Impact of pH on the Stability and the Cross-Reactivity of Ochratoxin A and Citrinin

Mycotoxins are secondary metabolites produced by several fungi contaminating crops. In several countries, the maximum permitted levels of mycotoxins are found in foodstuffs and feedstuffs. The common strategy of mycotoxin analysis involves extraction, clean-up and quantification by chromatography. In this paper, we analyzed the reasons of underestimation of ochratoxin A (OTA) content in wine, and overestimation of OTA in wheat, depending on the pH of the clean-up step and the simultaneous presence of citrinin (CIT). We demonstrated that the increase of pH by adding polyethylene glycol (PEG) to wine led to an underestimation of OTA by conversion of OTA into open ring ochratoxin A OP-OA. In comparing three methods of extraction and clean-up for the determination of OTA and CIT in wheat—(i) an inter-laboratory validated method for OTA in cereals using immunoaffinity column clean-up (IAC) and extraction by acetonitrile/water; (ii) a validated method using IAC and extraction with 1% bicarbonate Na; and (iii) an in-house validated method based on acid liquid/liquid extraction—we observed an overestimation of OTA after immunoaffinity clean-up when CIT is also present in the sample, whereas an underestimation was observed when OTA was alone. Under neutral and alkaline conditions, CIT was partially recognized by OTA antibodies.

As some species of Penicillium (such as P. verrucosum or P citrinum) produce both OTA and CIT [21,22], these two mycotoxins can be found simultaneously in cereals. Co-contamination by CIT-OTA was observed in samples of food such as rice [3,23], olives [24,25], wheat, [4,26,27] and whole meal [15]. Co-contamination of mycotoxins still poses a risk due to the possible synergy and their additive effects [15,20,28]. The development of methodologies allowing simultaneous extraction of these mycotoxins in different matrices has been encouraged [29]. In recent years, analytical methods for extraction and analysis of these two mycotoxins have been improved [30,31]. Extractions in liquid phase were largely studied for ochratoxin A [32], and most of the validated methods were based on the extraction of ochratoxin A and citrinin through the solubility of these compounds in organic solvents or alkaline solutions [25,33]. The most widely used technique is HPLC with fluorescence detection. Different excitation and emission fluorescence parameters (OTA 335 and 465 nm; CIT 331 and 500 nm) were used to achieve the optimal conditions of detection for each toxin [4]. However, for this analytical method, an efficient sample extract clean-up is mandatory to reach low detection limits and to protect the HPLC column. The most frequent clean-up method involves solid-phase extraction (SPE) of sample by using an immunoaffinity column [29,34]. This kind of SPE greatly facilitates the clean-up stage, usually providing high purity extracts that can be directly injected in the HPLC column [35,36], and are commonly used for the determination of mycotoxins in different matrices with adequate analytical efficiency. However, their disadvantages are that specific antibodies should exist for each mycotoxin. Until now, no IAC for the determination of CIT is available. A versatile method suitable for raw cereals and cereal products using partition was developed by Molinié et al. [4].
The development of new extraction and analysis techniques had been reported. The QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Safe) [37][38][39] method is a procedure that was adopted for the extraction of mycotoxin due to its quickness and low cost. In recent years, there has been a tremendous increase in reports on the determination of OTA and/or citrinin immunosensors based on different platforms. Rapid screening tests such as biosensors [40][41][42][43][44][45] and enzyme-linked immunosorbent assays (ELISA) [46][47][48] are emerging. A further development is their simplification, by developing colored immuno-tests, like rapid disposable membrane-based assay tests or clean-up tandem immune assay column.
The aim of the current study was to explain how the pH of extraction can lead to misinterpretation of OTA content and the cross-reactivity of CIT on IAC recognizing OTA. We analyze the reason of the underestimation of OTA content in wine after one treatment commonly used to eliminate anthocyans. We compare several methods of extractions and clean-up for OTA and CIT occurrence in wheat samples [4,35,36].

Underestimation of OTA Following PEG Clean-Up of Wine
Wine was treated either by poly (1-(2-oxo-1-pyrrolidinylethylene) (PVPP) according to the official method of IOW (International Organization of Wine) or with polyethylene glycol (PEG). PVPP was more efficient than PEG in terms of anthocyanin (respectively 90% and 72%) and polyphenol elimination (respectively 80% and 74%). The amount of OTA after these treatments was measured at 333 nm. Only 16% of OTA was lost after PVPP, whereas the proportion of OTA apparently lost during PEG treatment is over 80%, but simultaneously a new peak appears at 380 nm. When we analyzed the quantity of OTA bound to PEG, we did not find more than 10% of OTA. The effect of PEG concentration and pH on the recovery of OTA from wine using immunoaffinity column clean-up was already investigated by Visconti et al. 1999 [49], showing a loss of OTA. We wonder if this underestimation could not be due to an alkalization of the medium, leading to an open ring formation no longer recognized by antibodies, as previously described for breakfast cereals and coffee [50][51][52]. Indeed, the treatment of wine with PEG8000 has shifted from an acidified medium to an alkaline medium (pH recorded was around 9, with a green color), which confirmed the alkalization of the matrix) explaining the formation of the open ring form of OTA (OP-OA) for which the maximum of absorption is around 380 nm [53]. If we measured OTA content using 380 nm, we found 84% of the initial OTA wine content.

