A Novel Sensitive Cell-Based Immunoenzymatic Assay for Palytoxin Quantitation in Mussels

The marine algal toxin palytoxin (PLTX) and its analogues are some of the most toxic marine compounds. Their accumulation in edible marine organisms and entrance into the food chain represent their main concerns for human health. Indeed, several fatal human poisonings attributed to these compounds have been recorded in tropical and subtropical areas. Due to the increasing occurrence of PLTX in temperate areas such as the Mediterranean Sea, the European Food Safety Authority (EFSA) has suggested a maximum limit of 30 µg PLTX/kg in shellfish meat, and has recommended the development of rapid, specific, and sensitive methods for detection and quantitation of PLTX in seafood. Thus, a novel, sensitive cell-based ELISA was developed and characterized for PLTX quantitation in mussels. The estimated limits of detection (LOD) and quantitation (LOQ) were 1.2 × 10−11 M (32.2 pg/mL) and 2.8 × 10−11 M (75.0 pg/mL), respectively, with good accuracy (bias = 2.5%) and repeatability (15% and 9% interday and intraday relative standard deviation of repeatability (RSDr), respectively). Minimal interference of 80% aqueous methanol extract allows PLTX quantitation in mussels at concentrations lower than the maximum limit suggested by EFSA, with an LOQ of 9.1 µg PLTX equivalent/kg mussel meat. Given its high sensitivity and specificity, the cell-based ELISA should be considered a suitable method for PLTX quantitation.


Introduction
Palytoxin (PLTX), a complex marine poly-ol toxin, is one of the most toxic natural compounds. The discovery of PLTX dates back to the 1960s, when, in a tide pool of Hana Bay (Maui Island, Hawaii), Prof. Paul Helfrich collected samples of a toxic soft coral, subsequently identified as Palythoa toxica. Ten years later, the chemical structure of PLTX isolated from this coral was reported [1]. Later, PLTX and a series of its analogues were also identified in other Zoantharia belonging to the genera Palythoa [2][3][4][5][6][7] and Zoanthus [8], in benthic dinoflagellates of the genus Ostreopsis [9][10][11][12][13][14][15], and in cyanobacteria of the genus Trichodesmium [16]. Only a few of these analogues have been studied from a biological and chemical point of view, including (i) 42-hydroxy-PLTX (42S-OH-50S-PLTX), isolated from P. toxica [2,17], and its stereoisomer (42S-OH-50R-PLTX), isolated from P. tuberculosa [3]; (ii) ostreocin-D (OST-D) and PLTX binding on cultured cells was evaluated using a panel of different cell lines. Cells were exposed to PLTX for 10 min at 37 • C and the toxin binding was subsequently evaluated as described in the Materials and Methods section. The obtained results for each cell model were normalized on the protein content of each sample. Figure 1 shows the saturation curves of PLTX binding for each cell model (panel A). From these curves, Kd values and maximal binding were calculated, and their distribution was analyzed in the box plot of Figure 1  PLTX binding on cultured cells was evaluated using a panel of different cell lines. Cells were exposed to PLTX for 10 min at 37 °C and the toxin binding was subsequently evaluated as described in the Materials and Methods section. The obtained results for each cell model were normalized on the protein content of each sample. Figure 1 shows the saturation curves of PLTX binding for each cell model (panel A). From these curves, Kd values and maximal binding were calculated, and their distribution was analyzed in the box plot of Figure 1 (panels B and C). A median Kd of 8.1 × 10 -10 M (interquartile range = 2.2 × 10 -10 to 2.4 × 10 -9 M) and a median maximal binding of 0.015 (interquartile range = 0.0095 to 0.02738) were calculated. Binding parameters varied between the different cell lines, and HaCaT cells were the most sensitive cell line, as confirmed by the Kd values and maximal binding (1.4 × 10 -10 M and 0.043, respectively). On the contrary, the less sensitive cell models for PLTX binding were HepG2 cells (Kd = 6.5 × 10 -9 M; maximal binding = 0.009) and MCF-7 cells (Kd = not detectable; maximal binding = 0.003). For these reasons, the HaCaT cell line was chosen as the most sensitive model to set up the cell-based ELISA.

