The protein-conserved residues were determined from alignments using PSI-BLAST¹. The multiple sequence alignment and figure was generated with MEGA6 (Tamura et al., 2013) and BOXSHADE version 3.21². The LotP dimer structure was modeled using GalaxyWEB sever (Ko et al., 2012; Shin et al., 2014).

The strains used in this work are listed in Table 1. E. coli strains were cultured at 37°C in Lysogeny broth (LB) medium or M9 media supplemented with 0.1 mM of CaCl₂, 2 mM of MgCl₂, and 0.2% v/v glycerol. The culture media were amended with the appropriate concentration of antibiotics as recommended for each genetic vector used (Table 1).

Standard methods were used for restriction enzyme digestion and molecular recombination. E. coli cell transformation was done as previously described (Sambrook et al., 1989). Plasmids were isolated using the QIAprep^® Spin Miniprep Kit (Qiagen, Valencia, CA, United States), and PCR products were purified using QIAquick^® purification kits (Qiagen). Proteins were expressed with a His_6X-tag in the strain E. coli BL21 (DE3) and purified as previously described (Loto et al., 2017).

CLIBASIA_03135 was previously cloned in the plasmid p15TV-L (GenBank accession EF456736). Site-directed mutagenesis was performed into this clone/construction; the sequence of the primers used for each mutation is listed in Supplementary Table 1. The methods used for mutagenesis were described by Edelheit et al. (2009).

N- and C-terminal modified LotP constructs were cloned into the plasmid pCDF-1b (Novagen, Merck KGaA, Darmstadt, Germany) and p15TV-L. The sequences were amplified by PCR using the primers listed in Supplementary Table 1 and cloned in the vector using the restriction sites introduced within each primer. The coding sequence for the His.EK cleavage site was added at the N-terminal of LotP by cloning it into the plasmid pCDF-1b. FLAG- or His-tag was introduced to the N-terminal or C-terminal by modifying the primers used in PCR amplification. N-terminal(Lon)-LotP.His and N-terminal(Lon)-ΔNLotP.His were amplified using a two-step PCR with Fw_N1LotP or Fw_N2LotP as forward primer in the first amplification and Fw_N3LotP_NcoI in the second amplification. The reverse primer, Rv_LotP_HisStop_XhoI, was the same in all three PCRs.

Size-exclusion chromatography was performed as described previously (Loto et al., 2017). Superose 12 10/300 chromatography column (GE Healthcare, Chicago, IL, United States) was calibrated with cytochrome c from horse heart (12.4 kDa), carbonic anhydrase from bovine erythrocytes (29 kDa), albumin from bovine serum (66 kDa), alcohol dehydrogenase from yeast (150 kDa), and β-amylase from sweet potato (200 kDa). In the case of Superose 6 10/300 chromatography column (GE), apoferritin from equine spleen (443 kDa), and thyroglobulin from bovine thyroid (669 kDa) were added to the standard mix. Each construction was analyzed with samples from at least two different purifications process, and each sample was run at least three times to confirm the profile and elution volumes.

The cross-linking assay was done as previously described with minor modifications (Lee et al., 2004). Briefly, purified LotP protein (2 μg) was incubated with 2% formaldehyde at 25°C, and the reaction was stopped after 10 or 60 min with the addition of glycine to a final concentration of 0.125 M. Cross-linked products were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) (10%).

Purified LotP protein (∼0.2 mg/ml) was dialyzed with a solution of 5 mM of MgSO₄, 36.4 mM of Na₂HPO₄, and 3.6 mM of NaH₂PO₄. UV circular dichroism (CD) spectra (190–260 nm) was recorded at 25°C on a Circular Dichroism Spectrometer Model 400 (Biomedical, Inc., Lakewood, NJ, United States). We used a 1-mm path length cell at 1-nm step resolution and 10 nm/min scan speed. All spectra were corrected by subtracting the control measurement of dialysis buffer and normalized by the mean residue weight (MRW) of the proteins, path length (P, cm), and protein concentration (C, mg ml^–1). The final spectra were expressed in Δε (M^–1 cm^–1), where Δε = θ ^∗ (0.1 ^∗ MRW)/(P ^∗ C ^∗ 3,298). Machine units (θ) were obtained in millidegrees from the spectrometer.

