Purification and characterization of FtsZ from the citrus canker pathogen Xanthomonas citri subsp. citri

Abstract Xanthomonas citri subsp. citri (Xac) is the causative agent of citrus canker, a plant disease that significantly impacts citriculture. In earlier work, we showed that alkylated derivatives of gallic acid have antibacterial action against Xac and target both the cell division protein FtsZ and membrane integrity in Bacillus subtilis. Here, we have purified native XacFtsZ and characterized its GTP hydrolysis and polymerization properties. In a surprising manner, inhibition of XacFtsZ activity by alkyl gallates is not as strong as observed earlier with B. subtilis FtsZ. As the alkyl gallates efficiently permeabilize Xac membranes, we propose that this is the primary mode of antibacterial action of these compounds.

Xac is the causal agent of Asiatic citrus canker, a severe plant disease that affects citrus crops, decreases fruit production, and causes economic losses (Gottwald, Graham, & Schubert, 2002). Asiatic citrus canker affects all the commercially important citrus species and cultivars in use (Gottwald et al., 2002). Infected trees exhibit brownish crater-like lesions on aerial tissues that are sometimes surrounded by chlorotic halos. The disease leads to defoliation and premature fruit drop, which with time, decreases citrus fruit production.
The disease is currently present in South and North America, Asia, Africa, and Oceania (Behlau, Fonseca, & Belasque, 2016;Davis et al., 2015;Leduc et al., 2015;Stover et al., 2014). The state of São Paulo, Brazil, the largest producer of concentrated orange juice in the world, has now been declared an area of Risk Mitigation System according to current regulation (MAPA, Brazil, Normativa 37, September 2016), meaning the disease is now considered endemic in this area.
Infection control is subject to pressure from the orange production chain and currently involves plantation of less susceptible citrus cultivars, the use of windbreaks to avoid bacteria spreading by wind and rain as well as to prevent wind-induced damage to trees which also facilitates infection, and the eradication of symptomatic trees along with spraying copper-containing bactericides in a radius of 30 m of the symptomatic tree (Behlau, Canteros, Minsavage, Jones, & Graham, 2011;Gottwald et al., 2002). However, this strategy is costly and has limited effectiveness (Behlau, Canteros, Jones, & Graham, 2012;Behlau et al., 2016). Moreover, copper sprays are known to leave footprints on the environment and may help the emergence of resistant Xac strains (Canteros, 1999). Copper is not broken down in the environment and can be accumulated in plants and animals that live on copper-contaminated soils. Therefore, there is an urgent need for environmental friendly compounds that can be used to prevent the imminent spread of Xac (and associated economic damage to citriculture), as well as to prevent the accumulation of toxic compounds in the soil.
In an important manner, these compounds are not mutagenic or cytotoxic and therefore safe to use (Silva, Polaquini, Regasini, Ferreira, & Pavan, 2017). Alkyl gallates are esters of gallic acid, the main product of tannin hydrolysis. The hydrolysis of alkyl gallates produces gallic acid and the corresponding alcohols (or alkanols), which both are common components in many plants. Alkyl gallates have a head-and-tail structure similar to alkanols, suggesting that their antibacterial mode of action may be as surface-active agents affecting membrane integrity (Takai, Hirano, & Shiraki, 2011).
Using membrane permeability to propidium iodide as an indicator for disruption of membrane integrity, we showed that alkyl gallates indeed permeabilize the membrane and that the efficacy of this activity depends on the length of the alkyl chain (Król et al., 2015). However, elongation of Xac and B. subtilis exposed to alkyl gallates suggested that these compounds not only target membranes but also cell division, and we could show that this is indeed the case as these compounds inhibit ring formation of B. subtilis FtsZ in vivo and FtsZ polymerization and GTPase activity in vitro (Król et al., 2015;Silva et al., 2013).
To formally show that alkyl gallates target FtsZ from Xac (XacFtsZ) as well, we purified and characterized the protein and tested it in the presence of alkyl gallates. His-tagged XacFtsZ was previously investigated in a search for single-stranded DNA-based compounds as anti-canker agents (Ha, Lee, Hyun, & Yoon, 2015).
To the best of our knowledge, this is the first study of fully native XacFtsZ and the GTPase activity of the protein.

