and pEarlyGate100) using Gateway LR Clonase II following manufacturer instructions (Thermo Fisher Scientific, Waltham, United States).

The recombinant plasmid p15::HisLas1345 was used to transform E. coli ArticExpress competent cells for further expression and purification. Cells were grown at 30°C until OD₆₀₀ ~ 0.7. Induction was made using 0.5 mM IPTG and 6-h incubation at 19°C. The cell pellet was suspended in binding buffer (50 mM HEPES pH 7.5, 250 mM NaCl, 5% (v/v) glycerol, 2.5 mM TCEP and 5 mM Imidazol). The thawed cells were passed through a French Press (French pressure cell press 40 K, Thermo Scientific). Three runs at 800 pressure gauge (Phcp) were conducted for each sample. Samples were released at 15 drops/min. The clarified lysed was purified using Ni-NTA resin (Qiagen, Germantown, United States) as described previously (Pagliai et al., 2014). Purified protein was dialyzed against 10 mM HEPES pH 7.5, 50 mM NaCl, 10 mM MgCl₂, 2 mM CaCl₂, 5% (v/v) glycerol and 0.25 mM TCEP. Protein was concentrated using Vivaspin centrifugal concentrators (Sartorius, Bohemia, United States) treated with Triton X-100 and the purity was analyzed in gel. The purified protein was refolded by dilution into cold buffer (50 mM Tris, 50 mM NaCl, 0–0.2 mM CaCl₂, pH 7.8) on ice for 20 min prior to activity. The proteolytic activity of the purified protein (pHisLas1345) was evaluated under different amount of protein and the addition of divalent cations (Zn²⁺, Co²⁺, Cu²⁺ and Ca²⁺) in the chloride salt form (Supplementary Figure 1).

For analysis of secreted and intracellular proteins, bacteria (R. leguminosarum, S. marcescens and Xcc) were grown for 48 h at 28°C in TY, LB or PYM medium, respectively to an OD₆₀₀ ~ 0.6. The cell culture (3 ml) was centrifuged at 6,500 rpm and both the extracellular medium (supernatant) and the cellular precipitate were reserved for protein extraction. Supernatant (2 ml) was clarified with a 0.2 μm-syringe filter to obtained a cell free supernatant fraction (SF). Extracellular proteins were concentrated from the SF by precipitation with 10% (v/v) trichloroacetic acid (Merck, Darmstadt, Germany) and incubated on ice for 2 h. Proteins were recovered by centrifugation (12,000 rpm, 30 min) and washed twice with 500 μl acetone (Cicarelli, Santa Fe, Argentina). The precipitate was dried and the proteins were solubilized in the proper buffer for activity or immunodetection. Intracellular proteins were recovered from the soluble and insoluble cell fractions (CFs and CF_I, respectively) prepared from cellular precipitate. Cells were washed twice with 10 mM Tris, 5 mM EDTA pH 8 and lysed by sonication at a frequency of 20 kHz. The lysate was centrifuged (12,000 rpm, 30 min) at 4°C to recovered the pellet (CF_I) and the supernatant (CFs) fractions. Proteins in the CFs were precipitated with acetone (Cicarelli, Santa Fe, Argentina) on ice for 2 h and recovered by centrifugation (12,000 rpm, 30 min). CFs and CF_I pellets were resuspended in the proper buffer for activity or immunodetection.