Limited proteolysis of purified LotP proteins was done as described by Cellini et al. (2010), with minor modifications. Briefly, purified proteins were incubated with proteinase K (ratio of 300:1) at 25°C. Aliquots were taken at 0, 5, 15, 45, 90, and 240 min. The reaction was stopped with phenylmethylsulfonyl fluoride (PMSF), and the samples analyzed by SDS-PAGE (12.5%).

The protein–protein interaction between LotP, LotP mutants, and the Lon protease (CLIBASIA_00775) was assessed by cloning the genes and encoding each protein into a bacteria two-hybrid system. Briefly, the sequence of Lon protease and the modified versions of LotP from ‘Ca. L. asiaticus’ were amplified by PCR using the primers listed in Supplementary Table 1. Each sequence was fused to the ω subunit of the RNAP by cloning the fragment into the NdeI and NotI sites of the pBRGP-ω plasmid. The same genes were cloned and fused to the zinc finger DNA-binding protein of the murine Zif268 transcription factor using same restriction sites for the plasmid pACTR-AP-Zif. The recombinant clones were expressed into the reporter E. coli strain KDZif1DZ. β-Galactosidase activity was measured as previously described (Loto et al., 2017). Each assay was performed in biological and technical triplicates—induced with 20 μM of IPTG.

Two E. coli strains were used in the reported growth curves: BW25113 as the wild type (wt) and BW25113 with a knockout of the lon gene (ΔLon, strain JW0429). The strains were transformed with pBAD24_LotP and pCA24N_SulA (from strain JW0941, ASKA collection) as well as pBAD24_LotP^R104A, which was amplified and cloned using primers listed in Supplementary Table 1. The empty pBAD24 vector was used as a control. Growth curves were carried out using M9 media with the addition of 0.1 mM of CaCl₂, 2 mM of MgCl₂, and 0.2% v/v glycerol as the sole carbon source. Cells were incubated at 37°C and 200 rpm. Individual flasks were inoculated with aliquots obtained from overnight cultures grown in LB broth at starting OD_{600 nm} = 0.05. The induction of gene expression was done at the beginning of each assay with L-arabinose (0.001–0.2%) for assays using the pBAD24 plasmid. IPTG (100 μM) was used for pCA24N plasmid. Cellular growth was monitored at OD₆₀₀; all assays were done in triplicate.

Samples analyzed with SDS-PAGE (12.5%) were transferred onto a 0.45 μm of polyvinylidene difluoride (PVDF) membrane (Bio-Rad, Hercules, CA, United States) using a semi-dry blotting unit (Fisher Scientific, Hampton, NH, United States) at 450 mA for 40 min. The membranes were treated as described before (Loto et al., 2017). Primary antibodies used were rabbit anti-LotP, rabbit anti-FLAG (1:5,000, Sigma-Aldrich Corp., St. Louis, MO, United States) and mouse anti-His (1:3,000, Sigma-Aldrich). The secondary antibodies conjugated to horseradish peroxidase for chemiluminescence detection were anti-rabbit (1:20,000, Sigma-Aldrich) and anti-mouse (1:10,000, Sigma-Aldrich). Horseradish peroxidase (HRP) activity was detected using Amersham ECL^TM Western Blotting Detection Reagents (GE Healthcare, Pittsburgh, PA, United States). The bands in the membrane were visualized using the automatic imager FluorChem R (ProteinSimple, San Jose, CA, United States), and the band intensity was quantified by ImageJ software (Schneider et al., 2012).

