Surface Plasmon Resonance assay for label-free and selective detection of Xylella

: Xylella fastidiosa is among the most dangerous plant bacteria worldwide causing a variety of diseases, with huge economic impact on agriculture and environment. A surveillance tool, ensuring the highest possible sensitivity enabling the early detection of X. fastidiosa outbreaks, would be of paramount importance. So far, a variety of plant pathogen biomarkers were studied by means of Surface Plasmon Resonance (SPR). In this study, Multi Parameter-SPR (MP-SPR) has been used for the first time to develop a reliable and label-free detection method for X. fastidiosa . The real-time monitoring of the bio-affinity reactions is provided as well. Selectivity is guaranteed by bio-functionalizing the gold transducing interface with polyclonal antibodies for X. fastidiosa and it was assessed by means of a negative control experiment involving the non-binding Paraburkholderia phytofirmans bacterium strain PsJN. Limit of detection of 10 5 CFU/mL was achieved by transducing the direct interaction between the bacterium and its affinity antibody. The screening of X. fastidiosa plant pathogen is performed by means of Surface Plasmon Resonance. The study demonstrates the bacterium direct-assay, performed on a cm-sized gold surface modified with the anti- X. fastidiosa , covalently bound to the detecting interface. Unprecedent high selectivity is shown by assaying the Burkholderia phytofirmans non-binding bacterium. The affinity constant of the X. fastidiosa immunoassay is determined as well.


Introduction
The detection of pathogenic bacteria, virus or fungi aimed at the identification of infectious plant disease is one of the most critical tasks in the monitoring of agriculture relevant species worldwide. [1] A relevant pathogen which stirred the interest of researchers in different countries is the bacterium Xylella fastidiosa. It is in fact one of the most dangerous plant bacteria worldwide, causing a variety of diseases, such as Pierce's disease of grapevine, phony peach disease, plum leaf scald, citrus variegated chlorosis disease, and olive scorch disease, [2,3] as well as leaf scorch on almond and on shade trees. [4] X. fastidiosa has been isolated from more than 300 plant species all over the world, [5,6] although not all of these plants are susceptible to disease. [7] X. fastidiosa is currently present in several European countries, mainly in Italy, France and Spain, [8] where a number of areas are under eradication or containment strategies to avoid dissemination of this quarantine pest. [9] Eventually, international organizations, such as the European and Mediterranean Plant Protection Organization (EPPO), the International Plant Protection Convention of the Food and Agriculture Organization (IPPC-FAO) and the European Food Security Agency (EFSA), periodically release standards for diagnosis and detection of the pathogen, and update about the biology, epidemiology and control of the bacterium. [10,11] They also provide the updated hosts' list, due to the continuous increase of new host species, [12] as well as the guidelines for plant tissue sampling. [13] The transmission of the bacterium takes place through xylem-feeding vector insects that are widespread in the entire EU territory. [14] The spread of X. fastidiosa by insects does not require an incubation period in the vector, that can acquire the bacterium by feeding on the xylem fluid of an infected plant and transmit it to healthy plants, immediately after acquisition. All this calls for strict control measures, to be taken immediately after any new outbreak is detected. Therefore, in this perspective a powerful surveillance tool, ensuring the highest possible level of early detection of

Accepted Article
This article is protected by copyright. All rights reserved outbreaks of X. fastidiosa, would be of paramount importance. The sooner the infection is identified the more effective the intervention on diseased trees can be. Hence, the demand for sensors capable to detect plant pathogen with low detection limits and selectivity, is now increasing. [15] Several diagnostic protocols were tested for the detection of X. fastidiosa, such as enzyme-linked immunosorbent assay (ELISA), [16] polymerase chain reaction (PCR), [17,18] direct tissue blot immunoassay (DTBIA) [19] or loop-mediated isothermal amplification (LAMP). [20,21] The molecular amplification techniques currently available are very useful and play a key role in the preventive control and management of the disease. [10] Nevertheless, real-time PCR protocols involve a high cost per sample and require trained personnel. Moreover, cross-contamination among samples is a considerably important problem. [22] In this respect, highly selective and ultra-sensitive immunological techniques could be very useful, because of the reduced inhibitory effects from sample template contaminants as well as their suitability in large-scale testing. [23,24] Indeed, ELISA is a key-technology for large-monitoring programs due to easier management of large number of samples compared to genomic tests. Indeed ELISA is the workhorse in the programs carried out in Italy, where more than 150,000 plants were tested from October 2017 to April 2018, [25] However, the limit of detection of ELISA is at most of the order of 10 4 CFU/mL, while real-time PCR genomic detection goes down to about 10 2 CFU/mL. [26] However, these are all label-needing technologies. Namely, to measure the concentration of a given biomarker, a label (typically an enzyme) is needed to make the signal detectable. This increases the complexity of the overall assay procedures and adds steps and personnel cost to it, leading to a time-to-results of several hours. Therefore, the fast screening of plants that might be infected is difficult with such approaches. To this end reliable point-of-care approaches need to be developed. They usually require a very stable layer of capturing antibodies that requires full characterization. One of the most suitable approaches in this respect is Surface Plasmon Resonance (SPR), that is also a sensing technique on its own.

