Potential of Kale and Lettuce Residues as Natural Adsorbents of the Carcinogen Aflatoxin B1 in a Dynamic Gastrointestinal Tract-Simulated Model

Adsorption of the carcinogen aflatoxin B1 (AFB1) onto agro-waste-based materials is a promising alternative over conventional inorganic binders. In the current study, two unmodified adsorbents were eco-friendly prepared from kale and lettuce agro-wastes. A dynamic gastrointestinal tract-simulated model was utilized to evaluate the removal efficiency of the sorptive materials (0.5%, w/w) when added to an AFB1-contaminated diet (100 µg AFB1/kg). Different characterization methodologies were employed to understand the interaction mechanisms between the AFB1 molecule and the biosorbents. Based on adsorption results, the biosorbent prepared from kale was the best; its maximum adsorption capacity was 93.6%, which was significantly higher than that of the lettuce biosorbent (83.7%). Characterization results indicate that different mechanisms may act simultaneously during adsorption. Non-electrostatic (hydrophobic interactions, dipole-dipole interactions, and hydrogen bonding) and electrostatic interactions (ionic attractions) together with the formation of AFB1-chlorophyll complexes appear to be the major influencing factors driving AFB1 biosorption.


Introduction
Mycotoxins are low-molecular-weight (<1 kDa) substances synthesized as part of the secondary metabolism of various fungal species, including Aspergillus, Penicillium, Fusarium, and Alternaria. Most of these secondary metabolites are extremely toxic and have significant economic implications and public health concerns. Similar to other mycotoxins, aflatoxins are a group of highly toxic metabolites that can contaminate a wide variety of food and feedstuffs. The known aflatoxins are about twenty, while the four main aflatoxins are denoted as aflatoxin B 1 (AFB 1 ), aflatoxin B 2 (AFB 2 ), aflatoxin G 1 (AFG 1 ), and aflatoxin G 2 (AFG 2 ). AFB 1 is the most extensively studied fungal toxin because it is a potent liver carcinogen [1].
shown in Figure 1, the adsorption capacity of the agro-waste-based materia cantly higher than that of the non-commercial zeolitic mineral. In contrast, out the addition of sorbents (reference test) show a marked lack of AFB1 adso To date, only three studies have been conducted to evaluate the aflatoxin-so ity of different plant-derived materials using static or dynamic gastrointest procedures [6,10,11].  Figure 1. Adsorption capacities of the unmodified agro-waste-based sorbents and th cotoxin binder (zeolite) against AFB1 using a dynamic gastrointestinal tract-simul Boxes and whiskers not sharing a common superscript differ significantly (Tukey te In this context, Adunphatcharaphon et al. [10] evaluated the aflatoxin-a pacity of durian-fruit hulls (Durio zibthinus) using a three-step static proced ulated salivary, gastric, and intestinal fluids, respectively. After completing procedure (2 h), the acid-treated durian peel (0.5% w/v) reduced the AFB1 b by 95.1% at both gastric and intestinal levels. Moreover, Rasheed et al. [11] adsorption capacity of blueberry pomace (0.2% w/v) in simulated gastric (p testinal fluids (pH 7). The removal efficiencies were about 65% and 70%, resp hough these studies are consistent with our findings, these methodologie directly applicable to in vivo conditions because these models do not food/feed matrix effect. Furthermore, the disadvantages associated with methodologies for preparing the biosorbents is that they are high energy-co demand complicated procedures or highly specialized chemical substances Recently, our research group evaluated the AFB1-adsorption capacities prepared from the banana peel, Pyracantha leaves, and Aloe vera using a m mentalized model simulating the dynamic conditions in the gastrointestin general, the organic sorptive materials (1.5%, w/w) significantly reduced the ity of AFB1 in the intestinal section, being Aloe vera the biomaterial with the adsorption capacity (68.5%). However, comparing these results with those ure 1, the adsorption capacity by these novel low-cost and bio sustainable significantly higher than that of the Aloe vera powder, even with a less bioso (0.5% w/w). This observation suggests that AFB1 adsorption by the agro sorbents could be accomplished by different chemical and or physical mech