Escherichia coli strain BL21 (DE3) was transformed with p15TV-L_His.LotP, p15TV-L_His.EK.LotP, pCA24N_Lon (pCA24N plasmid carrying E. coli Lon protease, strain JW0429 ASKA collection), and pCDF-1b_FLAG.MBP.C-terminalSulA; empty plasmids were used as controls. p15TV-L_His.EK.LotP was cloned from the digestion of pCDF-1b_His.EK.LotP using NcoI and XhoI restriction enzymes and a ligation in p15TV-L digested with the same restriction enzymes. pCDF-1b_FLAG.MBP.C-terminalSulA was cloned using a PCR product obtained from a two-step PCR amplification digested with NcoI and SacI. The SulA fusion was expressed from the pCDF1-b in E. coli BL21(DE3), a protease negative strain. The E. coli Lon protease was expressed from the vector pCA24N, while His.LotP and His.EK.LotP were expressed from p15TV-L. Once the different combinations of the proteins were induced, the amount of remaining proteins for each case/combination was analyzed by Western blotting.

The in vivo degradation assay was carried out as described previously (Ishii et al., 2000). Briefly, flasks containing LB media were inoculated with aliquots from overnight cultures at OD_{600 nm} = 0.025. Genes were induced with 0.05 mM of IPTG (3 h of culture); and protein expression was stopped with tetracycline 30 min after induction. The samples obtained were treated with PMSF to prevent further proteolysis. MBP.SulA degradation was monitored by Western blotting using anti-FLAG antibodies.

Citrus sinensis cv. Valencia on Kuharski rootstock plants were obtained from Phillip Rucks Citrus Nursery (Frostproof, FL, United States). Plants were grown in a greenhouse with a relative humidity of 74% and a controlled temperature around 27°C. Plants (6–8 months old) were moved to an indoor growth chamber 2 weeks before infiltrations. The chamber was kept at 24°C with 14 h of light and 10 h of dark per day. The leaves were cleaned with 10% bleach followed by water prior to infiltrations. Purified protein, diluted to the working concentration with 10 mM of MgCl₂, was used to infiltrate citrus leaves. Identical buffer, with no protein, was infiltrated as a control in each leaf.

For proteomic analysis, six plants were used, two for each treatment group. Fifteen leaves per treatment group (His.LotP and His.EK.LotP.FLAG) were fully infiltrated in all lobes at a concentration of 8 μM. Biological replicates (one leaf per tree) were taken at 1, 2, and 3 dpi (days post infiltration) for a total of 18 leaves. Once collected, leaves were frozen immediately in liquid nitrogen and subsequently placed at –80°C until further use. Phenol extracted proteins were precipitated and were further washed twice with cold 80% acetone (stored at –20°C), incubating overnight each time. The final pellet was re-suspended in 500 μl of ITRAQ protein buffer (0.4 M of urea, 12.5 mM of TEAB, 1% Triton X-100, and 0.05% SDS). Precipitated proteins were further used for acetone precipitation. Protein concentrations were determined using the Bradford assay and run on a 12.5% SDS-PAGE gel to ensure protein extraction quality.

Proteins were dissolved in protein buffer [8 M of urea, 50 mM of Tris-HCl, pH 8.0, 50 mM of triethylammonium bicarbonate, 0.1% SDS (w/v), 1% Triton-100 (w/v), 1 mM of PMSF, 10 μg/ml of leupeptin, and 1% phosphatase inhibitor cocktail 2 and 3 (v/v)]. For each sample, a total of 200 μg of protein was reduced with 10 mM of tris(2-carboxyethyl)phosphine, alkylated with 20 mM of iodoacetamide, trypsin-digested (w/w for enzyme:sample = 1:100), and labeled according to the manufacturer’s instructions (Thermo Scientific, Rockford, IL, United States). Control samples with three different day points of Days 1, 2, and 3 were labeled with TMT tags 126, 127N, and 127C, respectively, and the LotP samples were labeled with TMT tags 128N, 128C, and 129N, respectively. The corresponding FLAG samples were labeled with 129C, 130N, and 130C, respectively. To increase the signal, mixed samples of LotP and FLAG were labeled with 131N, and mixed samples of controls were labeled with 131C, which works as an internal standard. Therefore, all experiments were performed with at least biological duplicates in time-point per group. Labeled peptides were desalted with C18-solid phase extraction and dissolved in strong cation exchange (SCX) solvent A [25% (v/v) acetonitrile, 10 mM of ammonium formate, and 0.1% (v/v) formic acid, pH 2.8].

