Lipid Profile of Xylella fastidiosa Subsp. pauca Associated With the Olive Quick Decline Syndrome

Lipids, components of the plasma and intracellular membranes as well as of droplets, provide different biological functions related to energy, carbon storage, and stress responses. Bacterial species display diverse membrane composition that changes in response to the different environmental conditions. During plant–pathogen interactions, lipids might have roles in several aspects such as recognition, signal transduction, and downstream responses. Among lipid entities, free fatty acids (FFAs) and their oxidized form, the oxylipins, represent an important class of signaling molecules in host–pathogen perception, especially related to virulence and defense. In bacteria, FFAs (e.g., diffusible signaling factors) and oxylipins have a crucial role in modulating motility, biofilm formation, and virulence. In this study, we explore by LC-TOF and LC-MS/MS the lipid composition of Xylella fastidiosa subsp. pauca strain De Donno in pure culture; some specific lipids (e.g., ornithine lipids and the oxylipin 7,10-diHOME), characteristic of other pathogenic bacteria, were revealed. Nicotiana tabacum was used for testing the ability of this pathogen in producing such lipids in the host. Different lipid compounds present a clear distribution pattern within the infected plant tissues compared to the uninfected ones.


INTRODUCTION
Among bacterial plant pathogens, Xylella fastidiosa is one of the most dangerous bacteria causing devastating diseases and showing an extensive natural host range. X. fastidiosa, with diverse modalities related to the subspecies and the host, causes different diseases such as Pierce's Disease (PD) of grapevine, phony peach, leaf scald of plum, citrus variegated chlorosis, olive quick decline, and leaf scorch in almonds, coffee, and oleander (Wells et al., 1987;Hopkins, 1989). Recently, Saponari et al. (2017) assessed the pathogenic role of X. fastidiosa subsp. pauca strain De Donno in olive and other susceptible host plants.
Lipids play important roles at various stages of host-pathogen interactions (van der Meer-Janssen et al., 2010;Siebers et al., 2016;Sohlenkamp and Geiger, 2016) and are crucial in determining the virulence of bacterial pathogens (Martinez and Campos-Gomez, 2016). Free fatty acids (FFAs) might also function as modulators of several pathways in bacterial cell-to-cell communication such as the diffusible signaling factor (DSF). Notably, DSF acts as regulator of biofilm formation and as virulence factor in several plant bacterial pathogens, as for instance X. fastidiosa (Dow, 2017). In X. fastidiosa, the DSF family can also participate in the inter-kingdom communication with plants or insects (vector) (Ionescu et al., 2016). Unsaturated fatty acids (FAs) may also act as substrates for oxidizing enzymes [e.g., lipoxygenases (LOXs) and dioxygenases (DOXs)] forming oxylipins that have been extensively studied in plantpathogen interaction (Blèe, 2002;Christensen and Kolomiets, 2011). The oxylipins, per se or conjugated with sugars and aminoacids, are bioactive molecules; the oxylipin jasmonic acid and its derivatives in plants mediate hormone-like functions and are involved in defense responses (Jones and Dangl, 2006). Notwithstanding their importance, the role of oxylipins is almost underestimated and understudied in phytopathogenic prokaryotes. Very recently, Martinez and Campos-Gomez (2016) show that the opportunistic bacterial pathogen Pseudomonas aeruginosa may transform monounsaturated FAs into mono-and di-hydroxylated derivatives during its interaction with the host (e.g., lettuce). In this pathogen, the oleic acid-derived oxylipins negatively control the motility of flagella, cause the upregulation of twitching motility and promote bacterial organization in micro colonies and the formation of biofilms in vitro and in vivo, controlling the virulence in the host (Martinez and Campos-Gomez, 2016).