Comparison of OTA/CIT Occurrence in Wheat Using Different Methods of Extraction and Clean-Up
A total of 33 wheat samples have been collected either in farms (18 samples) or in wheat cooperatives (15 samples) in northeastern France. OTA and CIT were analyzed either after liquid-liquid extraction in acidic conditions, using an in-house validated method [4] or after purification on the immunoaffinity column (OTA) following the RDT application note [35], or following the official method [36]. In addition CIT was also analyzed using the ELISA test. The analyses were done in triplicate on the same grinded aliquot of wheat. Table 1 shows OTA and CIT contents in these samples using both techniques of clean-up.
Only four samples (F10, F18, S2, S14) were positive for OTA after extraction in alkaline condition and purification onto the immunoaffinity column. The OTA amount ranged from 6.63 ± 2.33 µg/kg to 128.49 + 30.1 µg/kg. Eleven samples (F1, F2, F10, F 15, F18, F19, F20, S2, S9, S14, S16) were positive for OTA after extraction by acetonitrile/water and purification on IAC in PBS pH 7.4, and ranged from 1.00 ± 1.00 to 74.4 ± 8.2 µg/kg. Only three samples (F10, S2, S16) were positive for CIT after ELISA analysis. Table 1. OTA/CIT content in wheat purified on immunoaffinity column (IAC) [35,36] or by liquid partition [4], expressed in µg/kg (ppb). After acidic extraction and purification by liquid-liquid partition, half of the samples contained OTA ranging from a trace to 30.77 ± 1.54 µg/kg; and CIT ranging from a trace to 512.18 ± 61.5 µg/kg. Samples containing only OTA (F1, F2, F4, F6, F7, F14, F20, S11) were negative after extraction in alkaline condition and purification on IAC. Traces of OTA were observed in the three highest contaminated samples (F1, F2, F20) after extraction with ACN/water and purification on IAC. On the contrary, the samples where OTA was detected in large amount after IAC purification correspond to samples containing simultaneously significant amount of CIT (F10, F18, S2, S14). Three samples apparently positive for OTA after extraction with ACN/water and purification on IAC at pH 7.4 (F19, S9, S16) contained in fact only CIT. This data suggests possible interferences between OTA and CIT for OTA antibodies.

Recoveries and Confirmation of OTA Metabolites Formed
In order to confirm interference between CIT and OTA, we first pass pure mycotoxins onto IAC. The mycotoxins (OTA and/or CIT) dissolved in ultra-pure water solution or in water solution containing 1% bicarbonate were added in PBS pH 7.4 before IAC. In a second experiment, wheat samples were spiked with OTA alone or in the presence of CIT. One hundred grams of non-contaminated wheat were spiked with OTA at a final concentration equivalent to 100, 40, 7 or 3 µg/kg. Some samples were contaminated with OTA and CIT: (i) 40 µg/kg each toxins (ii) 5 µg/kg OTA ± 50 µg/kg CIT. The samples were analyzed in parallel with the three techniques: acidic extraction followed by liquid-liquid partition purification versus alkaline extraction and purification on IAC or ACN/water extraction and purification on IAC. The analyses were made in triplicate. Recoveries are shown in Table 2.  The amount of OTA recovered after IAC following alkaline extraction is less than 50% when the initial OTA amount was in the range of 7 to 40 µg/kg. When the initial concentration of OTA was 3 µg /kg, no more OTA was found after IAC following extraction in alkaline condition. When OTA was spiked on wheat the recovery was of the same order. The amount of OTA recovered after IAC following neutral extraction is better (around 60%) and not dependent of the amount of OTA. In contrast, the recoveries of OTA and CIT after purification by liquid/liquid partition are ranged between 75% and 90%.
When OTA and CIT were simultaneously passed through IAC either from aqueous solution or from wheat extract, OTA recoveries were much higher than 100%, whereas by liquid/liquid extraction, the recoveries ranged from 75% to 90%. When CIT alone is passed through the IAC, a peak eluting to the same retention time than OTA was detected. We confirm the presence of CIT by mass spectrum analysis (m/z negative [M − H] − 250 > 232; 217). Thus, CIT is partially recognized by OTA antibodies under neutral and alkaline conditions.