The peptides were fractionated using an Agilent HPLC 1260 with a polysulfoethyl A column (2.1 × 100 mm, 5 μm, 300 Å; PolyLC, Columbia, MD, United States). Peptides were eluted with a linear gradient of 0–20% solvent B [25% (v/v) acetonitrile and 500 mM of ammonium formate, pH 6.8] over 50 min followed by ramping up to 100% solvent B in 5 min.

A hybrid quadrupole Orbitrap (Q Exactive Plus) MS system (Thermo Fisher Scientific, Bremen, Germany) was used with high-energy collision dissociation (HCD) in each MS and MS/MS cycle. The MS system was interfaced with an automated Easy-nLC 1000 system (Thermo Fisher Scientific, Bremen, Germany). Each sample fraction was loaded onto an Acclaim Pepmap 100 pre-column (20 mm × 75 μm; 3 μm-C18) and separated on a PepMap RSLC analytical column (250 mm × 75 μm; 2 μm-C18) at a flow rate at 350 nl/min during a linear gradient from solvent A [0.1% formic acid (v/v)] to 30% solvent B [0.1% formic acid (v/v) and 99.9% acetonitrile (v/v)] for 95 min, to 98% solvent B for 15 min, and hold 98% solvent B for additional 30 min. Full MS scans were acquired in the Orbitrap mass analyzer over m/z 400–2,000 range with resolution 70,000 at 200 m/z. The top 10 most intense peaks with charge state ≥ 2 were fragmented in the collision cell with a normalized collision energy of 28. The quadrupole isolation window was 0.7 m/z. The maximum ion injection time for both the survey scan and the MS/MS scan was 250 ms, and the ion target values were set to 3^e6 and 1^e6, respectively. Selected sequenced ions were dynamically excluded for 60 s.

The raw MS/MS data files were processed by a thorough database searching approach considering biological modification and amino acid substitution against Uniprot Citrus database (downloaded on June 6, 2019; 196,268 entries) using the Proteome Discoverer v2.3 (Thermo Fisher Scientific), with the SEQUEST algorithm (Eng et al., 1994). The following parameters were used for all the searching: peptide tolerance at 10 ppm, tandem MS tolerance at ±0.02 Da, peptide charges of 2+ to 5+, trypsin as the enzyme, allowing one missed cleavage, TMT label and carbamidomethyl (C) as fixed modifications, acetylation (n-terminus), 2-hydroxyisobutyrylation (K), loss of methionine (n-terminus), oxidation (M), and phosphorylation (S, T, and Y) as variable modifications. For peptide confidence, we adopted the following cutoff values of Xcorr that are commonly used for the SEQUEST algorithm (Nesvizhskii et al., 2003): 2.31 for 2+, 2.41 for 3+, and 2.6 for 4+ and 5+ peptides. The false discovery rate (FDR) was calculated using the Percolator algorithm in the Proteome Discoverer workflow based on the search results against a decoy database and was set at 1% FDR. Duplicates were grouped, and statistical analyses evaluated as the quantitative ratio between each control and treatment per day. Furthermore, proteins identified which contained at least two peptides with a p-value of <0.05 and log2 fold ratio greater or less than 0.5.

Mercator was used for a genome-wide functional annotation of the UNIPROT Citrus Database (June 6, 2019), creating MapMan Bin codes. Pathway analysis was curated using PageMan that is fully integrated into the MapMan application.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Deutsch et al., 2017) partner repository with the dataset identifier PXD020912 and doi: 10.6019/PXD020912.

Maximum quantum yields of photosystem II (PSII) were estimated from chlorophyll fluorescence measurements using the Opti-Sciences PSK (Opti-Sciences, Hudson, NH, United States). Dark adaption time was determined by checking the value of the maximum PSII quantum yield, F_v/F_m, of citrus leaves at different dark-adapted times. It was determined that 15 min of dark adaptation was appropriate for a stable F_v/F_m of 0.80 with fluorometer settings as follows: saturation pulse intensity of 1, saturation pulse width of 1, modulated light intensity of 2, and gain of 6. Chlorophyll fluorescence was measured at the middle part of the abaxial side of the third lobe from the top of the leaf with the in situ portable fluorometer. The minimum fluorescence in dark-adapted state (F₀) was measured after 15 min dark adaptation. Measurements were taken daily. Based on the measured fluorescence signals, the variable fluorescence in dark-adapted state (F_v = F_m – F₀) can also be obtained.