Lipids play important roles in plant disease (Siebers et al., 2016;Sohlenkamp and Geiger, 2016); the membrane lipid compositions can vary among bacterial species depending on the environmental conditions to which the organism is exposed (Giles et al., 2011;Lewenza et al., 2011;Vences-Guzmán et al., 2011;Moser et al., 2014;Sohlenkamp and Geiger, 2016). Glycerophospholipids of the bacterial membranes present a hydrophobic tail composed by two FAs, a glycerol backbone, a hydrophilic head of phosphate and groups such as phosphatidylethanolamine (PE), phosphatidylglycerol (PG), cardiolipin (CL), lysyl-phosphatidylglycerol (LPG), phosphatidylinositol (PI), phosphatidic acid (PA), and phosphatidylserine (PS). Bacterial membranes present also phosphorus-free lipids and notably ornithine/glutamine lipids (O/GlnLs), diacylglyceryl-N,N,N-trimethylhomoserine (DGTS), sulfolipids, monoacylglycerol (MAG), diacylglycerol (DAG), triacylglycerol (TAG), glycolipids (GLs), and hopanoids (BHPs) (Sohlenkamp and Geiger, 2016). Phospholipids can be replaced within the membrane lipids of bacteria with phosphorusfree lipids; this remodeling has been thoroughly studied in Rhodobacter sphaeroides, Sinorhizobium meliloti, Agrobacterium tumefaciens, and Mesorhizobium loti (Benning et al., 1995;Geiger et al., 1999;Zavaleta-Pastor et al., 2010;Devers et al., 2011;Geske et al., 2013;Diercks et al., 2015). Bacterial phosphate-free membrane lipids and in particular OLs and their hydroxylated forms, are important for interaction with plants ( Vences-Guzman et al., 2013). Some bacteria form OLs only under phosphorous-limiting conditions; in others, OLs are formed constitutively. Vences-Guzman et al. (2015) estimate that about 50% of the bacteria can produce OL. A mutant of the plant pathogen Agrobacterium fabrum (formerly: A. tumefaciens C58), lacking hydroxy-OL or any OLs, anticipates the formation of tumors that are even bigger than those produced by the wild type upon plant infection. Vences-Guzman et al. (2013) hypothesize that the recognition of OL or hydroxy-OL might elicit plant defense responses; A. fabrum devoid of OL or hydroxy-OL, escape the plant immune system thus inducing an accelerated infection process. In Rhizobium tropici and Burkholderia cepacia, the 2-hydroxylation of OL is also involved in the growth at higher temperature condition (Taylor et al., 1998;Vinuesa et al., 2003;Vences-Guzmán et al., 2011). Among the phosphorus-free membrane lipids, BHPs are present in a wide variety of prokaryota and have structural similarities with eukaryotic sterols (Saenz et al., 2012). BHPs enhance the stability and impermeability of the bacterial membranes. Strains of Burkholderia cenocepacia, defective in BHPs production, display increased sensitivity to low pH, detergents, and various antibiotics and cannot produce flagella (Schmerk et al., 2011). Within phosphorous-containing lipids, some authors demonstrated that PC, which in bacterial membranes might account up to > 20% of total phospholipids (Klüsener et al., 2009), are involved in the virulence of Agrobacterium tumefaciens. Deletion of A. tumefaciens pmtA partly impaired the synthesis of PC, delayed tumor formation that is reduced in size (Wessel et al., 2006). Furthermore, the authors highlight that tumor formation is absent when the host is infected with the PC-free double mutant pmtA and pcs of A. tumefaciens. Pseudomonas syringae pv. syringae requires PC for virulence and specifically for secreting, through the type III secretion system, the effector HrpZ (Xiong et al., 2014).
In phytopathogenic bacteria, different type of lipids may drive the compatibility or incompatibility with the host. In relation to this, the present study first investigates by mass spectrometry the lipid composition of X. fastidiosa subsp. pauca strain De Donno under in vitro conditions. This approach allows individuating and identifying several lipid compounds produced by this pathogen in the cell as well as in the culture media. Second, the model plant Nicotiana tabacum was inoculated with this bacterial pathogen. Mass spectrometry analysis shows a differential accumulation of lipid entities among infected and non-infected plants; notably, some FFAs, complex lipids and oxylipins could play a role during plant colonization.

Bacterial Strains and Culture Conditions
Xylella fastidiosa subsp. pauca strain De Donno (CFBP 8402) (from here XfCFBP8402) grown for 7 days at 28 • C in buffered charcoal yeast extract (BCYE) was harvested from the medium (Wells et al., 1987) and suspended in sterile potassium phosphate buffer (0.05 mM pH 7.2; PBS) to a final concentration of approximately OD 600 of 0.5, corresponding to ca. 10 8 colony forming unit CFU/mL, as reported in Giampetruzzi et al. (2016). Aliquots (100 µL) of this bacterial suspension were used to inoculate PD2 liquid medium (Davis et al., 1980) and grown for 7 and 11 days at 28 • C and 100 rpm. A total of 10 independent experiments (n = 6 in each experiment) for the lipid extraction were performed.