Incubation Temperature and Fixing Agents
To improve sensitivity, the assay was carried out exposing HaCaT cells to PLTX, varying the fixing agents and incubation temperature (37-60 °C) with the primary and secondary antibodies. An increased signal (optical density, OD) was observed with increased incubation temperature up to 50 °C, which subsequently decreased at higher temperatures ( Figure 2). This trend was recorded also varying the following cell fixing agents: 4% paraformaldehyde (PFA), 4% PFA and 1% glutaraldehyde, neutral-buffered formalin (NBF) (Figure 2).
At the optimal temperature of 50 °C, the highest colorimetric reaction signal was recorded using 4% PFA as fixing agent, followed by NBF (significant differences starting from 1.1 × 10 −9 M PLTX as compared to the data recorded using 4% PFA, p < 0.05) and 4% PFA + 1% glutaraldehyde (significant

Incubation Temperature and Fixing Agents
To improve sensitivity, the assay was carried out exposing HaCaT cells to PLTX, varying the fixing agents and incubation temperature (37-60 • C) with the primary and secondary antibodies. An increased signal (optical density, OD) was observed with increased incubation temperature up to 50 • C, which subsequently decreased at higher temperatures ( Figure 2). This trend was recorded also varying the following cell fixing agents: 4% paraformaldehyde (PFA), 4% PFA and 1% glutaraldehyde, neutral-buffered formalin (NBF) (Figure 2). At the optimal temperature of 50 • C, the highest colorimetric reaction signal was recorded using 4% PFA as fixing agent, followed by NBF (significant differences starting from 1.1 × 10 −9 M PLTX as compared to the data recorded using 4% PFA, p < 0.05) and 4% PFA + 1% glutaraldehyde (significant differences starting from 1.2 × 10 −10 M PLTX as compared to the data recorded using 4% PFA, p < 0.01) ( Figure 2D). Thus, 4% PFA and 50 • C were chosen as the optimal fixing agent and incubation temperature with antibodies.
Toxins 2018, 10, x FOR PEER REVIEW 4 of 15 differences starting from 1.2 × 10 −10 M PLTX as compared to the data recorded using 4% PFA, p < 0.01) ( Figure 2D). Thus, 4% PFA and 50 °C were chosen as the optimal fixing agent and incubation temperature with antibodies.

Sequence of Cell Fixation and Exposure to PLTX
The possibility of changing the sequence of cell fixation and exposure to PLTX was also evaluated: the cell-based ELISA was carried out fixing HaCaT cells with 4% PFA for 30 min before exposure to PLTX or exposing the cells to the toxin before fixation, as described in the Materials and Methods section. Figure 3A shows the concentration-dependent curve for PLTX detection recorded in the two conditions; the optimal condition consists in cell exposure to PLTX before fixation, since the inverted sequence dramatically decreased the OD values, as expected.

Blocking Agent
The influence of different blocking agents on the assay signal was also evaluated. Figure 3B shows a PLTX calibration curve obtained using three blocking agents. The highest signal was recorded using a Tris-borate buffer (TBB) solution containing 10% horse serum (HS) as a blocking agent. A significant decrease in signal was recorded starting from 1.2 × 10 −10 M PLTX using a Trisbuffer saline (TBS) solution containing 0.2% Tween 20 and 1% or 2% dried milk powder.

Primary Antibody Dilution
To further increase the signal/background ratio, the influence of different dilutions of the primary antibody on the assay signal was evaluated. As shown in Figure 3C, no significant differences were observed among three primary antibody dilutions tested (1:750, 1:1500, and 1:3000).

Sequence of Cell Fixation and Exposure to PLTX
The possibility of changing the sequence of cell fixation and exposure to PLTX was also evaluated: the cell-based ELISA was carried out fixing HaCaT cells with 4% PFA for 30 min before exposure to PLTX or exposing the cells to the toxin before fixation, as described in the Materials and Methods section. Figure 3A shows the concentration-dependent curve for PLTX detection recorded in the two conditions; the optimal condition consists in cell exposure to PLTX before fixation, since the inverted sequence dramatically decreased the OD values, as expected.

Blocking Agent
The influence of different blocking agents on the assay signal was also evaluated. Figure 3B shows a PLTX calibration curve obtained using three blocking agents. The highest signal was recorded using a Tris-borate buffer (TBB) solution containing 10% horse serum (HS) as a blocking agent. A significant decrease in signal was recorded starting from 1.2 × 10 −10 M PLTX using a Tris-buffer saline (TBS) solution containing 0.2% Tween 20 and 1% or 2% dried milk powder.