We had previously reported that stabilized soluble LotP, carrying a histidine tag and a TEV cleavage site in its N-terminal (His.LotP), eluted from a Superose 12 column as a dimer of 44 kDa (∼12.8 ml) after purification. A deeper analysis of our samples showed a minimal fraction of higher-molecular-weight complexes, with a peak out of the resolution of the column (∼7.6 ml) (Figure 1A). In order to fully evaluate the importance of the high-molecular-weight complexes of His.LotP, we have analyzed native and site-directed mutants of newly purified samples. We had observed high-molecular-weight complexes using a Superose 6 column (∼300 kDa, ∼ 15.72 ml) corresponding to structures of 12 monomeric subunits (Figure 1B). The Superose 6 column has better resolution for protein of higher molecular weight. Aimed to identify critical residues holding these structures together, the native molecular weight of a library of site-directed mutants (18 in total) was also analyzed. The mutation sites were selected based on amino acid conservancy among LotP homologues (Supplementary Figure 1). Noteworthy, the single alanine replacement LotP^R104A shifted a large fraction of the multimeric units observed to monomers of 24 kDa (∼13.7 ml) (Figure 1C). The spatial location of this relevant amino acid in the predicted protein structure is shown in Figure 1D. The model analysis suggests that this amino acid is in a critical position connecting the two main domains of the protein.

The results obtained with the single LotP^R104A mutant prompted us to evaluate the contribution of the N- and C-terminal domains of LotP on the molecule oligomerization. Understanding the stability of these large-molecular-weight complexes will help us to better evaluate the biological role of the protein in the pathosystem. The two domains are well defined and separated by a flexible alpha helix of 23–28 amino acids (Figure 1D). To address the relevance of each domain in oligomerization, the recombinant proteins were modified as described in Figure 2A. Adding a FLAG (His.LotP.FLAG) or a histidine tag (LotP.His) to the carboxy terminal showed an identical oligomerization profile to that of His.LotP (Figure 2B and Supplementary Figure 2). Thus, we concluded that modifications to the C-terminal have no effect on the formation of multimers.

To address the importance of the N-terminal domain in the formation of protein multimers, the TEV site was replaced by an enterokinase (EK) cleavage site. The modified proteins, named His.EK.LotP.FLAG and His.EK.LotP, eluted as a unique peak of 44 kDa (12.7 ml) from the Superose 12 column corresponding to dimers in solution (Figure 2B).

Furthermore, we hypothesized that LotP has a dynamic behavior in solution where several oligomeric forms coexisting. To verify our hypothesis, His.LotP was mixed with a two-fold excess of His.EK.LotP.FLAG, and the samples were analyzed by size-exclusion chromatography. The results indicated that His.LotP interacts with His.EK.LotP.FLAG, shifting most of the multimeric complexes to a uniform population of dimers in solution (Figure 3A). The total amount of protein in the mixture remained constant, as no protein precipitation was observed after centrifugation. These observations suggested that the synthetic heterodimer remained in solution.

A time course cross-linking assay using the purified proteins, separately, indicated that both His.LotP and His.EK.LotP.FLAG can form multimers of approximately 25, 50, 75, and 100 kDa or of even higher molecular weights (Figure 3B). Still, after 1 h in solution, the proportion of dimers was significantly higher in the case of His.EK.LotP.FLAG, indicating either a higher stability of the dimers or slower rate of multimers formation for the recombinant protein. In summary, the results suggested that LotP has a dynamic behavior in solution—going from dimers to a steady formation of multimers.

To further understand the significance of multimer formation, we introduced additional modifications to the N-terminal domain of LotP (Figure 2A). Deletion of 13 amino acids in the amino terminal (ΔNLotP.His) and further comparative analysis with size-exclusion chromatography revealed that the deletion did not abolish multimer formation but shifted a large fraction of proteins to monomeric subunits (Figure 4A). When the His.EK site was inserted in the N-terminal of LotP^R104A (His.EK.LotP^R104A), only complexes of ∼112 kDa (hexamers) were observed (Figure 4A). These results suggest that changes in the overall scaffold of the N-terminal sequence may compensate for the effects of the punctual mutation. Lacking the structure of this protein, we are unable to elucidate this intriguing effect using closely related models.