Accepted Article
This article is protected by copyright. All rights reserved A wide variety of plant pathogen biomarkers, were detected through SPR, with different degree of sensitivity depending on the assay configuration. [27][28][29] However, none of these studies involve X. fastidiosa. Among other techniques, SPR holds the advantage of being a label-free detection enabling also real-time monitoring of bio-affinity reactions. Indeed, the local variation of the surface refractive index can be directly correlated to the amount of bounded species, for which the interaction kinetics can be also determined. [30] Thus, the SPR platform is particularly useful not only as a detection method per se, but also for assessing the biofunctionalization protocol of gold surfaces with specific antibodies, validating their capturing efficacy against target analytes. [31] Hence it can be very useful to characterize an electrode that can be used for further development of biosensing and bioelectronic methods. In fact, among the novel approaches, organic bioelectronic devices are of interest in many fields of application from clinical diagnostics [32][33][34][35][36] to agrofood, [33,[37][38][39] and represent ideal candidates in point-of-care testing, thanks to low-cost, rapidity and portability.
The aim of the present study is to develop a Multi Parameter-SPR study of a reliable, and label-free detection method for X. fastidiosa achieving limit of detections comparable to the labelneeding ELISA gold standard. Relevantly, to the best of our knowledge, the direct assay of X. fastidiosa with an SPR apparatus is here proposed for the first time. The system is also endowed with high selectivity by successfully functionalizing a gold transducing surface with the X. fastidiosa affinity antibodies. The selectivity is assessed by measuring the response of the nonbinding Paraburkholderia phytofirmans bacterium strain PsJN, which is a Gram-negative rodshaped bacterium like X. fastidiosa. The zero response over a wide range of concentrations, prove the unprecedented selectivity of the assay towards X. fastidiosa. Moreover, the polyclonal antibodies and the X. fastidiosa bacteria affinity reaction binding constant, being 3.5•10 7 M -1 , has been also measured for the first time. Table 1. Differences in performance between Multi-parameter SPR and state-of-the-art platforms for detection of Xylella Fastidiosa.

SPR Sensor surface modification
One main requirement for a high-performing biosensors, and specifically immunoassay techniques, is the reproducibility of sensing surface, on which the bio-recognition elements are immobilized. An effective method largely used in antibody-based sensors is the immobilization of these capturing antibodies on functional monolayers, whose terminal groups can be tailored for the required binding. [40] In the present work, gold sensor surfaces were modified by a self-assemblies of alkylthiols with different chain length, prior to the amine coupling of polyclonal antibodies for the recognition of X. fastidiosa (anti-XF). An Au coated (~ 50 nm) SPR slides (BioNavis Ltd) comprising a chromium adhesion layer (~2 nm) served as semi-transparent SPR substrate. To this aim a mixed self-assembled monolayer (SAMs) of 11-mercaptoundecanoic acid (11-MUA) and 3mercaptopropionic acid (3-MPA), in a molar ratio of 1:10, was used as it was previously demonstrated to be particularly suited to achieve a very dense grafting of the capturing antibodies.

Accepted Article
This article is protected by copyright. All rights reserved The protocol is schematically depicted in Figure 1. Indeed a mixture of SAMs with different chain length is known to be preferable for large biomolecules effective immobilization, since biomolecules are able to attach to the surface without undergoing conformational changes due to steric hindrance and therefore an enhanced surface coverage of functional proteins can be obtained. [41][42][43] More in details the protocol involved the activation of carboxyl terminal groups of [44] in a 2-(Nmorpholino)ethane-sulfonic acid (MES) buffer at controlled pH 4.8 ( Figure 1A-B). Then, the surface was rinsed with phosphate buffer solution (PBS) at pH 7.4 and further exposed to the anti-XF buffer solution at a concentration of 10 µg/mL ( Figure 1C). Besides a solution of ethanolamine 1 M was used to saturate possible unreacted esters on the SAM. The scheme of the final modified sensor surface is reported in Figure 1D.