Nicotiana tabacum Inoculation
Nicotiana tabacum "Petite Havana SR1" were propagated in a greenhouse as reported in the European Plant Protection Organization (EPPO) Standard PM 7/24 (2) (European Plant Protection Organization, 2016) and prepared before inoculation as described previously (Francis et al., 2008). Briefly, the apical part of the stem was cut while removing the lower juvenile leaves, thus remaining the sole three adult leaves (numbered as 1-3) in the lower portion of the plant. The plants were inoculated with XfCFBP8402 and control plants (named mock) were treated only with the buffer. The inoculum concentration was prepared at approximately 10 9 CFU/mL. The infections were performed at least in three independent experiments, each comprising infected (n = 20) and mock plants (n = 20). Each plant was inoculated (20 µL bacterial suspension or buffer) in the petioles near the axils of the three healthy adult leaves 1-3. A total of 15 days after inoculation (DAI), the petiole of the first leaf (numbered as 4) above leaves 1-3 was collected from each plant and the DNA was extracted according to the modified DNeasy R Mericon TM Food standard kit (Qiagen) following manufacturers' instructions. The presence of bacterial DNA was confirmed by real time PCR as indicated by Harper et al. (2010). Among the infected specimens, only the plants positive to XfCFBP8402 were selected for further analysis. The same real time PCR procedure was performed to confirm the mocks as XfCFBP8402-negative. A total of 30 DAI, the first two leaves (numbered as 5 and 6) upon the leaf 4 were collected from at least 15 XfCFBP8402-positive plants (from here named as "Xf-infected") and grouped in a bulk. The samples were separated in two parts: one (from here named "petiole") composed by the petioles and central vein and the other (from here named "leaf ") composed of the marginal leaf. Petioles and leaves were lyophilized and ground with liquid nitrogen. The same collection procedure was followed for the mock plants. The bacterial DNA amount in the petioles and leaves of XfCFBP8402-infected samples as well as in the mock samples was evaluated as reported in Modesti et al. (2017). In particular, DNA was extracted using the modified DNeasy R Mericon TM Food standard kit (Qiagen) following the manufacturers' instructions and amplified by quantitative real time PCR (Harper et al., 2010). Lipid extraction from the petioles and leaves of XfCFBP8402-infected and mock plants was performed as described below.

Lipid Extraction
Lipids were extracted from pelleted cells and lyophilized culture filtrate of XfCFBP8402 at different DAI (7-11 DAI), and from tobacco plants, Xf-infected and mock, at 30 DAI. The internal reference standards tricosanoic acid and 9-HODEd 4 , for the analysis were added at a final 2-µM concentration. The extraction was performed on pelleted bacterial cells or 30 mL of lyophilized culture filtrate or 20 mg of lyophilized plant material that were extracted with 2 mL of iPrOH: H 2 O: EtOAc (1:1:3 v/v); butylated hydroxytoluene (0.0025% w/v) was added to avoid oxidation. After centrifugation, the ethyl acetate upper phase was collected in a clear tube and dried with nitrogen. The extraction was repeated on the initial matrix adding 1.2 mL of EtOAc and then vortex-mixing. After centrifugation, the upper phase was recovered and transferred to the collection tube together with the previously extracted fraction, and dried under nitrogen flux. The dried samples were resuspended in MeOH (100 µL).

Nomenclature and Abbreviations Used to Describe Lipid Components
Notation of lipids common to eukaryote and prokaryote such as MAG, DAG, TAG, glycerophosholipids was done as reported in Camera et al. (2010). In relation to bacterial-specific lipids, BHPs was used for indicating the hopanoids while OL was the abbreviation for ornithine lipids (OLs). In these bacterial lipids, the ornithine head group is bound via its N α -amino group to a 3-OH FA (R 1 ), with a second FA chain (R 2 ) esterified to the 3-OH group of the first FA. Thus, OL3-OH (18:1/19:1) indicates an ornithine esterified with an oleic acid that is subsequently esterified with a non-adecenoic acid (Sohlenkamp and Geiger, 2016). Notation for FAs and oxylipins (OM/D/TrE) is reported as indicating the carbon number (CN) and the number of double bond (DB) equivalents (e.g., C18:1 were oleic acid and HODE hydroxyoctadecenoic, respectively). The short notation for free FA is FFA (Liebisch et al., 2013).