Primary Antibody Dilution
To further increase the signal/background ratio, the influence of different dilutions of the primary antibody on the assay signal was evaluated. As shown in Figure 3C, no significant differences were observed among three primary antibody dilutions tested (1:750, 1:1500, and 1:3000). Thus, the cell-based ELISA was subsequently carried out using the highest dilution (1:3000; 0.5 µg/mL final concentration) of the primary antibody. Thus, the cell-based ELISA was subsequently carried out using the highest dilution (1:3000; 0.5 µg/mL final concentration) of the primary antibody.

Characterization of the Cell-Based ELISA
The cell-based ELISA was subsequently characterized by evaluating the limit of detection (LOD)

Characterization of the Cell-Based ELISA
The cell-based ELISA was subsequently characterized by evaluating the limit of detection (LOD) and quantitation (LOQ) of PLTX, as well as the accuracy, repeatability, and specificity of the assay. The calibration curve of PLTX is shown in Figure 4A: the working range is 1.4 × 10 −11 to 1.1 × 10 −9 M, whereas the estimated LOD and LOQ of PLTX were 1.2 × 10 −11 M (32.2 pg/mL) and 2.8 × 10 −11 M (75.0 pg/mL), respectively. The working range was analyzed by linear regression, plotting the theoretical PLTX concentrations used in the assay against the measured toxin concentrations, and the obtained results revealed a good correlation coefficient (r 2 = 0.9894; n = 10) ( Figure 4B). A mean bias value of 2.5% (range: −5.1 to 9.8%) was obtained ( Table 1).
The intraassay repeatability was estimated over six replicates carried out in one day, while the interassay repeatability was evaluated over 10 replicates carried out over a six-month period. Good correlation coefficients were calculated, with r 2 = 0.9770 for intraassay and r 2 = 0.9985 for interassay ( Figure 5). Moreover, the intraday and interday repeatability coefficients (relative standard deviation of repeatability, RSDr) were 12% and 15%, respectively (Table 1).
The calibration curve of PLTX is shown in Figure 4A: the working range is 1.4 × 10 −11 to 1.1 × 10 −9 M, whereas the estimated LOD and LOQ of PLTX were 1.2 × 10 −11 M (32.2 pg/mL) and 2.8 × 10 −11 M (75.0 pg/mL), respectively. The working range was analyzed by linear regression, plotting the theoretical PLTX concentrations used in the assay against the measured toxin concentrations, and the obtained results revealed a good correlation coefficient (r 2 = 0.9894; n =10) ( Figure 4B). A mean bias value of 2.5% (range: −5.1 to 9.8%) was obtained ( Table 1). The intraassay repeatability was estimated over six replicates carried out in one day, while the interassay repeatability was evaluated over 10 replicates carried out over a six-month period. Good correlation coefficients were calculated, with r 2 = 0.9770 for intraassay and r 2 = 0.9985 for interassay ( Figure 5). Moreover, the intraday and interday repeatability coefficients (relative standard deviation of repeatability, RSDr) were 12% and 15%, respectively (Table 1).  Table 1. Bias values (%) for PLTX analysis by the cell-based ELISA, intraday (n = 6, 1 day) and interday (n = 10, 6 months) repeatability (relative standard deviation of repeatability, RSDr %), and mean of PLTX concentrations measured by the assay.

Inhibition of PLTX Binding by Ouabain
Ouabain (OUA) is a known inhibitor of PLTX in vitro effects due to its binding to the common molecular target, Na + /K + ATPase [54][55][56]. Thus, to confirm the specific detection of PLTX by the cell-based ELISA, the assay was performed also exposing the cells to 1 mM OUA for 10 min before their exposure to PLTX (1.4 × 10 −11 to 1.1 × 10 −9 M). As can be seen in Figure 6, a significant reduction of the assay signal was recorded when cells were pre-exposed to 1 mM OUA.

Inhibition of PLTX Binding by Ouabain
Ouabain (OUA) is a known inhibitor of PLTX in vitro effects due to its binding to the common molecular target, Na + /K + ATPase [54][55][56]. Thus, to confirm the specific detection of PLTX by the cellbased ELISA, the assay was performed also exposing the cells to 1 mM OUA for 10 min before their exposure to PLTX (1.4 × 10 −11 to 1.1 × 10 −9 M). As can be seen in Figure 6, a significant reduction of the assay signal was recorded when cells were pre-exposed to 1 mM OUA.