SF, CFs and CF_I pellets were resuspended in loading buffer [60 mM Tris, 10% (v/v) glycerol, 180 mM β-mercaptoethanol, 0.003% (w/v) bromophenol blue and 2% (w/v) SDS, pH 6.8], separated by 12% sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by staining with Coomassie brilliant blue R-250 (Genbiotech, Buenos Aires, Argentina). Western blot was carried out using standard techniques. Briefly, after SDS-PAGE, proteins were transferred onto pre-wetted polyvinylidene difluoride membranes (PVDF- Immun-BlotV^R, BioRad, CA, United States). Immunodetection was performed using polyclonal antibodies against His tag (1:6,000 dilution, #SAB1306085 Sigma, Merck, Darmstadt, Germany), Rhizobium adhering-protein A1 (RapA1; 1:5,000 dilution; Vozza et al., 2016) or E. coli Glucose-6-phosphate dehydrogenase (G6PD; 1:500 dilution; Giró et al., 2006) and then revealed using a peroxidase-conjugated goat anti-rabbit IgG (H + L; 1:3,000 dilution, #1706515, BioRad, Des Plaine, IL, United States) and Pierce™ ECL Western Blotting Substrate (Thermo Scientific, Petaluma, CA, United States) according to the manufacturers. Chemiluminescence was detected using an CCD camera (ChemiDoc™ XRS⁺, BioRad, Des Plaine, IL, United States). For probe stripping and rehybridization, the filters were incubated with stripping buffer (0.05 M Tris, 1% (w/v) SDS, 0.8% β-mercaptoethanol, pH 6.8) for 45 min at 30°C and washed five times with TBS (20 mM Tris, 150 mM NaCl, pH 7.6) supplemented with 0.05% (v/v) Tween (Promega, Fitchburg, WI, United States).

Arabidopsis thaliana and Nicotiana benthamiana were grown under controlled conditions in a growth chamber with a temperature of 25°C–27°C and a photoperiod of 16 h light/8 h dark and light intensity of 150–200 μE/s m². All plant inoculations involved a minimum of 10 plants per strain tested and one leaf per plant was inoculated. Bacterial suspensions of Xcc (10⁷ CFU/ml in 10 mM MgCl₂) were inoculated by pressure infiltration (Siciliano et al., 2006). Disease progression (water-soaking and necrosis) was monitored phenotypically and by bacterial growth curves. Bacterial population was determined in five samples according to the method described by Rigano et al. (2007). Each sample was obtained from three leaf disks of 1 cm² collected randomly from different inoculated leaves. Symptom development were assayed by conductivity measurements using leaf disks submerged in milli-Q water following protocols previously described (Chiesa et al., 2019). Six samples were measured, each one obtained from three leaf disks of 1 cm² collected randomly from 10 inoculated leaves. Photosystem II quantum efficiency (ϕPSII) was measurements at 25°C on one dark-adapted leaf from 15 plants using MultispeQ V 2.0 (PhotosynQ INC, East Lansing, MI, United States). These assays were repeated three times.

To evaluate the function of Las1345 in plants, N. benthamiana leaves were agroinfiltrated with A. tumefaciens GV3101 transformed with the corresponding plasmid (Table 1), to overexpress Las1345 with and without a GFP tag. Molecular techniques and agroinfiltration were performed as described previously (Sánchez et al., 2010; Enrique et al., 2011). GFP versions were analyzed to study protein localization by confocal laser scanning microscopy (CLSM). Nucleus and membrane were marked with DAPI (Thermo Fisher Scientific, Waltham, United States) and FM 4-64 (Sigma, Merck, Darmstadt, Germany), respectively. Images were taken 2 days after agroinfiltration with a Zeiss LSM 880 confocal laser scanning microscope (Carl Zeiss Microscopy GmBH, Jena, Germany) using the following parameters, GFP (488/527 nm), FM 4-64 (515/604 nm) and DAPI (405/449 nm). For CLSM analysis, each construct was infiltrated into two leaves of 10 plants.

For plant protein extraction, six leaf disks (1 cm²) collected randomly from different agroinfiltrated plants were pulverized in pre-chilled pestle and mortar using liquid nitrogen and the powder was resuspended in 0.5 ml of extraction buffer (50 mM Tris, 100 mM KCl, 10% (v/v) glycerol, pH 7.5). Supernatant was recovered after centrifugation (12,000 rpm, 20 min) at 4°C and kept at −80 for protease activity measurements.