Lipid Analysis
The HPLC runs and accurate mass measurements of lipids were conducted with a G6220A TOF-MS, (Agilent Technologies, United States) equipped with an electrospray interface operating in the negative as well as in positive ion scan mode (m/z 100 1200) as indicated in Camera et al. (2010) (Supplementary  Figures S1A,B). Our attention was focused on specific classes of lipids and namely: MAG, DAG, TAG, BHP, OL, some of the most represented phospholipids (i.e., PC, PE, and PG), FFA, and oxylipins.
Some lipid classes were further analyzed (fragmentation analysis) by liquid chromatography (HPLC 1200 series rapid resolution) coupled to the triple quadrupole G6420A (Agilent Technologies, United States) equipped with an electrospray ionization (ESI) source. The same elution conditions (phases, gradient, and column) described above were used.
Several lipid classes were searched through neutral loss (NL) and/or precursor ion mode in the XfCFBP8402 pelleted cells, culture filtrate and plant samples. Specifically, PC can be identified by scanning in positive ion mode for precursors of m/z 184.2 while dehydrated OLs at m/z 115.2 (Supplementary Table S1A). By scanning precursors in negative ion mode it was possible to individuate several species of OLs (precIon at m/z 131.2, Zhang et al., 2009;Moore et al., 2016). PEs were searched by NL at m/z 141.1 (Brügger et al., 1997). Product ion (PI) scan mode was used for characterizing mass fragmentation pattern of several lipid compounds (Supplementary Table S1B Table  S1C) in the samples. Standard curves in the range of 0.1 µM to 4 mM were originated for each FA listed in Supplementary Table S1C. Rate was linear within the range and regression curves [limit of detection (LOD) and limit of quantification (LOQ)] were originated to calculate the exact amount of each FA in real samples (Supplementary Table  S2). Further, in plant samples, a multiple reaction monitoring (MRM) method (with the same chromatographic settings) was adopted to analyze the most abundant lipid entities in bacterial cells; the fragmentor voltage (F) and collision energies (CE) were optimized for each compound as previously described. Bacterial and plant samples were analyzed for the presence of oxylipins as reported in Ludovici et al. (2014). Flow injection of authentic standards comparison with the literature (Yang et al., 2009;Strassburg et al., 2012;Ludovici et al., 2014;Martinez and Campos-Gomez, 2016) was used to confirm MRM transitions; for some compounds not reported in previous studies, fragmentation patterns were identified in bacterial cell extracts (e.g., jasmonates; Balcke et al., 2012; Supplementary Figure S2).
The MRM data were processed using the Mass Hunter Quantitative software (B.07.00 version) (Supplementary

Statistics
Average RT, peak areas, and mass accuracy of identified lipids were calculated for the analyzed samples as indicated in Camera et al. (2010). XLSTAT Version 2015.3.01.19199 (Addinsoft, Paris, France) as statistic package. In each experiment, datasets were pooled and compared using Student's t-test, and the differences were considered significant when the p-value was < 0.05.

Lipid Profile
The lipidomic profile of cells and cultural filtrate of XfCFBP8402 was acquired at different time intervals of bacterial growth, i.e., 7 and 11 DAI in liquid media (see the Section "Materials and Methods"). Results of TIC chromatogram are reported in Supplementary Figures S1A,B. LC-TOF chromatograms were searched for specific classes of lipids namely, MAG, DAG, TAG, OL, PE, PC, BHPs, FFA, and oxylipins. The analysis was performed basing on referenced data (e.g., Talbot et al., 2003;Han, 2016;Sohlenkamp and Geiger, 2016, inter alia), databases 1,2 and by comparison with authentic standards. To assign properly the entities found we followed an approach exploiting MS/MS analysis.

Ethanolamine Glycerophospholipids
In our samples, the [M+H] + ion of 2-diheptadecanoylsn-glycero-3-phosphoethanolamine (PE 34:0; m/z 720.5538; C 39 H 79 NO 8 P) was present at a comparable RT (17.8 min) (Supplementary Figure S8A). For structural elucidation, under positive ion mode we detected fragment ions in the PI scans of the standard PE 17:0/17:0 as well as in XfCFBP8402 cell extracts consistent with the abundant NL of phosphoryl ethanolamine ion (m/z 141.1) (Supplementary Figures S8B,C).  Figures S8D-I). The "O-" prefix is used to indicate the presence of an alkyl ether substituent, whereas the "P-" prefix is used for the 1Z-alkenyl ether substituent. It was not possible to deduce PE identities since fragmentation did not provide insights in FA composition (loss of ethanolamine).