Accepted Article
This article is protected by copyright. All rights reserved The SPR apparatus allowed the real-time monitoring of each step of the binding process, by measuring the angular variation of the plasmon peak minimum vs time, as shown in the sensogram spectrum reported in Figure 2A. The functionalization was performed by manual injections of each reagent, in a 100 µL single channel cell. The gold exposed area of about 0.4 cm 2 , was sampled simultaneously into two different points with two laser beams at wavelength of 670 nm. The two traces, shown as red and blue curves in Figure 2, give an idea of the high homogeneity of the layer.
Once the sensor was placed in the sample-holder, the surface was rinsed with water and MES until a stable trace was measured, that was taken as the baseline. Then, the EDC/NHSS solution in MES was injected and kept in contact with the sensor for 15 minutes. The variation of surface refractive index produces the SPR angle change observed in the sensogram. Then, the PBS solution was injected, and a new stable baseline was acquired before injecting the PBS solution of the anti-XF to be bound on the SAM. Figure 2B shows the anti-XF grafting that is completed in 3.5 hours, when the trace levels-off. After PBS rinsing, the ethanolamine (EA) solution was injected and kept in contact with the sensing area for 45 minutes. Finally, PBS was injected to rinse the surface and remove not covalently bound species. The angle shift (∆θSPR) recorded starting from the initial PBS baseline to the final rinsing after EA represents the amount of effectively bound antibodies. [45]  The elicited protocol lead to a very dense layer of the anti-XF with a minimal consumption of regents thanks to the optimization of the method previously performed. [46,47] Indeed, by employing the MES buffer at pH 4.8, more efficient reactions between the EDC/NHSS and the carboxyl terminal groups can be obtained. [45,48] Then, for the reaction with the activated esters (NSS) to occur, the antibody needs to be brought in close proximity with the surface, promoting the electrostatic attraction with the oppositely charged remaining surface carboxyls. [45] Two factors should be considered in the antibody biding reaction: the charge on its amino groups could be kept positive if the pH is 0.5-1 units below the isoelectric point (pI) of the antibody; and the negative charge of the carboxyl groups on the sensor surface could be preserved by a surface pH above 4.
Hence, a two-step immobilization protocol has been chosen to fit the best experimental condition, in which the MES buffer is replaced by PBS at pH 7.4 before injecting the solution of antibodies (pI ̴ 7/8). [49] Indeed, by using 100 µL of a anti-XF 10 µg/mL solution in PBS, an average angle shift Δθ = (0.16 ± 0.04) deg was recorded for 10 replicate measurements. This SPR angle variation was used to calculate the surface coverage of the bio-recognition element by means of the de Feijter's Equation 1: The surface coverage (Γ, expressed in ng/cm 2 ) is proportional to the thickness of the deposited ligand (d), to the difference in refractive index of the adlayer (n) and the bulk medium (n0) and to the refractive index increment (dn/dC). Moreover, the instrument response is correlated to the refractive index variation by the equation: where k is the wavelength dependent sensitivity coefficient, and ΔθSPR is the measured angular shift.
For a thin layer (<100 nm) and at a λ = 670 nm, the ratio dn/dC is approximated to 0.182 cm 3 /g,

Accepted Article
This article is protected by copyright. All rights reserved whereas k•d ≈ 1.0•10 −7 cm/deg; [50] thus by introducing Equation 2 into Equation 1, the surface coverage can be estimated as: In Table 2 the experimental angle shift measured after anti-XF binding and after the saturation with EA, are given along with the surface coverages estimated from Equation 3. The antibody surface coverage is as high as 89 ± 21 ng/cm 2 , corresponding to (3.6 ± 0.8) •10 11 molecules/cm 2 . This represent a homogeneous layer of densely packed biorecognition elements, which provides a highly packed transducing surface for the interaction with the target species. [47]  The response registered for the antibody coverage in this study was comparable with those reported in the literature, with equivalent working conditions, for which an angular response of ∆θSPR ≈ 0.18 deg was recorded. [51,52] The error bars have been estimated as the relative standard deviation of the surface coverages on two different replicates and four different sampled areas, to provide an estimation of the homogeneity of the biolayer uniformity. Indeed, SPR sensograms were acquired from four different points on two different SPR slides.