Protease activity was assayed using milk or azocasein (#A2765, Sigma, Merck, Darmstadt, Germany) as substrates as described previously (Bruna et al., 2018). For qualitative analysis, the samples were inoculated on LB agar plates supplemented with skim milk at 2% (w/v) and incubated for 16 h at 30°C. Distinct clearing of the milk around the colony was used as a protease activity indicator. For quantitative analysis, protease activity was measured using azocasein as substrate. Samples of SF, CF_S and CF_I (50 μl) were mixed with 10 μl of 1% (w/v) azocasein and 140 μl of phosphate-buffered saline (PBS) and incubated for 1 h at 37°C. The reaction was stopped by addition of 80 μl of 10% (v/v) trichloroacetic acid, and the mix was incubated on ice for 15 min. Protease activity was repeated five times using six replicates of each strain. The same protocol was used to evaluate protease activity with both the purified HisLas1345 protein from E. coli and plant protein extracts. For these analyzes, measurements were normalized by protein quantification using the Bradford assay kit (BioRad, Des Plaine, IL, United States) and repeated three times.

Swimming, swarming and sliding motility assays were carried out as previously described (Malamud et al., 2011, 2013). Briefly, overnight cultures were normalized to an OD₆₀₀ ~ 0.8 and 3 μl were used to inoculate 0.25% (w/v) agar NYGB medium plates (swimming) or 0.5% (w/v) agar NYGB medium (sliding/swarming). Plates were incubated 72 h at 28°C. Motility was assessed quantitatively by measure the circular halo formed by the growing bacterial cells in 10 different culture plates. This assay was repeated three times.

Exopolysaccharide (EPS) xanthan quantification was performed as described previously (Vojnov et al., 1998). To measure EPS xanthan production, strains were grown in PYM medium supplemented with 1% (w/v) D-glucose at 28°C for 24 h with shaking (156 rpm) in 250-mL flask. EPS xanthan was precipitated from culture supernatants using 2 vol of ethanol. The precipitated EPS xanthan was collected, washed with ethanol, dried and weighed. Each assay included eight to 10 replicas and the experiment was repeated three times.

To analyze biofilm structure, GFP-expressing Xcc cells were transformed with either pBBR2::Las1345 or pBBR2 as the empty vector control. Both strains were grown at 28°C on PYM medium supplemented with 50 μg/ml kanamycin. Cultures were diluted in Y minimal medium (YMM) to reach OD₆₀₀ ~ 0.02 and aliquots of 500 μl were transferred to 8-wells chambers containing a 1-mm thick borosilicate glass (Nunc, Wiesbaden, Germany), as described by Rigano et al. (2007). Strains were incubated for 4 days at 28°C. Congo red (10 μM) was added to YMM to visualize extracellular components, such as exopolysaccharides, curli amyloid proteins and adhesins which conform the biofilm matrix (Serra and Hengge, 2017). Congo red staining has been used to detect amyloid fibers in Xanthomonas spp. (Oh et al., 2007; Kraiselburd et al., 2012). Biofilm formation was monitored using a Zeiss LSM880 confocal laser scanning microscope (Carl Zeiss Microscopy GmBH, Jena, Germany) by excitation at 488 nm with the argon laser line and 20x NA = 0.8 Plan-Aphocromat objective (Carl Zeiss Microscopy GmBH, Jena, Germany). To analyze the tridimensional structure of the biofilm, the XY area was screened in 35 Z-intervals (Z-stack) automatically captured up to a 25.5 μm range at green (488/520 nm, pinhole 0.84 A.U.) and red (543/619 nm, pinhole 0.70 A.U.) channels, respectively. Biofilm images were obtained by ZEN BLACK software (Carl Zeiss Microscopy GmBH, Jena, Germany) and analyzed with COMSTAT2 (Heydorn et al., 2000; Vorregaard, 2008).¹ Each assay includes triplicates (3 wells per strain) and it was repeated three times.