Analysis of Specific Lipid Groups
Basing on the lipid entities found above, we measured the relative abundance of several MAG, DAG, TAG, OL, BHP, PG, PC, and PE in the cell extracts of X. fastidiosa CoDiRO strain CFBP8402 at different times of growth (7-11 DAI) (Supplementary Figure S1A, Table S4). The Figure 1 shows that the relative abundance of lipid entities differed significantly among 7 and 11 DAI for the sole PG (p < 0.01) and PE (p < 0.05).

Targeted Analysis of FFA and Oxylipins
The FFA profile shows the presence of multiple compounds (Figure 2 and Supplementary Table S5): main FFA was stearic (C18:0), followed per abundance by oleic (C18:1), palmitic acid (C16:0), linoleic (C18:2), and heptadecenoic acid (C17:1). Unsaturated FA (UFA) and polyunsaturated fatty acids (PUFA) such as myristic, palmitoleic, and linolenic acid (C14:0, C16:1, and C18:3, respectively) even if in minor amount were present in our samples. The abundance of the different FFA was similar in cell and in culture filtrate, with the sole exceptions of palmitoleic (XfDSF2) and oleic acid which are more abundant at 11 DAI in the culture filtrate than in the cell extracts (Figure 2). In our samples, we also detected XfDSF1 in trans and cis forms (C14:1 t and c, respectively; Figure 2). Every FFA, significantly (p < 0.001) changed during bacterial growth. Xylella fastidiosa produced several oxylipins deriving from oleic, linoleic and linolenic acid ( Table 9 and Supplementary  Table S6). Some oxylipins are produced at significant, micromolar amount, in the bacterial cells as well as in the culture filtrate. In Table 9, we showed the sole oxylipins present in significant amount; a more comprehensive view of the complete profile is reported in Supplementary Table S6. Among the most abundant compound, the products of oxylipins derived   Results are the mean ( ± SE) of 10 independent experiments (n = 6 in each experiments).

Infection of a Model Plant: Nicotiana tabacum
The presence of XfCFBP8402 in experimentally inoculated tobacco tissues was confirmed through real-time qPCR. Its distribution differed significantly (p < 0.001) among petioles and leaves (Figure 3); the pathogen was not detected in the mock plants. In particular, the DNA of XfCFBP8402 was higher in the leaves than in the petioles of Xf-infected plants (Figure 3). The lipid profile of the leaves and petioles of the Xf-infected and mock plants was studied. The analysis was focused on those lipid entities that were most abundant in the previous in vitro screening of XfCFBP8402 (see above; Figures 4A-C). The results were reported as the amount of lipid entities in the XfCFBP8402infected plants compared to uninfected ones (mock). Raw data (relative abundance of each lipid entities into plant tissues) are presented in Supplementary Tables S7-S9.
After 30 DAI, all FFA were more abundant in Xf-infected samples than in the mock, in the leaves as well in the petioles ( Figure 4B). In particular, C14:1 (XfDSF1) was particularly abundant in the petioles while the C16:1 (XfDSF2) in the leaves.
The Figure 4C shows the oxylipins derived by oxidation of oleic, linoleic and linolenic acid, respectively. In the Figure 4C, the distribution of the oxylipins appeared consistently different between the petioles and the leaves; notably, mostly upregulated in the petiole and downregulated in the leaves. In our experimental conditions, 10-HpOME was more abundant in the petioles than in the leaves; conversely, 7,10-diHOME is more abundant in the leaves than in the petioles. The oxylipins derived by linoleic acid are more abundant in the petioles than in the leaves, made exception 12(13)-epOME and 13-HODE. In the petioles of the Xf-infected plants, 9-HpOTrE is the prevailing oxylipin derived from linolenic acid together with JA, whereas 13-HpOTRE is more abundant in the leaves.