| Cloning, protein expression, and purification
Our initial attempts at XacFtsZ expression failed (Król et al., 2015), but a close inspection of the Xac genome annotation revealed that the ftsZ start codon was incorrectly assigned and that the ftsZ gene included an additional 27 bases at the 5′ end. Thus, the Xac ftsZ gene was amplified from an expression vector (pCXZ-FtsZ, a gift from H. Ferreira) lacking the upstream sequence and cloned into the pET15b vector (Novagen). The following primers were used: forward-GAGCCCATGGCACATTTCGAACTGATTGAAAAAATG GCTCCCAACGCGGTCATCAAGG, where the NcoI site is underlined and the additional 27 bases of Xac ftsZ are in bold; reverse-AGTTCATATGCGACGCAGCCGACGCTCCTCAG, where the NdeI site is underlined and the stop codon is in bold. Escherichia coli DH5α was used as a host for DNA manipulation. The resulting plasmid, pET15b-XacFtsZ, was sequenced to ensure that the ftsZ coding sequence was correct.
XacFtsZ was expressed in E. coli BL21 (DE3) cells containing the pET15b-XacFtsZ vector. The cells were cultivated in LB medium containing 100 μg/ml ampicillin at 37°C until OD 600 was 0.6-0.8 and then in LB medium containing 100 μg/ml ampicillin and 1 mM IPTG at 30°C for 4 hr. After that, the cells were harvested and stored frozen until the purification, for which they were resuspended in a TRIS50 buffer (see below). The cells were disrupted by sonication for 8 min at amplitude 7, in 8 s on/off intervals (MSE Soniprep 150) with concomitant cooling on an ice/ethanol mixture. The insoluble fraction was removed by centrifugation at 70,000g for 45 min at 4°C. XacFtsZ was purified by ammonium sulfate precipitation from the soluble fraction in the TRIS50 buffer. The protein precipitated at 20% of saturated ammonium sulfate, which was added gradually to the soluble fraction which was stirred gently while being kept on ice.
The pellet of XacFtsZ was centrifuged at 3,200g for 20 min at 4°C, resuspended, and dialyzed against TRIS50 buffer in order to remove residual ammonium sulfate. This resulted in a pure (>95%) FtsZ fraction, as confirmed by SDS-PAGE gel analysis. The protein was stored at −80°C in small aliquots, which, after defrosting, were either used immediately or after a few days of storage at 4°C. After a few days, the thawed protein was discarded, and once thawed, protein was never refrozen.
Sedimentation, light scattering, and GTPase assays were performed according to literature protocols (Ingerman & Nunnari, 2006;Król & Scheffers, 2013;Margalit et al., 2004). FtsZ was polymerized at 30°C in an appropriate buffer with 5 mM MgCl 2 and 1 mM nucleotide (or a corresponding volume of buffer) for 5 min in a water bath and for approximately 7 min until samples were spun down at 186,000g. GTPase activity was measured with a continuous, regenerative coupled assay that detects the decrease of NADH absorption used for the regeneration of GDP to GTP (Ingerman & Nunnari, 2006;Margalit et al., 2004). The reaction was performed at 30°C at different FtsZ concentrations in TRIS50 buffer with 5 mM MgCl 2 and 400 μM GTP, in the presence or absence of an alkyl gallate, as indicated in the results section. The assay components for GTP regeneration were 20 U/ml pyruvate kinase/lactic dehydrogenase enzymes from rabbit muscle (Sigma-Aldrich), 1 mM NADH (Sigma-Aldrich), and 2 mM phospho(enol) pyruvic acid monopotassium salt (Sigma-Aldrich). The reaction was followed by the measurement of 340 nm absorbance on a BioTek Synergy ™ Mx Microplate Reader. For the analysis of the nucleotide bound to FtsZ, the protein was precipitated with 10% TCA and the resulting supernatant was loaded on a MonoQ HR 5/5 1 ml column (GE Healthcare) in 10 mM KH 2 PO 4 /K 2 PO 4 pH 8.0 buffer and run in a gradient formed by 50 mM mM KH 2 PO 4 /K 2 PO 4 pH 7.4, 1M NaOH. The concentrations of the corresponding nucleotide were calculated based on reference GTP, GDP, and GMP solutions.

| Calibration of protein concentration
Quantitative amino acid analysis of FtsZ was performed by Eurosequence (the Netherlands). Commercial BSA (Thermo Fisher) and calibrated FtsZ were used to calibrate Bradford (Thermo Fisher), bicinchoninic acid (Thermo Fisher), and DC (Bio-Rad) colorimetric assays.

| Electron microscopy
XacFtsZ in the polymerization buffer was incubated at 30°C, and samples were collected at different time points and applied to an electron microscopy grid as described in (Król & Scheffers, 2013).
The grids were examined in a Philips CM120 electron microscope equipped with a LaB 6 filament operating at 120 kV. Images were recorded with Gatan 4000 SP 4K slow-scan CCD camera.