The expression levels of three (filE, fliF and flgL) flagellum assembly-related genes of Xcc were analyzed by qPCR. Xcc/Las1345 and Xcc/pBBR2 cells were grown in YMM at 28°C to reach an OD₆₀₀ ~ 0.8. The cells (10 ml) were collected and total RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific, Walthman, United States), following treatment with RNase-free DNase (Promega, Wisconsin, United States). Reverse transcription was performed using RevertAid reverse transcriptase (ThermoFisher Scientific, Carlsbad, United States), 2 μg of DNase-treated total RNA and 0.5 μg/μl random hexamers, according with the manufacturer’s instructions. cDNAs were used for qPCR using Hot FIREPol^® EvaGreen qPCR mix plus (Solis BioDyne, Tartu, Estonia) and 0.25 μM of each primer. Quantitative PCRs (qPCR) were performed for 42 cycles according to the following conditions: denaturation at 95°C for 15 s, annealing at 60°C for 30 s and extension at 72°C for 40 s. After amplification, melting curves were performed to exclude artifactual amplifications. The 16S ribosomal RNA was used as internal control. Primers were designed based on the Xcc 8004 (CP029484.1) genome sequence. Flagellar genes were amplified with the following primers, fliE_fw 5′-AGCTTCAGCGAGACCTTGCG-3′ and fliE_rv 5′-CAGATCGGCACTGGGGTCAC-3′; fliF_fw 5′-AAGTTCCAAGAGCGCCACCC-3′ and fliF_rv 5′- ACGATCTTGCCCTTGGCACC-3′; flgL_fw 5′- ATGGCAATGCGCCTTTCATC and flgL_rv 5′- GCGGATGCGCATGAAGATTT-3′; 16S_fw 5′-AGGACCTTCGGGCTT-3′ and 5′-TGTCTCAGTTCCAGTG-3′ (Yan et al., 2019).

Data were analyzed according to Student’s t-test through InfoStat Software v2017 (Di Rienzo et al., 2017), excepting protease activity in bacterial cell fractions and EPS production data that were subjected to one-way analysis of variance (ANOVA). Two-way ANOVA was used to analyze the data from protease activity in plant crude extract. In both cases, means were analyzed using Tukey’s test p < 0.05).

Genome comparison between CLas and CaL solanacearum (CLso, causing zebra chip disease in potato) revealed that the CaL serralysin-like gene (Figure 1A) is largely conserved in terms of size and localization, being next to a gene cluster encoding the T1SS (Li et al., 2012; Ravindran et al., 2018). In this work we have included sequence comparison with the other two HLB-associated CaL species, africanus strain PTSAPY (CLaf) and americanus strain São Paulo (CLam; Wulff et al., 2014; Lin et al., 2015). A putative serralysin gene was found in CLaf genome (Figure 1B). Comparative protein sequence analysis indicates that Las1345 (665 amino acids, aas) has 63.77 and 65.67% amino acid identity with the serralysin-like protein of CLsoB (577 aas) and CLaf (642 aas), respectively. These two proteins are longer than the characterized serralysins from Serratia spp. (PrtA and Ser1, 504 aas each) and Dickeya chrysanthemi (former Erwinia; PrtC, 479 aas; Figure 1B). Las1345 share 30.41%, 31.23%, and 31.30% amino acid identity with PrtA, Ser1 and PrtC, respectively. Las1345 conserves the N-terminal metzincin-type metalloprotease motif HExxHxxGxxHP, including the last proline, characteristic for serralysins. The three histidines in this motif are zinc ligands and the glutamic acid serves as catalytic base (Baumann, 1994, 2019). Twenty-three residues downstream from this motif, the conserved met-turn motif (SxMSYF/W) is found. Three-dimensional prediction shows that the N-terminal part of Las1345 has a folding topology similar to PrtA (Figure 1C). PrtA, Ser1 and PrtC have four conserved motifs (GGxGxDxUx) in the RTX (repeats-in-toxins) domain (Figure 1B). T1SS substrates that have these specific repeats belong to the RTX- protease family (Spitz et al., 2019; Hui et al., 2021). Three-dimensional structure of the RTX domain of PrtA shows that this domain consists of four parallel β-rolls, where two GG repeats coordinate one Ca²⁺ ion by interaction of the side chain of the aspartate residues and the carbonyl oxygens of the amino acids forming the repeat. Calcium ion binding is implicated in secretion and extracellular folding of the protein (Baumann, 2019; Spitz et al., 2019; Hui et al., 2021). Unlike PrtA, Las1345 has only one non-canonical RTX repeat, GSSGND, located between G₃₇₀ and D₃₇₅. It has been hypothesized that this consensus motif, instead of the typical nonapeptide repeat, may also be considered as a potential binding site for Ca²⁺ (Linhartová et al., 2010). Interestingly, the GSSGND motif in the RTX domain of Las1345 also folds in a right-handed β-roll (GSS) and connects a short strand as in PrtA. However, this RTX repeat is not recognized as a Ca²⁺ binding site by the modeling server (Figure 1C). Moreover, prior to the predicted non-cleaved C-terminal T1SS secretion signal (DFS) for serralysins, Las1345 exhibits a duplicated 84-residue sequence (Figure 1B).