Mussel Matrix Effect
The suitability of the cell-based ELISA for quantitation of PLTX in mussels was evaluated by assessing matrix interference on the assay sensitivity. Dilutions (1:2, 1:5, 1:10, 1:50) of the PLTX-free mussel extract in PBS were spiked with known PLTX amounts. Each extract was then analyzed by the cell-based ELISA and the OD values were compared to those obtained by analyzing the same PLTX concentrations without matrix. The minimum extract dilution that did not interfere with the assay was 1:10 (Supplementary Figure S2). At the same dilution, the solvent alone did not interfere with the assay (data not shown). The linear regression analysis between the results obtained by analyzing the 10-fold diluted mussel extract spiked with PLTX and those obtained by analyzing the Figure 6. Effect of cell pre-exposure to ouabain on PLTX detection by the cell-based ELISA. Each point represents mean ± SE of three different experiments. Statistical differences: *** p < 0.001 as compared to PLTX (two-way ANOVA and Bonferroni post test).

Mussel Matrix Effect
The suitability of the cell-based ELISA for quantitation of PLTX in mussels was evaluated by assessing matrix interference on the assay sensitivity. Dilutions (1:2, 1:5, 1:10, 1:50) of the PLTX-free mussel extract in PBS were spiked with known PLTX amounts. Each extract was then analyzed by the cell-based ELISA and the OD values were compared to those obtained by analyzing the same PLTX concentrations without matrix. The minimum extract dilution that did not interfere with the assay was 1:10 (Supplementary Figure S2). At the same dilution, the solvent alone did not interfere with the assay (data not shown). The linear regression analysis between the results obtained by analyzing the 10-fold diluted mussel extract spiked with PLTX and those obtained by analyzing the same PLTX concentrations without matrix (Figure 7) revealed a good correlation coefficient (r 2 = 0.9734) and an excellent mean bias value (mean bias = −1.4%; Table 2). In addition, the estimated LOQ for PLTX in mussels was 3.5 × 10 −11 M, equal to 9.3 µg/kg mussel meat.

Recovery of PLTX from Mussels
Recovery experiments were carried out to determine the efficiency of extraction and quantitation of PLTX from mussel meat. Aliquots of PLTX-free mussel homogenate were spiked with PLTX (1.4 × 10 −11 to 1.1 × 10 −9 M) and subsequently extracted as described above to obtain extracts containing 0.1 g meat equivalent/mL. Extracts were diluted 1:10 and analyzed by the cell-based ELISA. Recovery of PLTX from mussels ranged between 94% and 110% (coefficient of variability: 13-17%; Table 3).   Table 2. Bias values (%) for PLTX detected in 80% aqueous methanol mussel extract spiked with the toxin after 1:10 dilution compared to theoretical PLTX concentrations (n = 10).

Recovery of PLTX from Mussels
Recovery experiments were carried out to determine the efficiency of extraction and quantitation of PLTX from mussel meat. Aliquots of PLTX-free mussel homogenate were spiked with PLTX (1.4 × 10 −11 to 1.1 × 10 −9 M) and subsequently extracted as described above to obtain extracts containing 0.1 g meat equivalent/mL. Extracts were diluted 1:10 and analyzed by the cell-based ELISA. Recovery of PLTX from mussels ranged between 94% and 110% (coefficient of variability: 13-17%; Table 3). Table 3. Recovery of PLTX from mussels analyzed by the cell-based ELISA.