| Native Xac FtsZ production and purification
In order to produce and purify FtsZ from Xanthomonas in a soluble form, the Xac ftsZ gene was cloned into the pET15b vector, which should allow production of the native protein. XacFtsZ proved to be expressed as soluble protein in E. coli BL21 cells and ammonium sulfate precipitation was sufficient to obtain the protein that did not need any further purification steps, as established by SDS-PAGE analysis (not shown). This provides a very fast and cheap purification of XacFtsZ that can be produced in high yield (about 100 mg of pure protein from half a liter of LB medium). To our knowledge, this is the first time when XacFtsZ was obtained without any purification tag.
FtsZ proteins have a quite unique amino acid composition, with a very low number of aromatic amino acids-they contain no tryptophans and only a few tyrosines (XacFtsZ only one, Figure 1). Therefore, the absorbance at 280 nm is not an accurate method to determine the concentration of the protein. Various colorimetric methods produce a certain error that is related to the colorimetric reaction detecting, for example, aromatic amino acids or cysteines (of which XacFtsZ only contains two). In a traditional manner, the concentration of FtsZ from different species was assessed by calibrating a number of colorimetric assays against the standard protein, which is normally BSA Lu, Stricker, & Erickson, 1998). Here, we performed a quantitative amino acid analysis of FtsZ that was further used to calibrate three colorimetric assays against commercial BSA. This gave the following correction factors: 0.60 for the Bradford assay, 0.69 for the BCA assay, and 0.76 for the DC assay. We decided to use the Bradford assay as our standard method to determine FtsZ concentration, due to its reproducibility and simplicity.

| XacFtsZ polymerizes independently of external GTP
In vivo, FtsZ exists in an equilibrium between monomers and polymers. This equilibrium is dynamic and controlled by many protein and nonprotein factors influencing cell division, but also the gen-  & Erickson, 2002;Sun & Margolin, 1998). In vitro however, FtsZ is able to polymerize in certain simplified conditions that usually require GTP that is bound between monomers of the polymer in the presence of magnesium ions. The amount of polymers formed is traditionally quantified in a sedimentation assay (Bramhill & Thompson, 1994). In a surprising manner, XacFtsZ showed a significant pellet in the absence of any of the two nucleotides, GTP and GDP, as long as magnesium was present (Figure 3). This was not the case if this ion was absent, even if GTP or GDP was added. The most significant pellets were formed in the presence of magnesium and any of the two nucleotides, but there was no striking difference in the amount of pellet formed in the presence of GTP or GDP.
This atypical behavior prompted us to hypothesize that XacFtsZ copurifies with a GTP nucleotide bound to the protein, which would allow polymerization to be triggered by the addition of magnesium ions, as shown before for certain mutants of E. coli FtsZ (Scheffers, de Wit, den Blaauwen, & Driessen, 2001). Therefore, nucleotides that copurified with XacFtsZ were extracted from the protein with 10% TCA and analyzed using FPLC with the use of GTP, GDP, and GMP as standard references. Purified XacFtsZ had 0.9 mol of bound GDP (but no GTP or GMP) per 1 mol of protein, which makes it similar to FtsZ from E. coli (RayChaudhuri & Park, 1992;Scheffers et al., 2001;Zorrilla, Minton, Vicente, & Andreu, 2000). Furthermore, the amount of added magnesium influenced the amount of XacFtsZ pelleted in the sedimentation assay ( Figure 4). In the absence of any external nucleotide, the pellet started to be visible between 0.5 and 1.0 mM of MgCl 2 and continued to increase above 20 mM MgCl 2 . To see whether this effect was specific for Mg 2+ or also occurred with other divalent cations, we used Mn 2+ which, as we previously showed, stabilizes BsFtsZ polymers (Król, de Sousa Borges, Kopacz, & Scheffers, 2017). The effect was similar when XacFtsZ was allowed to polymerize in the presence of MnCl 2 , but here a lower concentration was needed for the pellet to be visible and the protein was almost fully sedimented already at 10 mM MnCl 2 .
In addition, the sedimentation properties of XacFtsZ were analyzed in different buffers (see Materials and Methods and Figure 5). In all buffers of pH lower than 7.0, significant amounts of pellet were observed in the buffer only. Considerable pellets were also formed in the buffers of higher pH when salt concentrations were high. It was concluded that the protein is not stable in these conditions. On the other hand, the behavior of XacFtsZ was fairly similar in HEPES and TRIS buffers at 50 mM KCl, independently of the pH.
F I G U R E 3 Sedimentation assay of XacFtsZ. FtsZ was polymerized for approx. 12 min (5 min in water bath + ~7 min until samples started to spin down at 186,000g). 5.2 μM FtsZ was polymerized at 30°C in TRIS50 buffer with 5 mM MgCl 2 and 1 mM nucleotide (or a corresponding volume of buffer), where indicated. W-whole sample input, N-supernatant fraction and P-pellet fraction F I G U R E 4 Influence of magnesium (a) and manganese (b) on FtsZ sedimentation. The procedure was performed as in Figure 3. N-supernatant fraction and P-pellet fraction, input samples were not loaded on SDS-PAGE