These differences between Las1345 and characterized serralysins such as PrtA may influence Las1345 secretion via the T1SS and extracellular folding. However, the metzincin-type metalloprotease domain and the non-classical RTX repeats still supports the possibility of protease activity in Las1345.

CaL is a member of the order Rhizobiales and it is phylogenetically related to the Rhizobium-type genus (Young et al., 2001; Duan et al., 2009; Kuykendall et al., 2012a). In order to study the putative secretion of Las1345, the symbiont R. leguminosarum was used as surrogate model, considering the T1SS similarity between CaL spp. and Rhizobium (Duan et al., 2009; Kuykendall et al., 2012a). We first evaluated the putative secretion of Las1345 using cell fractionation in R. leguminosarum bv. viciae strain A34 (A34; Figure 2A). RapA1, an extracellular T1SS-secreted and surfaced associated protein from R. leguminosarum (Vozza et al., 2016) and the bacterial intracellular G6PD protein (Giró et al., 2006) were used as subcellular fractionation controls. An N-terminal His-tagged version of Las1345 (HisLas1345) was only detected in the cellular fraction of R. leguminosarum, suggesting either that the C-terminal domain of this protein is not recognized by the inner membrane ABC transporter of the T1SS of R. leguminosarum or that the N-terminal tag prevents Las1345 secretion to the extracellular medium (Figure 2A). Interestingly, accumulation of HisLas1345 in cellular fraction correlated with an increased protease activity, measured using azocasein as substrate (Figure 2B). To discard a tag-effect during the secretion process, an untagged version of Las1345 was expressed in R. leguminosarum and protease activity was measured in the supernatant fraction. In both tagged and untagged versions, Las1345 did not increase proteolytic activity in the supernatant fraction of A34, suggesting that Las1345 might not be secreted, at least in this surrogate system (Figure 2C). Moreover, in both versions of Las1345, an increase in proteolytic activity was detected both in the soluble and insoluble cellular fractions (CF_S and CF_I). Interestingly, proteolysis was increased by more than 50% in CF_S of both A34/Las1345 (100 ± 3.15) and A34/HisLas1345 (78.45 ± 3.42) compared to the A34/pBBR2 (9.67 ± 2.33) and only 10%–20% in CF_I (Figure 2C).