Discussion
In recent years, Ostreopsis cf. ovata has bloomed with increasing frequency in temperate areas such as the Mediterranean Sea and along the Atlantic coast of Portugal [31,34]. Peculiar climate changes and marine conditions, high availability of nutrients, and hydrographic conditions characterized by low wave energy may play important roles in this phenomenon [57]. Concomitant with Ostreopsis blooms, PLTXs have been detected in microalgae, aerosolized seawater, and edible marine organisms [30,58,59]. Despite no foodborne poisonings being attributed to PLTX and analogues in this area to date, these toxins can represent a significant public health concern. Therefore, there is a need to develop new detection methods able to quantify PLTXs in seafood at concentrations lower than the maximum limit suggested by EFSA (30 µg PLTX/kg of shellfish meat) [35]. Over the years, various detection methods have been developed, some with limited sensitivity or other limitations, such as significant matrix effects. We developed a cell-based immunoenzymatic assay (cell-based ELISA) that combines the high binding potency of PLTX with the sensitivity and specificity provided by anti-PLTX antibody detection and evaluated it for its ability to accurately quantify PLTX in mussels.
Initially, the sensitivity of different cell lines to PLTX binding was evaluated to select the most suitable cell line. To this aim, different cell lines derived from colon cancer (LoVo, Caco2, HCT-116), pancreatic cancer (PANC-1), hepatic cancer (HepG2), and breast cancer (MCF-7, MDA-MB-231) were considered. In addition, nontumor cell lines were used, such as the HaCaT skin keratinocyte cell line and the immortalized human hepatic (IHH) cell line. A wide distribution of binding parameters (Kd and maximal binding) was observed, suggesting a wide sensitivity range among the different cell lines. Consequently, the HaCaT cell line turned out to be the most sensitive to PLTX binding, corroborating literature data showing that HaCaT keratinocytes are among the cell models most sensitive to the toxin [60]. For these reasons, this cell line was chosen as the most suitable model to develop the cell-based ELISA.
To optimize the method, the influence of the incubation temperature of the antibodies on the assay signal as well as the effects of different cell fixing agents were evaluated. The highest signal was obtained using 4% PFA as a fixative solution and incubating the primary and secondary antibodies at 50 • C. The decreased signal recorded after increasing the incubation temperature of the mAb is probably due to a thermal denaturation of the antibodies [61]. For the fixation phase, we chose formaldehyde, a commonly used fixative that easily diffuses into cells, allowing optimal fixation [62].
Once the final protocol of the assay was defined, an intralaboratory validation demonstrated good sensitivity, accuracy, and repeatability. As expected, with the combination of two sensitive features, PLTX binding to Na + /K + ATPase of HaCaT cells and its detection by monoclonal antibodies, the sensitivity of the cell-based ELISA was higher than that of other PLTX immunoassays [46,48]. In addition, this assay was almost 35 time more sensitive than the indirect sandwich ELISA (estimated LOD = 1.1 ng/mL) developed by our group [47]. With respect to the latter, a significant improvement was achieved in terms of not only sensitivity but also assay time (less than 3 h compared to about 6 h for the sandwich ELISA). Moreover, the sensitivity of the cell-based ELISA, expressed as LOD value (estimated at 1.2 × 10 −11 M, 32.2 pg/mL) was more than 400 times higher than that of LC-HRMS [36]. Similarly, the estimated LOD was more than 10-fold more sensitive than an SPR-based immunoassay [50] and about twofold more sensitive than the carbon nanotube-based biosensor (70 pg/mL) developed by Zamolo et al. [51]. On the other hand, the sensitivity of the cell-based ELISA was lower than that of the immunoassay (0.5 pg/mL) developed by Garet et al. [49], who employed a single-chain antibody isolated by phage display technology. However, cross-reactivity with other toxins was not investigated and, as discussed later, the method was affected by variable toxin recovery from mussels, hindering its use for PLTX quantitation in this matrix.
The specificity of the cell-based ELISA for PLTX was demonstrated by the lack of cross-reactivity toward other marine algal toxins contaminating seafood [63] and potentially co-occurring with PLTX in edible marine organisms. Moreover, the interference of mussel extract was minimal, as demonstrated by the 1:10 extract dilution that did not interfere with the assay. The 80% aqueous methanol was chosen as an extraction solvent based upon previous studies demonstrating that it is the most suitable for PLTX extraction as assessed by LC-HRMS [64] and the sandwich ELISA [47]. The estimated LOQ of the cell-based ELISA for PLTX in mussels was 9.3 µg PLTX/kg mussel meat, about three times lower than the safety limit suggested by EFSA (30 µg PLTXs/kg of shellfish meat). This result, together with the excellent toxin recovery from mussels (94-110%), suggests the ability of this novel cell-based ELISA to quantify PLTX in mussels with good precision. This is further supported by the good recovery recorded also at very low concentration (1.37 × 10 −11 M), even lower than that of the estimated LOQ, strengthening the ability of the cell-based ELISA to accurately quantify the toxin in mussels at very low concentrations.
On the contrary, despite its higher sensitivity, the immunoassay of Garet and colleagues [49] suffers from variable toxin recovery from mussels (64-113%), impairing its suitability for PLTX detection in this matrix. Similarly, significant interference by mussel matrix was observed for other methods, such as the hemolytic assay (LOQ = 640 µg PLTX equivalent/kg mussel meat) [42] and the lactate dehydrogenase-based hemolytic biosensor [43], for which a 1:50 dilution of the mussel extract was necessary. Furthermore, the sensitivity of the cell-based ELISA to quantify PLTX in mussels is higher than that of other antibody-based assays, such as the flow cytometry-based immunoassay (LOQ from 374 to 4430 µg/kg), which required at least a 30-fold extract dilution to avoid matrix effect [50]. In addition, the estimated LOQ of the cell-based ELISA for PLTX in mussels was close to that of other detection methods, such as the sandwich ELISA (11 µg/kg meat) [47], the carbon nanotube-based biosensor (2.2 µg/kg meat) [51], and LC-HRMS (15 µg/kg meat) [36]. However, it must be considered that the cell-based ELISA is a simple, fast, and inexpensive method. Moreover, due to its high sensitivity in buffer solution, it could be further exploited for its suitability to detect and quantify PLTX in other matrices.