SPR assay of Xylella fastidiosa
To the best of our knowledge, the direct assay of X. fastidiosa with SPR has been performed for the first time. Indeed, the SPR direct-assay configuration for large cell detection, such as

Accepted Article
This article is protected by copyright. All rights reserved bacteria, has been very seldom investigated, due to its limited sensitivity of direct cell-capture. [51] In fact, one of the major limiting factor in SPR assay is represented by the fluid forces that have to be overcome before particles can reach the sensor surface where they are captured. [52] Once cells have found their target, the antibody-cell-binding affinity must be able to withstand the effect of those forces, generated by the laminar flow of the SPR apparatus. Remarkably, in the present work SPR configuration has been optimized to overcome this main limitation. Indeed, the X. fastidiosa directassay has been performed operating the SPR sensor in static conditions, preventing the effect of any fluid forces. Moreover, the optimized SPR cell is endowed with a wide field sensing area. Therefore, the high density of anchored antibodies on the cm 2 -area allowed to enhance the capturing efficiency of the detecting interface toward the target bacteria. The biofunctionalized SPR sensor slide was tested against the specific binding of the X. fastidiosa in a range of concentration from 10 5 CFU/mL to 2•10 8 CFU/mL. The relevant sensogram is reported in Figure 3.
The assay was carried out by injecting X. fastidiosa solutions in PBS at different concentrations. Each solution was let to interact with the anti-XF functionalized interface for 40 minutes (Figure 3A), minimum time required to reach the equilibrium between bounded species at the surface and unbounded bacteria in solution. Upon equilibrium, the bacterium excess was removed by rinsing with the PBS buffer solution. The signal after each rinsing was compared to the initial baseline, acquired in PBS buffer solution, taken as the zero-level signal in the sensogram.
Also in this case, the exposed sensing area was sampled in two different points (blue and red signal in Figure 3A). The SPR data reported as red squares in Figure 3B are the averages over three replicates, for which a relative standard deviation, RSD, of 1% was computed. Relevantly all the analysis were performed at controlled temperature, due to the dependency of the refractive index from temperature changes. [43] Indeed, the MP-SPR used in this study is equipped with a thermostatic apparatus that was set at room temperature, to avoid any discrepancy between the temperature value of SPR sensor surface and the solution injected.

Figure 3. A) Exposure of the anti-XF functionalized surface to X. fastidiosa at increasing concentrations; black arrows correspond to sample injections and purple dotted line refers to the buffer level. Red and blue lines refer to two sampling points measured simultaneously. B) Comparison of X. fastidiosa (red square) and Burkholderia phytofirmans (black circle) SPR responses against the anti-XF functionalized surface. The average signal and standard deviation for four replicates analysis is reported.
The selectivity of the assay was evaluated by exposing the sensing interface to a nonbinding species. To this end, the interaction of the anti-XF functionalized surface with the Paraburkholderia phytofirmans (P. phytofirmans) strain PsJN, a gram-negative bacterium was assayed. The SPR angle shift vs. the P. phytofirmans concentration is reported in Figure 3B as Accepted Article black circle. The selectivity of the biosensing platform has been successfully demonstrated, as the negative control experiment involving the P. phytofirmans showed a maximum angle shift below 0.01 deg, being only 4% of the signal registered for the X. fastidiosa assay. Therefore, although some spurious non-specific adsorption could not be ruled out considering the aggregating nature of these bacteria, [6] the major contribution to the SPR signal can be ascribed to the binding between X. fastidiosa and its specific antibodies. Thus, the selectivity of the assay was estimated as the ratio between the angle-shift measured for P. phytofirmans and X. fastidiosa respectively, resulting in a value as low as ∆θBPhy / ∆θXF = 0.05 ± 0.01. This is among the lowest measured values as the closest best is as high as 0.23. [26,53,54] The limit of detection (LOD) of the assay was evaluated considering the linear portion of the calibration curve given in Figure 3. To this aim the linear regression on the data of angle shift vs.
analyte concentration was reported in Figure S1. The fit of the linear portion of the curve (highlighted range 10 5 -10 7 CFU/mL) is given as a red dotted line in Figure S1. In the same figure the black circles are the data coming from the negative control experiment involving the P. phytofirmans bacterium assay, and the black dotted line is its average signal (sC). The LOD was evaluated as the average signal of the negative control experiment (sC) plus three times its standard deviation (σC). This level is hence = + 3 = 3.18 • 10 −3 . The comparison of this level with the interpolating linear regression results in a LOD of (3.3 ± 0.2) •10 5 CFU/mL. This well compares with the detection limit found with SPR for different bacteria species in similar test conditions, as well as with ELISA gold standard. [51,55] Indeed, the signal-to-noise ratio of the SPR apparatus is not sufficiently high to detect bacteria concentrations below 10 4 CFU/mL. [56][57][58] On the other hand, the maximum bacteria concentration assayed was 2•10 8 CFU/mL, corresponding to the concentration needed to fully cover the biofunctionalized sensing surface. The latter has been estimated considering the number of X. fastidiosa covering the 0.4 cm 2 wide gold area, evaluating