DISCUSSION
Within the last decade, an increasing number of studies identified lipids in phytopathogenic bacteria and described their role in the interaction with the hosts. Lipids are essential constituents of the cells providing different functions ranging from structural to energy storage and signal mediators (Lim et al., 2017). Escherichia coli has been the standard model to study bacterial lipids for years; very recently, Sohlenkamp and Geiger (2016) reported the diversity of membrane lipids in a multitude of eubacteria. This "burst" in lipid surveys in bacteria relies also on the stepping up of analytical methods essentially based on coupling chromatography with mass spectrometry.
In the present study, we have addressed an analytical method integrating exact mass data (LC-TOF) with mass fragmentation (LC-MS/MS) to detect multiple classes of lipids in the XfCFBP8402 associate with severe epidemics in centenarians olive trees in Southern Italy. Currently, a holistic approach is under way for containing this outbreak. Studying the bacterial lipid diversity could aid the individuation of the lipid entities potentially involved in the interaction with host. Through this Different authors (Sohlenkamp and Geiger, 2016;Rowlett et al., 2017) report that nutrient availability, accumulation of products of metabolism, pH variation, oxygen levels, biofilm formation, and surface adhesion significantly affect the lipid composition of bacteria grown under in vitro conditions. In our study, we analyze the lipid composition of the bacterial culture at 7 and 11 DAI to check if lipid entities composition change during the growth phase as reported by Feil and Purcell (2001). Notably, within this time frame (7-11 DAI) in our growing conditions, XfCFBP8402 shifts from a planktonic behavior to form the characteristic cell ring at the air-liquid interface. By combining TOF and MS/MS analysis, we succeed in identifying different entities within each lipid class: 18 TAGs,13 DAGs,15 MAGs,29 PGs,20 PCs,7 PEs,8 BHPs,11 OLs, 23 FFA, and 24 oxylipins, respectively. The relative abundance of MAG, DAG, TAG, OL, BHP, PG, PC, and PE indicated that MAG and PL fractions prevailed in bacterial cells. With current data, we can suggest that lipid changes occurring between 7 and 11 DAI may influence the cells physicochemical properties like membrane-protein topology, inner and outer membrane transport, modulate the interaction among bacterial cells (e.g., quorum sensing), shift from planktonic growth to biofilm formation (Bakholdina et al., 2004;Zhang and Rock, 2008;Bisbiroulas et al., 2011).
The FFA fraction represents either an important reservoir of signal molecules per se or as precursor readily converted to oxylipins (Yaeno et al., 2004;Siebers et al., 2016). Within the class of FFA, DSFs are gaining momentum. DSF families have been described in different bacterial pathogens; these (mono)unsaturated FAs (UFA) are involved in interspecies and inter-kingdom signaling and recognition. In Xylella and Xanthomonas, different authors indicate that DSF signals are fine-tuned during interaction with the host plants since they are recognized as elicitors, thus triggering innate immune response in plants (Torres et al., 2007;Chatterjee et al., 2008a;Kakkar et al., 2015). X. fastidiosa utilizes one or more of such signal molecules to shape its lifestyle depending on cell density (Ionescu et al., 2016;Dow, 2017). In X. fastidiosa, RpfF -a bifunctional crotonase with both dehydratase and thioesterase activities -produces a mixture of DSF species (from C14:1 to C19:1) during the invasion of the xylem of host plants (Wang et al., 2004;Ionescu et al., 2016). X. fastidiosa DSF-deficient mutants are more virulent but less capable of colonizing the insect vector and infecting healthy plants (Newman et al., 2004;Chatterjee et al., 2008b). Currently, at least two DSF, XfDSF1 (C14:1) and XfDSF2 (C16:1) were thoroughly investigated; XfDSF2 is apparently more active as signaling molecule compared to XfDSF1 (Ionescu et al., 2016).
In this paper, we show that in culture filtrate (and in cell extracts as well) of XfCFBP8402 strain, XfDSF2 is more abundant compared to XfDSF1 that, in turn, is more abundant at 7 DAI than 11 DAI of growth. In general, this strain appears competent in producing several UFA (up to C24:1) at least as cell components (FA in complex lipids); indeed, oleic (C18:1) and palmitic acid (C16:1) are the main UFA in FFA fraction. Potentially, other XfDSFs could have a role in different aspects of XfCFBP8402 lifestyle; in fact, both C17:1 and C18:1 accumulate at significant levels in the culture filtrate at 11 DAI. It is suggested elsewhere that different UFAs can have different activities in other strains of X. fastidiosa: from quorum sensing to toxicity (Ionescu et al., 2016).