Several LotP constructs were analyzed for structural changes using CD and a limited proteolysis assay with proteinase K. The far-UV CD spectrum representing the ratio of α-helices and β-sheets present in the secondary structure shows no apparent changes in any of the analyzed proteins (Figure 4B). Proteinase K assays demonstrated that mutant R104A suffered less degradation than other LotP variants, suggesting small changes in protein structure, making the protein less prone to degradation (Supplementary Figure 3).

The formation of multimers suggested that this protein could form higher-level structures like the active conformation of many proteolytic complexes. The structural similarities of the main protein scaffold with the substrate binding domain of the Lon protease prompted us to evaluate protein interactions between LotP and the Lon protease.

It was previously reported that LotP from ‘Ca. L. asiaticus’ interacts with different components of the proteostasis network found in E. coli, Sinorhizobium meliloti, and ‘Ca. L. asiaticus’ (Loto et al., 2017). The Lon protease, similar in structure to LotP, belongs to this same big group of proteins (Cho et al., 2015). Given this connection, the ‘Ca. L. asiaticus’ Lon protease was cloned, and its interaction with LotP was evaluated using a bacterial two-hybrid system (Vallet-Gely et al., 2005). A significant interaction between Lon protease and LotP was observed (Figure 5A).

After the determination of this interaction, we studied the importance of the N-terminal of these proteins, as well as residue R104 of LotP, in this interaction. When comparing the Lon protease of ‘Ca. L. asiaticus’ with other Lon proteases, there is a distinct extension to the N-terminal (Supplementary Figure 4). The N-terminal of this Lon protease contains several additional charged amino acids (EDESKDR), thereby showing sequence similarity to the EK sequence used in the LotP constructs described above. We hypothesized that this sequence could regulate the interaction between these two proteins. We constructed two chimeras by adding the N-terminal of the Lon protease (20 amino acids) to LotP.His and ΔNLotP.His, creating N-terminal(Lon)-LotP.His and N-terminal(Lon)-ΔNLotP.His, respectively. Surprisingly, both proteins eluted as dimers when analyzed by size-exclusion chromatography (Figure 5B). These results were comparable with those described for the construct His.EK.LotP and His.EK.LotP.FLAG (Figure 2B)—all with important modifications in the N-terminal. These results confirmed the critical contribution of the N-terminal domain sequence in oligomerization.

To assess if the N-terminal region mediates LotP–Lon interaction, we designed a new two-hybrid assay. The experiment was carried out using LotP and a Lon protease missing the first 23 amino acids of its N-terminal (ΔNLon). In this assay, the ω subunit was fused to LotP, while the Zif domain was fused to either Lon protease constructs, as this orientation provides optimal interfacing with LotP. The interaction of ΔNLon with LotP decreased by 50% when compared with interactions between the control Lon and LotP (Figure 6A). These results clearly demonstrate the importance of the Lon N-terminal sequence for the interaction between these two proteins. Interestingly, when mutant LotP^R104A was used in the same two-hybrid analysis, the β-galactosidase activity was arrested (Figure 6B).

To acquire in vivo evidence in support of the molecular mechanisms described above, we designed a series of assays using the E. coli Lon protease as a central interaction target. As expected, LotP expression in a wt E. coli strain slightly affects cellular growth. Interestingly, this effect was not observed when LotP was expressed in a Δlon isogenic strain (Figure 7A). When LotP^R104A was expressed, it affected the growth of the wt strain but did not affect the growth of the Δlon mutant strain (Figure 7B). A Western blotting analysis indicated comparable protein abundance in all strains analyzed, validating the results observed (Supplementary Figures 5A,B). These results suggest that the presence of LotP may boost the activity of the Lon protease with deleterious effects in the cell physiology. It has been reported that high amounts of Lon protease negatively affect the growth of E. coli (Sonezaki et al., 1994).