Effect of the pH on OTA Quantification and UV Spectra in Presence or Absence of CIT
We recorded the spectra of OTA (Figure 3) or CIT alone at various pH (4, 7, 8 or 12), and in the presence of OTA (Figure 4), in order to understand how the pH of extraction influences OTA or/and CIT analysis.
At pH 4, the spectrum of OTA shows a peak at 333 nm (Figure 3). At pH 7, OTA exists in protonated OTA form corresponding to the peak at 333 nm and in dianion form having a maximal absorbance at 380 nm [50,53,54]. At pH 8, the spectrum shifts to a higher wavelength (around 380 nm) with an increase of the molar absorbance which corresponds to the dianionic form of OTA, and open ring OTA (OP-OA) [50]. At pH 12, the spectrum shifts even to a higher wavelength (around 385 nm) with an increase of optical density, corresponding exclusively to OP-OA. OTA possesses two pKa (pKa = 4.4 (carboxyl group) [54]; pKa = 7.3 (phenol hydroxyl group) [53]). These data confirm those obtained by Verrone et al. 2007 [55].   (4, 7, 8 and 12). Curve B represents the spectrum of CIT in mixture with OTA at pH 4. Curve C represents the spectrum of CIT in mixture with OTA at pH 7. Curve D represents the spectrum of CIT in mixture with OTA at pH 8.
OP-OA is not recognized by antibodies and leads to an underestimated amount. The conversion is reversible. If the pH is lowered, OP-OA is converted into OTA ( Figure 5).  [50][51][52].
Whatever the pH, CIT presented a peak at 320 nm ( Figure 4A), but the intensity of the UV signal is lower at pH > 7 compared to pH 4. The decrease of absorbance could be due to the transformation of citrinin into citrinin H2 ( Figure 6) [55] as it was already observed with OTA. Figure 6. Ring opening of citrinin (citrinin H2) induced by increase of pH [56].
When CIT and OTA are present in the mixture, one or two peaks are observed, depending on the pH ( Figures 4B,C,D). At pH 4, the spectrum of the mixture shows only one peak at 320 nm ( Figure 4B). At pH > 7 two peaks are observed. At pH 7, the main peak is at 320 nm, and a small peak at 380 nm. At pH > 8, the two peaks have the same intensity ( Figure 4D). The appearance of a peak at 380 nm at pH 7 when both toxins are present together can be explained not only by the formation of OP-OA, but also by the formation of ochratoxin quinone (OTHQ) for which the maximum of absorbance is of 380 nm [52]. Chemically, citrinin is a quinone derivative which is pro-oxidant agent susceptible to transform OTA into OTHQ [28]. Simultaneously to the formation of OTHQ, OTA is also converted into OTB (dechlorinated OTA) for which the maximum of absorbance is 318 nm (very close to that of CIT) [52].
The underestimation of the OTA amount under alkaline pH results to the opening of the lactone cycle of OTA ( Figure 5), which is no longer recognized by the anti-OTA antibodies [50][51][52]. In presence of CIT, an additional interference occurs with the formation of both OTB and OTHQ that have a similar absorbency to CIT and OP-OA, respectively. OTB is even more so recognized by OTA antibodies [52].

Confirmation by HPLC MS/MS of the Formation of OP-OA; OTHQ and OTB
In order to confirm the hypothesis of oxidation of OTA by CIT, the solution of OTA and the mixtures of OTA and CIT at different pH were analyzed by an HPLC run using the gradient system described by Faucet et al. 2006 [57]. An example of the HPLC profile is shown in Figure 7. Every peak was collected and analyzed by Nano-ESI-IT-MS in negative mode as described by Faucet-Marquis et al. 2006 [57]. The spectral data are presented Table 3. The formation of these metabolites depends on the conditions of pH and mixtures (Table 4). We confirm that OTA is transformed into OP-OA when the pH increases above 7. The conversion is total when the pH is around 12. At a pH above 7, CIT allows oxidation of OTA inducing formation of OTB (dechlorinated OTA) and OTHQ (quinone derivative). In addition, a part of OTA is transformed into OP-OA.

Preparation Standard Solution
Standard solutions of OTA and CIT were prepared by dissolving 10 mg of OTA or CIT in 1 mL of methanol. Series of working standards from 0.2 to 100 ng/mL of mycotoxin/mL were prepared by dilution in methanol and were used to calibrate the LC detector response. The OTA stock solution was determined by absorbance at 333 nm and calculated with the molar extinction coefficient ε of 5500 mol/cm. CIT stock solution was determined by absorbance at 321 nm and calculated with the molar extinction ε of 5490 mol/cm.