Moreover, FFA can function as substrates to form oxylipins. The biosynthesis of oxylipins occurs both constitutively and consequently to abiotic and biotic stresses. Oxylipins modulate several processes in prokaryotic and eukaryotic cells. In mammals, plants, and fungi, the biosynthetic pathways and physiological roles of some oxylipins are almost well defined: e.g., prostaglandins in mammals, jasmonates in plants, hydroxyacids in fungi (Blèe, 2002;Andreou et al., 2009;Christensen and Kolomiets, 2011;Scala et al., 2014). According to the available literature, oxylipins might derive from LOX as well as DOX activities, although they can also be produced by non-enzymatic chemical oxidation of FAs (Christensen and Kolomiets, 2011;Banthiya et al., 2016;Martinez and Campos-Gomez, 2016). For a long time, oxylipins studies had been limited to eukaryotes (Andreou et al., 2009); nevertheless, the first two prokaryotic LOX sequences were described (Vance et al., 2004) and numerous oxylipins were subsequently characterized; some of these oxylipins regulate the host-pathogen interaction (Porta and Rocha-Sosa, 2001). Recently, in P. aeruginosa the cellular function of oxylipins has gained attention. Among these, (10S)-hydroxy-(8E)-octadecenoic acid (10-HOME), and 7S,10Sdihydroxy-(8E)-octadecenoic acid (7,10-diHOME) are depicted as crucial in determining a switch in bacterial lifestyle; namely, these oxylipins are required for biofilm formation when bacteria interacts with host cells (Martinez and Campos-Gomez, 2016). The authors here suggest that the expression of oxylipin forming genes might alter lipid signaling during the interaction with the host, impact biofilm formation, and improve the invasiveness of the bacteria. The simultaneous detection of oxylipins in the cells as well as in the cultural filtrate suggests indeed that although oxygenation may occur inside the cell, oxylipins are transported through the outer membrane and accumulate in the medium (Martinez et al., 2013).
In this paper, we show that different oxylipins deriving from oleic, linoleic, and linolenic acid are detected both in the cells and in the cultural filtrate of X. fastidiosa CoDiRO CFBP8402. Among the most abundant compounds, oxylipins derived from oleic and linolenic acid are dominant. Notably, 10HpOME, 10-HOME, and epOMEs even with different intra/extra-cell distribution prevailed over the others oxylipins. Our results confirm previous findings in P. aeruginosa reporting that oxylipins mainly accumulate in the extracellular medium and synthesis ascribed to DOX and diol synthase activities (Martinez et al., 2013;Martinez and Campos-Gomez, 2016). Other, less represented, oxylipins such as 13-HODE 9-HODE, 8,13-diHODE, 13HOTrE, and methyl jasmonic acid, whose synthesis is related to LOX enzymes (Christensen and Kolomiets, 2011) are found into the cells as well as in the culture filtrate. Intriguingly, the wellknown plant stress hormone (Nomura et al., 2005) -methyl jasmonate -is produced and, overall, secreted into consistent amount. This compound could play a role in modulating plant defenses if produced even during plant infection. The evidence that these oxylipins were more abundant in culture filtrate, suggesting their secretion and pave the way for hypothesizing a role in the infection path of X. fastidiosa similar to that played in P. aeruginosa.
The long latency period of X. fastidiosa for the perennial hosts, such as citrus and olive, makes arduous studying this pathogen in planta. Some authors demonstrated that tobacco can be used as model plant for X. fastidiosa subsp. pauca, although different varieties of this Solanacea show different level of infection and symptoms development (Pereira et al., 2017).
FIGURE 5 | Suggested overview of oxylipin pathway in XfCFBP8402. Black font represents oxylipins that were found in the samples and gray font represents oxylipins that were expected in the samples but not found. Continuous lines represent single step in production of the oxylipins, whereas dotted lines indicate the presence of multiple steps to achieve the compounds indicated in figure.
In our study, N. tabacum Petit Havana SR1 artificially infected with XfCFBP8402 strain is used to monitor the bacterial ability to multiply and spread into tobacco and to study the variation of target lipid entities into XfCFBP8402 infected plants compared to uninfected ones.
The tobacco Xf-infected plants show feeble symptoms after 15 DAI: that are, wrinkling of the lower older leaves along the leaf margin. This wrinkling develops across the leaf 's surface in the interveinal spaces resulting in leaf deformation and in leaf early yellowing (Supplementary Figure S11). The distribution of XfCFBP8402 is diverse within the analyzed tissues: more abundant into leaves (laminar part) than in the petioles (and central vein) of the samples. This scenario is almost clearly illustrated by the lipid profile of the two parts of the leaf. The lipid entities whose abundance varying dramatically upon bacterial infection may be ascribed to both plant and pathogen repertoire with some exception such as the bacterial ornithine lipid OL1 and the hopanoid BHT. Namely, plant change their lipid profile following pathogen attack (Siebers et al., 2016) as well as bacteria reshape their lipids while infecting the host (Torres et al., 2007;Nascimento et al., 2016;Siebers et al., 2016).