Conclusions
In conclusion, the combination of PLTX binding to Na + /K + ATPase with PLTX detection by specific antibodies allowed the development of a novel and sensitive cell-based immunoenzymatic assay for PLTX detection. This combination resulted in higher sensitivity as compared to other very sensitive methods. Hence, the cell-based ELISA described here is sensitive, repeatable, and accurate, and is able to quantify PLTX in mussels at concentrations lower than the maximum limit suggested by EFSA (30 µg PLTX/kg shellfish meat). Thus, the high sensitivity, specificity, and rapidity make the cell-based ELISA a suitable method for PLTX screening in mussels during monitoring programs. Further studies are in progress to characterize the ability of the cell-based ELISA to detect and quantify PLTX in other frequently contaminated matrices, such as other shellfish, fish, crabs, and other seafood.

Evaluation of Mussel Matrix Effect
The suitability of the cell-based ELISA for PLTX quantitation in mussels was assessed using different extracts from edible parts of Mytilus galloprovincialis collected in the Gulf of Trieste (Trieste, Italy). Shucked mussel meat (200 g) was homogenized using an Ultra-Turrax (Ika-Werk; Staufen, Germany) at 14,000 rpm and room temperature, until a homogeneous pulp was obtained (about 5 min). Mussel homogenate (1 g) was extracted 3 times with 3 mL 80% aqueous MeOH by Ultra-Turrax homogenization (14,000 rpm, 3 min) followed by centrifugation at 5500 rpm for 30 min. The supernatants were then pooled and the volume adjusted to 10 mL with 80% aqueous methanol to obtain 0.1 g mussel meat equivalent/mL. The extract was analyzed by LC-MS/MS (Thermo-Fisher, San Josè, CA, USA) to confirm the absence of PLTX before the matrix effect evaluation. Then, dilutions (1:2, 1:5, 1:10, and 1:50 in PBS, v/v) of the PLTX-free mussel extract were spiked with known PLTX concentrations to prepare a series of matrix matched samples at PLTX concentrations ranging from 1.4 × 10 −11 M to 1.1 × 10 −9 M. These samples were then analyzed by the cell-based ELISA and compared to the same PLTX concentrations in solutions free of mussel matrix.

Evaluation of PLTX Recovery from Mussels
To evaluate the recovery of PLTX from mussels, samples of the PLTX-free mussel homogenate were spiked with known amounts of PLTX and then extracted as described above to obtain extracts containing different theoretic concentrations of PLTX (1.4 × 10 −11 to 1.1 × 10 −9 M). Each extract was then analyzed by the cell-based ELISA, as previously described.

Statistical Analysis
PLTX concentrations are reported as mean ± SE of at least 3 independent experiments performed in triplicate, unless otherwise specified. For binding experiments, dissociation constant (Kd) was calculated by a 1-site binding hyperbola nonlinear regression analysis using GraphPad Prism software version 6.0 (GraphPad, Inc., San Diego, CA, USA). Maximal binding was evaluated as the maximal optical density (OD) normalized for the µg of proteins of each sample. Linearity (r 2 ) of the calibration curve was estimated by linear regression analysis, also using GraphPad Prism version 6.0. Data were compared by 2-way ANOVA followed by Bonferroni post test, and significant differences were considered at p values < 0.05.
The optimized assay was characterized according to the international principles defined by the Eurachem Guide [65]: limits of detection (LOD) and quantitation (LOQ) were estimated as PLTX concentration corresponding to the average of the optical density of 10 blank values plus 3 or 10 times the standard deviation, respectively. Repeatability was expressed as relative standard deviation of repeatability (RSDr), measured as % ratio between the standard deviation of independent results and their mean value. Independent results obtained by the same operator in 1 day (intraassay RSDr; n = 6) and by different operators within a 6-month period (interassay RSDr; n = 10) were evaluated. Accuracy was measured as % bias (n = 10), calculated as % difference between PLTX concentration measured by the assay and the theoretical concentration in the sample divided for theoretical PLTX concentrations.