PEG Treatment
Ten milliliters of wine spiked with OTA (40 µg/mL) were mixed with 10 mL of a solution containing PEG8000 (1%) and NaHCO 3 (5%). This mixture was incubated during 30 min at room temperature on a rocker. Afterwards, it was centrifuged at 8000 rpm for 15 min.
The spectral analysis (200-800 nm) of the supernatant allows the quantification of the adsorbing effect of PEG for anthocyanins, and the occurrence of polyphenols in wine.

PVPP Treatment
PVPP, poly(1-(2-oxo-1-pyrrolidinyl)ethylene), is an organic synthetic polymer obtained from the polymerization of N-vinyl-2-pyrrolidone. It has the same structure than PVP (pyrrolidinylethylene) but the catalyst used for the polymerization was either sodium hydroxide or N,N'-divinylimidazolidone. This compound is not soluble in water and has a very high affinity to polyphenols. This characteristic is highly affected by the molecule's polymerization degree.
According to the IOW (International Organization of Wine), 10 mL of wine (spiked or not with OTA from 5 µg/mL to 20 µg/mL) were diluted with 20 mL of water and supplemented with 10 mg/mL of PVPP.

Sampling
Thirty-three wheat samples were collected in northeastern France. Eighteen were collected in the farms and 15 in cooperatives.

OTA Extraction from Wheat Samples
OTA was analyzed on wheat samples using two methods of extraction and clean-up: (i) an in-house validated method for the simultaneous extraction/clean-up of OTA and CIT [4]; and (ii) the immunoaffinity column (IAC) [35,36].

Solvent Extraction Clean-Up and Partioning Purification
In brief, 20 g of milled sample was acidified with an aqueous solution of potassium chloride (4%) acidified to pH 1.5 with sulfuric acid. The mixture was homogenized and extracted with acetonitrile (ACN). After filtration on Whatman No. 4 paper, the sample is defatted twice with N-hexane. The sample was then purified by liquid-liquid partition.

IAC Clean-Up
Two different extractions were made before IAC: one using alkaline extraction, another using a neutral extraction 3.4.2.2.1. OTA Extraction by Bicarbonate The first method was described in the IAC column (Rhone Diagnostics technologies, Application note Ref No. A9-P14.V1, 1999 [35]). In brief, 10 g of wheat are blended with 200 mL of 1% sodium bicarbonate solution. The sample is filtered on Whatman No. 4 paper. Twenty mL are transferred to the IAC column (3 mL/min). The column is washed with 20% methanol/water (v/v); flow rate 5 ml/min. OTA is eluted passing slowly 1.5 mL acetic acid/methanol (98/2 v/v); and after 1.5 mL water.

OTA Extraction by Methanol/Water
The second method follows the validated method described by Entwisle et al. 2000 [36]. In brief, OTA is extracted from cereals in ACN/water (60/40). The extract is filtered on Whatman No 4 paper. Forty-four mL of PBS pH 7.4 are added to 4 mL of filtrate, and the mixture is transferred onto the IAC. After washing the column with PBS, the OTA is eluted with methanol/acetic acid (98/2) and analyzed by HPLC spectrofluorimetry. [4] OTA and CIT are analyzed on RP HPLC using C18 column PRONTOSIL 120 (25 cm × 0.46 cm) with inner porosity of 3 µm, under isocratic condition (mobile phase: orthophosphoric acid at 0.33 M/ACN/propan-2-ol (600:400:55), flow rate 0.8 mL/min). Detection is performed with a programmable Merck HITACHI FL Detector L-7485 (excitation 340 nm, emission 465 nm for OTA; 331 and 500 nm for CIT). [35,36] One Hundred µL of extract are injected in HPLC using Spherisorb ® ODS2 (25 cm × 0.5 cm) with inner porosity of 5 µm. The mobile phase is ACN/water/acetic acid (51/47/2 v/v/v); flow rate 1 mL/min. Detection was performed with a programmable Merck HITACHI FL Detector L-7485 (excitation 340 nm, emission 465 nm for OTA).
CIT was either extracted by liquid-liquid extraction simultaneously with OTA as described above [4] or analyzed using Ridascreen ® .

Conclusions
Extraction of OTA in alkaline medium or addition of PEG to wine increasing pH induced an underestimation of OTA content in wheat or wine, respectively. The analyses of OTA/CIT contents in wheat, following purification onto immunoaffinity column (IAC) after alkaline or neutral extraction, led to an overestimation of OTA. This bias is due to the interferences between OTA and CIT for OTA antibodies, but also to the similarity of OTHQ and OTB formed by auto-oxidation of OTA induced by CIT. The pH is crucial for reliable detection of OTA and CIT, and should be below 7 for both accurate liquid/liquid extraction and IAC clean-up. CIT can be partially recognized by OTA antibodies.