To verify our hypothesis, we performed several assays using SulA, a well-known Lon protease target. This protein is a member of the SOS regulon, and its expression is triggered by DNA-damaging agents (Simmons et al., 2008). Once expressed, it inhibits bacterial septation by binding to FtsZ, stagnating the cellular growth. However, SulA is quickly degraded by the Lon protease, allowing for immediate resumption of division (Ishii et al., 2000). SulA was expressed from the plasmid pCA24N, while either LotP or LotP^R104A were expressed using pBAD24. After SulA expression was induced with 100 μM of IPTG, the growth of the E. coli wt strain was significantly affected. In the absence of the Lon protease (Δlon mutants), the overexpression of SulA completely arrested the cellular growth. The wt growth was partially restored when LotP or LotP^R104A was induced (Figure 7C). A Western blotting revealed that SulA was only visualized when expressed in absence of LotP or LotP^R104A (Supplementary Figure 5C). These results suggest enhanced Lon protease activity when LotP is expressed. Additional assays are necessary to elucidate the results observed with the mutant protein.

To analyze and validate the biological relevance of LotP on Lon protease activity, a new assay was designed. SulA is known to become insoluble in the E. coli cytoplasm when it is highly expressed (Ishii et al., 2000). We used a chimerical SulA substrate to measure Lon’s protease activity that does not affect E. coli growth. The chimerical substrate was done using SulA carboxyl terminal (11 amino acids) fused to the maltose-binding protein for further purification (MBP.SulA). The results obtained indicated faster degradation of MBP.SulA in the presence of LotP (Figure 8A, lane 3) when compared with the control (Figure 8A, lane 2). Once His.EK.LotP (dimer) was co-expressed with Lon, the MBP.SulA band (Figure 8B, lane 4) at 45 min was similar to the Lon control (Figure 8B, lane 2). These results suggest that the modified protein, His.EK.LotP, had no additional affect over Lon protease activity.

The ability of LotP to be highly expressed in citrus plant during ‘Ca L. asiaticus’ pathogenicity and its capability to interact with a variety of chaperones led us to assess its potential effects on the citrus plant response.

Healthy citrus plant leaves were infiltrated with purified His.LotP protein at different concentrations. His.LotP infiltration produced a chlorotic phenotype clearly visible at 3 dpi (days post infiltration) (Figure 9A). To study whether photoinhibition was responsible for chlorosis development, photosynthetic efficiency of the PSII (Fv/Fm) was measured. This index is a sensitive indicator of plant photosynthetic performance and can be used as a direct indication of plant stress and/or photoinhibition. In fact, a decrease in F_v/F_m has been observed in HLB-infected leaves (Sagaram and Burns, 2009; Cen et al., 2017). Three different concentrations of His.LotP were infiltrated into citrus leaves. Infiltrated His.LotP demonstrated a dose-dependent effect in a range of 4–12 μM (Supplementary Figure 6A). A concentration of 8 μM was selected as the minimal protein concentration required for a reproducible and measurable response. After His.LotP infiltration, F_v/F_m decreased significantly in the first 2 dpi compared with the values measured in a different lobe of the same leaf that was infiltrated with the vehicle control. A zigzag pattern was displayed over a 10-day period until F_v/F_m levels reached values comparable with those of the control (Figure 9B). The ground fluorescence (F₀) remained relatively constant during the assay (Figure 9C); however, the F_v/F₀ followed a similar zigzag pattern until it reached the levels similar to those of the control at 9 dpi (Figure 9D). These results clearly demonstrated a decreased photosynthetic rate after His.LotP infiltration. In order to evaluate if LotP oligomerization could be responsible of such effects, several citrus leaves were infiltrated with the mutant protein His.EK.LotP.FLAG. No significant changes in the F_v/F_m parameter were found using this modified protein (Supplementary Figure 6B). This suggested that the biological activity of LotP depends on the ability to form high molecular complexes. Altogether, analysis of LotP infiltrations suggests that the protein may interfere with the proper chloroplast function in citrus leaves, provoking a chlorotic phenotype—two main features observed in HLB-infected plants (Bové, 2006).