Functional Groups Involved in the Aflatoxin Adsorption
In this research, the sorbents prepared from kale and lettuce agro-wa In this context, Adunphatcharaphon et al. [10] evaluated the aflatoxin-adsorption capacity of durian-fruit hulls (Durio zibthinus) using a three-step static procedure with simulated salivary, gastric, and intestinal fluids, respectively. After completing the digestion procedure (2 h), the acid-treated durian peel (0.5% w/v) reduced the AFB 1 bioaccessibility by 95.1% at both gastric and intestinal levels. Moreover, Rasheed et al. [11] evaluated the adsorption capacity of blueberry pomace (0.2% w/v) in simulated gastric (pH 2.5) and intestinal fluids (pH 7). The removal efficiencies were about 65% and 70%, respectively. Although these studies are consistent with our findings, these methodologies may not be directly applicable to in vivo conditions because these models do not consider the food/feed matrix effect. Furthermore, the disadvantages associated with the proposed methodologies for preparing the biosorbents is that they are high energy-consuming and demand complicated procedures or highly specialized chemical substances.
Recently, our research group evaluated the AFB 1 -adsorption capacities of biosorbents prepared from the banana peel, Pyracantha leaves, and Aloe vera using a multicompartmentalized model simulating the dynamic conditions in the gastrointestinal tract [6]. In general, the organic sorptive materials (1.5%, w/w) significantly reduced the bioavailability of AFB 1 in the intestinal section, being Aloe vera the biomaterial with the highest AFB 1 adsorption capacity (68.5%). However, comparing these results with those shown in Figure 1, the adsorption capacity by these novel low-cost and bio sustainable sorbents was significantly higher than that of the Aloe vera powder, even with a less biosorbent content (0.5% w/w). This observation suggests that AFB 1 adsorption by the agro-waste-based sorbents could be accomplished by different chemical and or physical mechanisms.