As shown in Figure 1, PrtA from S. marcescens shows homology with Las1345, mainly in the catalytic domain. Therefore, a S. marcescens prtA mutant strain (prtA) was transformed with pBBR2::HisLas1345 and pBBR2::Las1345 to evaluate Las1345 secretion and its proteolytic activity in this surrogate model. prtA complemented by trans expression of prtA from the pBBR2::prtA plasmid or transformed with pBBR2 were used as positive and negative controls, respectively. Similarly to the findings observed in Rhizobium, HisLas1345 was only detected in the cellular fraction of prtA (Figure 3A) in contrast with the positive control (PrtA) which was detected in the supernatant fraction (Figure 3A). As expected, based on the cellular localization, protease activity using azocasein as substrate was only detected in the CF_S fractions of prtA/HisLas1345- and prtA/Las1345, increasing the proteolytic activity as compared to the control (pBBR2; Figure 3B). The absence of protease activity in the SF of prtA/Las1345 and prtA/HisLas1345 (Figure 3C) reinforces the results obtained for the serralysin-like protein of CLso in S. liquefaciens (Ravindran et al., 2018). Proteolytic activity of HisLas1345 was also confirmed using the purified protein (pHisLas1345) from E. coli (Supplementary Figure 1).

Taken together, these results indicate that Las1345 is not secreted by bacteria phylogenetically related to CLas, such as R. leguminosarum, nor by another surrogate bacterium such as S. marcescens. However, this protein shows proteolytic activity in the soluble cell fraction when expressed in these surrogates. We hypothesize that the long C-terminal region and the absence of multiple calcium-binding domains in the RTX motif of Las1345 allows this protein to be in its active conformation within the cell but hinders secretion to the extracellular space.

Las1345 secretion to the extracellular medium could not be detected in surrogate bacterial models. However, this does not rule out Las1345 secretion in CLas via other non-conserved secretion pathways, or the putative role of Las1345 inside the plant cell. Phenotype changes have been observed via ectopic expression of multiple Sec secreted-dependent effectors in planta, even though the secretion of these proteins have not been demonstrated (Pitino et al., 2016; Clark et al., 2020; Pang et al., 2020; Du et al., 2021). Additionally, several extracellular metalloproteases from phytopathogenic bacteria, including AprA and Prt2/3, take part in modulating the plant immune defense, particularly through processes associated with plant cell wall degradation (Dow et al., 1990, 1993; Pel et al., 2014). Thus, observing whether Las1345 affects or not the plant cell in the hypothetical case that it could be secreted, could help to gain insight into the real scenario where Las1345 is functional. To explore the effect that Las1345 expression could eventually have in plant tissues, N. benthamiana leaves were infiltrated with an Agrobacterium culture harboring 35S-directed untagged and C-terminal tagged versions of Las1345. No phenotypic change between N. benthamiana plants expressing Las1345 constructs (35S::Las1345GFP or 35S::Las1345) and the control (35S,GFP) were observed. Moreover, plant cell integrity and conductivity were not affected by Las1345 expression in planta (Figure 4A), suggesting that cellular membrane damage is not taking place in N. benthamiana leaves when Las1345 is expressed. Fluorescence microscopy analyzes showed that Las1345GFP co-localizes with the plasma membrane dye FM 4–64, indicating that Las1345 is localized in these subcellular structures of the plant cell, particularly correlated with discrete cytosolic structures as a result of plasma membrane internalization (Figure 4B). Interestingly, no increase in protease activity was detected in Las1345-expressing tissues (Figure 4C). These results suggest that the intracellular environment in the leaf tissue is not favorable for the intrinsic proteolytic activity of Las1345.