| XacFtsZ forms very short polymers
The morphology and dynamics of the polymers formed by XacFtsZ were observed with transmission electron microscopy ( Figure 6) and static light scattering. The polymers were rather short, but straight when compared to FtsZ from other species (Buske & Levin, 2012;Chen et al., 2012;Król & Scheffers, 2013;Pacheco-Gómez, Roper, Dafforn, & Rodger, 2011;Scheffers, 2008). Immediately upon addition of MgCl 2 and GTP (Figure 6a), only a small amount of very short polymers was observed that did not increase sig- FtsZ polymerization properties were tested with a static light scattering method in many different conditions, where buffer, GTP, magnesium, and protein concentrations varied. However, we were never able to observe the characteristic FtsZ polymerization and depolymerization peak, even when divalent cations (Ca 2+ , Mn 2+ ) or the poly-cation DEAE-dextran were added to promote bundling. This could be due to the very short length of polymers formed by XacFtsZ that were observed in the electron microscopy experiments.

| Alkyl gallates are mild inhibitors of XacFtsZ
Alkyl gallates were recently shown to be able to act as broadspectrum antimicrobial agents that can kill Xac (Silva et al., 2013).
This activity can be attributed to the surfactant properties of alkyl gallates that disrupt the bacterial membrane; however, several alkyl gallates are also able to interact directly with FtsZ from B. subtillis (BsFtsZ) and cluster the protein resulting in inhibition of GTPase activity and bundling of FtsZ polymers (Król et al., 2015). Therefore, we wanted to examine whether alkyl gallates act similarly on XacFtsZ.
In a previous study, pentyl, hexyl, heptyl, and octyl gallates were selected for testing with BsFtsZ, based on their ability to elongate Xac cells and disrupt their cell division machinery (Król et al., 2015;Silva et al., 2013). In the presence of 50 μg/ml of either alkyl gallate, only residual GTPase activity of BsFtsZ was detected, resulting in a nearly sixfold decrease compared to the control performed in the absence of the alkyl gallate. This was not the case with XacFtsZ, where a reduction in activity was found, but less significant and differing between the alkyl gallates used. The presence of hexyl and heptyl gallate caused approximately 30% and 60% decrease in XacFtsZ activity, while pentyl and octyl gallate showed only approximately 15% of reduction (Figure 7).
Because heptyl gallate showed the strongest inhibition of XacFtsZ GTPase activity, the activity was also measured at variable induce XacFtsZ clustering or aggregation because the pellet fraction was also higher in the absence of magnesium. However, this interaction seems to be XacFtsZ specific, because the compounds did not influence the pellet formation in the case of BSA.

| Alkyl gallates permeabilize Xac membranes
As the inhibitory effect of the alkyl gallates on XacFtsZ was relatively mild, we analyzed whether, as in the Gram-positive B. subtilis, alkyl gallates can disrupt Xac membranes. A permeabilization assay clearly showed that at MIC 50 pentyl gallate permeabilized 42% of Xac cells, and the other alkyl gallates permeabilized 94% or more Xac cells (Table 1). This shows that the antibacterial activity of the alkyl gallates against Xac is primarily caused by the effect of the alkyl gallates on the membranes. We recently reported a similar membrane permeabilization for mono-acetylated versions of these alkyl gallates (Savietto et al., 2018). F I G U R E 8 The size of the sedimented pellet of XacFtsZ in the presence of alkyl gallates. 10 μM XacFtsZ or BSA (control) was sedimented as in Figure 3 in the presence of 100 μg/ml of alkyl gallate or 1% DMSO (vehicle solvent). The percentage of pelleted protein (BSA or FtsZ) was calculated with ImageJ from SDS-PAGE analysis. The data are a mean of three repeats, and the bars represent standard deviation

Permeabilized cells (mean ± SD)
Control ( Results from two independent replicates, n represents the number of cells counted in each experiment. polymerization by static light scattering, which we could not reproduce with native XacFtsZ. Using GTPase activity as a robustly reproducible and quantitative indicator of FtsZ activity, we were able to show that XacFtsZ is inhibited by alkyl gallates although not to a similar extent as B. subtilis FtsZ. Although treatment of Xac with alkyl gallates caused cell elongation (Silva et al., 2013), filamentation of cells to more than double the length of the cell, as expected for a true cell division inhibitor, was not found for Xac, and only at reduced concentrations of alkyl gallates in B. subtilis (Król et al., 2015).
Combined with our observation that Xac membranes are permeabilized by the alkyl gallates, we conclude that the killing of Xac by alkyl gallates, especially at concentrations around MIC 50 or higher, is primarily mediated through membrane permeabilization, not by its inhibition of XacFtsZ.

ACK N OWLED G M ENTS
We would like to thank Henrique Ferreira (Instituto de Biociências,

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA ACCE SS I B I LIT Y
All data are included in the main manuscript. Raw data and materials are available on request.