The symptoms displayed after His.LotP infiltration prompted us to evaluate its overall effect on the plant proteomic response. To maximize the possibilities to identify changes in the proteomic profile, samples collected during the first 3 days of the response were evaluated. The infiltration assay was repeated using a solution containing 8 μM of His.LotP in identical conditions as described before. The F_v/F_m was measured every day, and leaf samples were collected during the first 3 dpi. Collected samples were processed and analyzed with TMT11 MS/MS. The FDR was calculated using the Percolator algorithm in the Proteome Discoverer workflow based on the search results against a decoy database and was set at 1% FDR. Duplicates were grouped, and statistical analyses evaluated as the quantitative ratio between each control and treatment per day. Furthermore, proteins identified, which contained at least two peptides with a p-value of <0.05 and log2 fold ratio greater or less than 0.5, were used for final analyses. Overall, a total of 14,958 unique proteins were identified.

A total of 220 proteins were found to be differently abundant in the period analyzed (first 3 dpi) (Supplementary Table 2). The subsets of proteins with significant differences compared with the control were grouped according to their biological roles and separated by more and less abundant proteins (Figures 10, 11). To better describe the plant response after His.LotP infiltration, the proteins identified were divided into three groups, namely, immediate response (1 dpi), intermediate response (2 dpi), and delayed response (3 dpi) (Table 2). These names do not describe a canonical plant response; they were merely used to group the proteins identified in the window of time examined.

A subset of 49 proteins significantly increased their abundance at 1 dpi. This group is characterized by the abundance of RNA processing proteins involved in cleavage capping and mRNA polyadenylation. The proteins included in this group were CPSF30, HMG-B (high motility group B proteins), LE ribonuclease, CPFS6, polyadenylate-binding protein 11 (PAIP), and UBP1, suggesting that the plant immediate response is directed to enhance the expression and stabilize critical mRNAs. Interestingly, we also found CPB20/80. These cap binding proteins are involved in the biogenesis of miRNA controlling several plant processes (Kim et al., 2008). With less intensity, but consistently induced, the plant specific PROS regulator was also highly represented. PROS activates SKD1 ATPse activity of the ESCRT III system, controlling the endosomal vesicle trafficking in plants. Two chloroplast proteins were also highly induced after LotP infiltration. Those proteins were part of the Calvin cycle and the photosystem I, supporting the impact of LotP on photosynthesis, as described above. Curiously, APX and Cu/ZnSOD, two important members of the reactive oxygen species scavenging system, were repressed together with a group of glycosylases, peptidases, and oxidases.

The response observed at 2 dpi was less intense in terms of peptide abundance but consists of a larger variety of proteins when compared with the 1 dpi response. Likewise, the RNA processing proteins were dominant. In this group, we also detected a significant increase in the number of transcription factors that respond to biotic and abiotic stresses (Alfin, C2H2-ZF, NTL9, and TCP). Interestingly, GeBP and CsWRKY21 regulators, highly abundant 1 dpi, reached statistical significance at 2 dpi. There was, as well, a significant abundance of proteins involved in protein quality control. These include the lectin chaperone CRT, the HPS60 co-chaperone HSP20, and HSP90. A subset of proteins involved in organelle reconstruction and cell transport were also identified together as well as proteins involved in cell cycle control and autophagy. Hydrolases involved in carbohydrates metabolism, oxidoreductases, and a subgroup of poorly characterized proteins described as potential defensins were negatively affected by the presence of LotP.

Only 11 proteins were observed to be differentially abundant in leaves collected at 3 dpi. This observation relates to a recovery of the PSII activity as previously described. It is important to indicate that three of them were chaperones, while the transcription factors GeBP and CsWRKY21 were still highly expressed. Consistently with the less abundant proteins described in the first two groups, defensins, dehydrogenases and oxidoreductases were negatively affected.

The chimerical His.EK.LotP.FLAG, a protein mainly forming dimers, was used in a parallel assay. As it was expected, the pattern of proteins affected was similar, but the intensity of the response for each protein affected was less intense judging by the relative abundance of the proteins detected. These results reinforce the idea that biological activity of LotP depends on the ability to form high-molecular-weight multimers.