Although growth was similar between Las1345-expressing R. leguminosarum and S. marcescens and the same bacterial strains transformed with empty pBRR2 plasmids, morphological characteristics of colonies on the surface of agar-solidified media were different (Supplementary Figure 2). Differences in colony phenotypes were also observed when Las1345 was expressed into the phytopathogenic bacteria Xcc, whereas again both control and Las1354-expressing bacteria exhibit similar growth kinetics. Las1345-expressing Xcc developed dull and dry colonies with slightly thin edges, characteristic of defective cell motility, as compared with the smooth and thick-edges colonies found in the control (Figure 5A). This phenotype resembles that of Xanthomonas mutants deficient in the production of xanthan, the main EPS produced by xanthomonads, which affect the extracellular matrix and biofilm development (Yun et al., 2006; Rigano et al., 2007; Torres et al., 2007; Malamud et al., 2011). The Xcc/Las1345 colony phenotype, together with the fact that the proteolytic activity was also detected in the CF of Xcc/Las1345 (Supplementary Figure 3), lead to the hypothesis that the proteolytic activity may influence the assembly or the structure of the extracellular matrix. Accordingly, significant differences in EPS production were observed between Xcc/pBRR2 and Xcc/Las1345 (8.66 ± 3.04 g/g wet cell and 3.26 ± 1.81 g/g wet cell, p < 0.05) when grown in liquid medium (Figure 5B). EPS xanthan contribute to cell motility, cell-to-cell aggregation, bacterial fitness, and virulence in many different Xanthomonas species (Yun et al., 2006; Rigano et al., 2007; Torres et al., 2007; Malamud et al., 2011, 2013; Yaryura et al., 2015; An et al., 2020). This prompted us to study the effect(s) of Las1345 expression in the extracellular matrix of Xcc.

Bacterial motility is mainly controlled by the flagella which is responsible for swimming and swarming (Nakamura and Minamino, 2019). Another type of cell motility, sliding, is a flagella-independent type motility in which xanthan acts as a surfactant or surface-wetting agent to facilitate the movement of bacterial cells (Murray and Kazmierczak, 2008; Malamud et al., 2011). Therefore, we hypothesized that the differential macrocolony phenotype found between Xcc/Las1345 and Xcc/pBRR2 cells may be associated with differences in motility either associated with the flagella or with the formation of the extracellular matrix. Motility of Xcc/Las1345 was assayed in 0.25/0.5% (w/v) agar NYGB over a 2/3-day period. Sliding and swarming motility of Xcc was reduced in Las1345-expressing cells at 48 h (pBBR2 2.14 ± 0.53 cm vs. Las1345 0.61 ± 0.21 cm, p < 0.0001, Student’s t-test), in agreement with the reduced amount of EPS xanthan production compared with the Xcc/pBRR2 control (Figure 5C). A similar phenotype was observed in a gumB mutant of Xanthomonas citri subsp. citri (X. citri), as was reported by Malamud et al. (2011). Interestingly, Xcc/Las1345 also displayed very little swimming motility at 72 h (pBBR2 6.65 ± 0.0.84 cm vs. Las1345 4.31 ± 0.46 cm, p < 0.0001, Student’s t-test). However, Xcc gumB mutant showed a swimming motility similar to the wild-type (Xcc/pBRR2; Figure 5C).

In accordance with these results, the relative expression of flagellum assembly-related genes, such as fliE and fliF (Kan et al., 2018) was significant diminished in Xcc/Las1345 compared with control (Xcc/pBRR2). In contrast, the expression levels of flgL -associated with the hook structure- were similar between Las1345-expressing cells and the control (Figure 5D).

Taken together, all these results suggest that Las1345 expression influences the characteristics of the bacterial extracellular matrix by altering xanthan production, flagellum assembly and biofilm development.