Complex lipids (e.g., DAG) as well as FFA (e.g., oleic acid) follow in the petioles and leaves similar trends with some notable exception. For instance, OL1 (OL 15:0-OH/19:0 cyclo) is more abundant in leaf margins; thus, we can here suggest that OL1 may act as Pathogen-Associated Molecular Patterns (PAMP) and trigger plant reactions in this part of the leaf where bacterial load is high (as indicated by qPCR; Vences-Guzman et al., 2013). TAG 52:2 (16:0/18:1/18:1) and PG 32:1 (16:1/16:0) may represent a source of diffusible factors (e.g., C16:1 and 18:1) in presence of lipase activity. X. fastidiosa subsp. fastidiosa strain Temecula1 (causal agent of the PD in grapevine), secretes the lipase LesA during the infection of the host. LesA accumulates, inversely with bacterial titer, in leaf margins and positively contributes to develop PD symptoms such as leaf scorching and chlorosis in grapevine. Notably, LesA decreases, in a gradient-shaped manner, toward the petiole where Xylella formed a biofilm network (Nascimento et al., 2016). According to our evidences, we can suggest that XfCFBP8402 uses lipase and the products of its activity as virulence factors. For instance, starting from PG 32:1 (16:1/16:0), a putative lipase could release C16:1 (XfDSF2) that is actually more abundant in the leaf than in the petiole. Further, Figure 4B shows that different FFA are accumulating in the Xfinfected plants and that the C14:1 (XfDSF1) is more abundant in the petioles than in the leaves (Supplementary Tables S4-S6) suggesting that the two DSF have a different role during the interaction with the host. We can argue that XfDSF1 is associated with the formation of biofilm that should occur within the petiole but not into the leaf margins.
The FFAs, as indicated elsewhere, provide substrates for the formation of oxylipins. Oxylipins are indicators of plant reactions during stress responses (Blèe, 2002;Christensen and Kolomiets, 2011;Scala et al., 2014;Siebers et al., 2016). The oxylipin profile -whose synthesis overview is reported in Figure 5 is completely different in the two leaf parts: enhanced within the petiole while depressed in the leaves. Notable exceptions (i.e., oxylipins enhanced into leaves) are represented by 7,10-diHOME and, the jasmonic acid precursor, 13-HpOTrE. We should here hypothesize that the first may be produced by XfCFBP8402 similarly to P. aeruginosa where this oxylipin appears playing a critical role in the initial stages of biofilm formation (Martinez and Campos-Gomez, 2016). 13-HpOTrE levels rise during tobacco infection; JAs, indeed, appear at steady state or feebly decreased. We can here speculate that XfCFBP8402 could interfere actively with the JA-pathway for avoiding plant resistance. Further studies are needed to investigate thoroughly the putative role of these molecules in the interaction with the host plants and finding the enzymes involved in their production.
It emerges a scenario in which the complex lipid profile of XfCFBP8402 -partly elucidated in this study -contributes in several ways to its lifestyle and to its relation with the host. In particular, we can here suggest that other DSFs than the already described ones (i.e., XfDSF1 and 2), as well as other lipid entities characterized in this study (e.g., oxylipins), are worth of further investigation in X. fastidiosa pathogenesis. Since the XfCFBP8402 ability to colonize tobacco appears related to the diffusion of lipid factors, it can be argued that lipid entities such as OL1, TAG 52:2, C18:1, and 7,10-diHOME may constitute an arsenal of molecules that actively contribute to plant-pathogen cross-talk.

AUTHOR CONTRIBUTIONS
VS performed the conception and design, data interpretation, coordination of contributes, and paper preparation and revision. MR performed the experiment management, data interpretation and elaboration, and paper preparation and revision. MS performed the data management and data acquisition. NP performed the plant infection experiment management and paper revision. VM performed the PCR data acquisition and interpretation. SLu performed the plant infection data acquisition. SLo performed the conception and design, data interpretation, and paper preparation and revision.

FUNDING
The work has been funded by the European Union Horizon 2020 research and innovation program under the grants agreement no. 727987 XF-ACTORS (Xylella fastidiosa Active Containment Through a multidisciplinary-Oriented Research Strategy the mass spectrometer QQQ 6420 Agilent Technologies was acquired by funding from Sapienza "medie attrezzature" -C26G14XYEX).