Accepted Article
This article is protected by copyright. All rights reserved the bacteria footprint as rod-like shaped capsules, with radius of 0.25 µm and an end-to-end length of 1.90 µm. [2,53,54] To complete the characterization, the binding affinity equilibrium constants were also evaluated with SPR using a non-regenerative approach. [60] Indeed, surface regeneration between consecutive analyte injections to remove the bound analytes cannot be used in case of stable ligandanalyte complexes, such as antibodies-bacteria binding interactions, where the dissociation bindingpairs rate (koff) falls in the range of 10 -4 s -1 . [61] In fact, with such systems the regeneration may fail in removing all the bound analyte molecules; as a result, the SPR signal would not be close enough to the baseline after the regeneration. Moreover, harsh regeneration reagents may destroy the bioactivity of the ligand molecules. Consequently, a non-regenerative approach has been herein proposed to investigate the interaction kinetics between the X. fastidiosa and its affinity antibody.
To this aim, a clean gold sensor surface was allocated in the SPR apparatus and exposed to a 2•10 8 CFU/mL concentration of X. fastidiosa for 50 min to deposit a layer of the bacterium. As shown in  Figure S3, the

Accepted Article
This article is protected by copyright. All rights reserved values for the maximum SPR response and the affinity constant (KA) of the species can be calculated from the intercept and slope of the linear fit. The resulting KA value obtained is 3.5•10 7 M -1 , which is comparable with the affinity constant value found in works for bacteria detection by antibodies, with comparable testing conditions. [61,62]

Conclusions
The SPR platform has been herein proposed for the first time as a label-free, fast and reliable detection method for X. fastidiosa, achieving limit of detection comparable to the labelneeding ELISA gold standard (10 5 CFU/mL). Moreover, SPR has been proposed as powerful tool to assess the optimized biofunctionalization protocol of gold surfaces with anti-XF, validating their capturing efficacy against X. fastidiosa. The selectivity of the sensor surface has been assessed as well, by comparing the cross-reactivity of the P. phytofirmans non-binding bacterium strain PsJn, showing unprecedented performance in both sensor stability and selectivity. Moreover, the binding affinity between anti-XF and the X. fastidiosa bacteria has been also tested in the SPR apparatus, obtaining an affinity constant value comparable with those reported in literature for bacteria detection with antibody/antigen assays. Remarkably, the SPR platform hereby presented paves the way for further development of a wide-field bioelectronic sensor, to accomplish an efficient presymptomatic diagnosis of diseases caused by X. fastidiosa. In fact, diagnostic tools for early detection of X. fastidiosa are urgently required. Indeed, such goal relies on the capability to obtain stable biofunctionalized gold transducing interface with purified Immunoglobulin (IgG) selective for X. fastidiosa. As a future perspective, the reliable and cost-effective biofunctionalization protocol herein developed will be employed to immobilize trillions of anti-XF antibodies on the gold gate electrode of an Electrolyte gated field effect transistor, to achieve ultra-sensitive and selective detection of X. fastidiosa. 10-day-old colonies of X. fastidiosa, subsp. pauca De Donno strain, sequence-type ST53 and 2-dayold colonies of Paraburkholderia phytofirmans PsJN were scraped from plates and dispersed in sterile potassium phosphate buffer (0.05 M, pH 7.2) to prepare ten-fold serial dilutions. Bacterial cells concentrated from 10 8 CFU/mL to 10 4 CFU/mL were tested for both bacteria. The effective bacteria concentrations, expressed in CFU/mL, were determined using plate count as the reference
Purified-IgG against X. fastidiosa (analytical specificity validated by EPPO standard) at concentration of 1 mg/mL were provided by Agritest an diluted to 10 µg/mL.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.