We have previously shown that GFP-labeled Xcc grown in chambered cover slides, develop after 2 days microcolonies showing intimate lateral interactions, and after 4 days a typical biofilm structure (Torres et al., 2007). Under identical culture conditions, we analyzed the in vitro biofilm tridimensional structure of a GFP-labeled strain of Xcc harboring either the pBBR2::Las1345 or the pBBR2 empty vector. After 1 day of culture, Xcc-GFP expressing Las1345 cells were flattened over the bottom and reached the highest z-stacks positions (Las1345; 11828.21 nm vs. pBBR2; 5522.50 nm) without any observable attachment as opposed to control cells (Xcc-GFP/pBBR2), which were grouped at the bottom (lower z-stack) and only contacted the glass surface via one cell pole (Figure 6). By day two, Xcc-GFP/Las1345 cells shaped a homogenous layer, where few cells are interconnected, forming small cellular aggregates separated by extensive water spaces. In contrast, Xcc-GFP/pBRR2 cells developed more complex structures with heterogeneous sizes and shapes, in which bacteria were interconnected side-by-side (Figure 6). Biofilm structure parameters, such as biomass (volume per unit area) and roughness coefficient (biofilm heterogeneity) were analyzed by COMSTAT2. Higher biomass levels were observed in Xcc-GFP/Las1345 expressing Las1345, indicating an increased cell number per slide when compared with Xcc-GFP/pBBR2. The changing in the spatial biomass distribution can be associated with alterations in the hydrodynamic of Las1345-induced biofilm matrix that may result in enhanced availability of nutrient to the cells. In agree with this, the roughness coefficient was increased in the control (Xcc-GFP/pBBR2), showing an intricated biofilm matrix forming channel-like structures that may promote a hydrodynamic biofilm (Figure 6). Similar biofilm structures were also observed in the gumB mutant, as reported previously (Torres et al., 2007). Moreover, by day four of growth, structural changes on Las1345-expressing Xcc biofilms were associated with differences on the extracellular matrix disposition, surrounding cell-to-cell contacts, as shown by congo red (red) fluorescence pattern (Figure 6). Red-fluorescence signals were barely detected over the highest green network developed by Xcc-GFP/Las1345, suggesting that other extracellular matrix components, such as amyloid fibers or adhesins, were modified by Las1345 expression. In contrast, cultures of Xcc-GFP/pBBR2 cells showed extensive red areas at the cell boundaries (Figure 6). These results suggest that Las1345 expression in Xcc increases cell aggregates by changing extracellular matrix components to remodel biofilm structures.

We hypothesize that Las1345 could degrade intracellular precursors of the extracellular matrix, or alter the secretion of proteins or EPS to modulate biofilm development, that eventually would translate into a switch to a planktonic status, that subsequently would favor dispersal of bacteria from the biofilm structure.

The presence of plant-associated biofilms is correlated with pathogenicity in Xanthomonas. As previously reported (Yun et al., 2006; Rigano et al., 2007; Torres et al., 2007; Conforte et al., 2019), changes in both structure and quantity of EPS reduce virulence in host plants. To further understand whether the changes in Xcc biofilm structure as a consequence of Las1345 expression interfere with disease progression, leaves of A. thaliana and N. benthamiana were inoculated with bacterial suspensions of Xcc/Las1345 and Xcc/pBBR2. Disease symptoms (water-soaking and necrosis) were observed in leaves inoculated with Xcc/pBBR2 at 3 dpi. In contrast, the disease area was reduced in leaves infected with Xcc/Las1345 (Figure 7A). Quantification of the bacterial population in planta revealed no significant differences between the Xcc/pBBR2 and Xcc/Las1345 strain growth until 36 or 48 h post inoculation (hpi), in A. thaliana or N. benthamiana, respectively (Figure 7B). However, Xcc/pBBR2 population gradually increased more than 10 times over the following monitoring days, while no Xcc/Las1345 could be recovered at soon as 3 dpi. Suppression of the Xcc-associated disease development was shown in N. benthamiana leaves infected with Xcc gumB mutant (Yun et al., 2006).

As compared with the areas inoculated with Xcc/pBBR2, Xcc/Las1345-infected areas had less plant cellular membrane damage, measured by electrolyte leakage, whereas photosystem II quantum efficiency (ϕPSII) was greater (Figures 7C,D), indicating that Las1345 expression compromised Xcc-associated disease development.