Functional Groups Involved in the Aflatoxin Adsorption
In this research, the sorbents prepared from kale and lettuce agro-wastes adsorbed variable amounts of AFB 1 , suggesting different binding mechanisms. Therefore, sorbents were further characterized to obtain information about the interaction between the functional groups present in the sorptive materials and the AFB 1 molecule. The experimentally measured transmittivity spectra and the functional groups of the sorptive materials via at-Toxins 2021, 13, 771 4 of 15 tenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy are shown in Figure 2. Significant differences in five IR-active vibrations were detected among the biosorbents, including: (i) the broad frequency band related to hydroxyl groups at 3281 cm −1 , (ii) the medium to weak frequency vibrations of alkyl chains at 2916 and 2850 cm −1 , (iii) the strong vibration of the carboxyl group at 1613 cm −1 , (iv) the strong vibration of the C=C bond in aromatics at 1406 cm −1 , and (v) the strong molecular vibration of the C-O bond at approximately 1031 cm −1 (Figure 2, Profile a). These functional groups play essential roles in AFB 1 adsorption, which is in close agreement with our previous works [4,6,14,15].  (Figure 2, Profile a). These functional groups play essential roles in AFB1 adsorption, which is in close agreement with our previous works [4,6,14,15]. On the other hand, the non-commercial zeolitic mineral ( Figure 2, Profile b) showed the characteristic band at around 3623 cm −1 associated with Al 3+ -OH in the octahedral sheet [16]. The zeolite was also hydrated, which was demonstrated by significant water absorption bands at 3388 and 1632 cm −1 . The very strong band centered at 997 cm −1 and the medium band located at 791 cm −1 are related to the antisymmetric and symmetric stretch vibrations of Si-O-Si bonds. The absorption band at 598 cm −1 was attributed to the presence of heulandite [17]. The band at 513 cm −1 can be assigned to the presence of the double rings of the zeolite [18]. Finally, the absorption band at 440 cm −1 is associated with the bending vibration of Si-O bonds [19].
Furthermore, in the IR spectrum, the positions and numbers of the absorption bands of the agro-waste-based sorbents were essentially the same. However, significant differences in intensity (relative transmittance) were detected. In this context, band intensity and band area are two parameters commonly used to calculate a particular chemical bond concentration. Band area is often used because this parameter provides minor variation, considering that several overlapped absorption bands may occur. In this research, several chemical bonds were investigated to clarify the binding mechanisms among the biosorbents. Figure 3 shows the bond indexes present in the sorptive materials. For the agrowaste-based sorbents, five bond indexes were calculated: the OH stretch, the CH2 and CH3 stretch, the C=O stretch, the C=C stretch, and the C-O stretch. In the biosorbent prepared from kale, it was found that the OH and C-O indexes were significantly lower than those in the lettuce biosorbent ( Figure 3, Profile a). This decrease in OH and C-O indexes can cause low hydrophilicity in the kale biosorbent, which would make its surface more favorable to the adsorption of AFB1 molecules. Moreover, the CH2 and CH3, C=O, and C=C indexes were significantly higher in the kale biosorbent. The high number of hydrophobic groups may also help the kale biosorbent to remove AFB1 efficiently. It has been reported that an increased number of hydrophobic groups (such as methyl and aromatic groups) On the other hand, the non-commercial zeolitic mineral ( Figure 2, Profile b) showed the characteristic band at around 3623 cm −1 associated with Al 3+ -OH in the octahedral sheet [16]. The zeolite was also hydrated, which was demonstrated by significant water absorption bands at 3388 and 1632 cm −1 . The very strong band centered at 997 cm −1 and the medium band located at 791 cm −1 are related to the antisymmetric and symmetric stretch vibrations of Si-O-Si bonds. The absorption band at 598 cm −1 was attributed to the presence of heulandite [17]. The band at 513 cm −1 can be assigned to the presence of the double rings of the zeolite [18]. Finally, the absorption band at 440 cm −1 is associated with the bending vibration of Si-O bonds [19].
Furthermore, in the IR spectrum, the positions and numbers of the absorption bands of the agro-waste-based sorbents were essentially the same. However, significant differences in intensity (relative transmittance) were detected. In this context, band intensity and band area are two parameters commonly used to calculate a particular chemical bond concentration. Band area is often used because this parameter provides minor variation, considering that several overlapped absorption bands may occur. In this research, several chemical bonds were investigated to clarify the binding mechanisms among the biosorbents. Figure 3 shows the bond indexes present in the sorptive materials. For the agro-wastebased sorbents, five bond indexes were calculated: the OH stretch, the CH 2 and CH 3 stretch, the C=O stretch, the C=C stretch, and the C-O stretch. In the biosorbent prepared from kale, it was found that the OH and C-O indexes were significantly lower than those in the lettuce biosorbent ( Figure 3, Profile a). This decrease in OH and C-O indexes can cause low hydrophilicity in the kale biosorbent, which would make its surface more favorable to the adsorption of AFB 1 molecules. Moreover, the CH 2 and CH 3 , C=O, and C=C indexes were significantly higher in the kale biosorbent. The high number of hydrophobic groups may also help the kale biosorbent to remove AFB 1 efficiently. It has been reported that an increased number of hydrophobic groups (such as methyl and aromatic groups) results in a highly hydrophobic surface, which is favorable to the adsorption capacity of sorbent materials [20]. sorbent materials [20].
For the non-commercial zeolitic mineral, only three bond indexes were calculated: the OH stretch, the Si-O-Si stretch and the Si-O bond ( Figure 3, Profile b). In general, the bond index of Si-O-Si was significantly higher than the others, which means that the Si-O-Si surface is strictly hydrophobic [21]. Although, the AFB1 adsorption into clays is mediated by weak electrostatic attractions, other mechanisms such as moderate electron donor-acceptor attraction and a strong calcium-bridging linkage are also responsible for the AFB1 adsorption [22].

Point of Zero Charge (pHpzc) and Zeta Potential (ζ-Potential)
The net electrical charge of the sorbent surface plays an essential role during sorption. In this regard, pHpzc gives helpful information about the surface charge of the biosorbents and ensures that electrostatic interaction is one of the mechanisms that drive adsorption. In solution, at pH > pHpzc, the sorbent surface is negatively charged and could interact with positively charged sorbate species. In contrast, at pH < pHpzc, the sorbent surface becomes positively charged (the acidic water donates more protons than hydroxide groups) and starts trapping negatively charged species. Figure 4 (Profile a) shows the pHpzc of the agro-waste-based sorbents and the inorganic mycotoxin binder (zeolite). It appears from this figure that the pHpzc, that is, the pH at which the adsorption of potential positively or negatively charged species is identical, lies in the vicinity of pH 6 for both kale and lettuce biosorbents. A pHpzc near pH 9 was recorded for the zeolitic material. In this context, Nava-Ramírez et al. [4] reported a pHpzc value of 5.65 for a biosorbent prepared from lettuce. Such a difference might be due in part to differences in the maturity stage, atmospheric conditions during growing, and harvest phase. Interestingly, both agro-waste-based sorbents have high negative-charged surfaces under intestinal pH simulation; as a result, these sorptive materials have significant AFB1 uptakes in the in vitro digestive model. On the other hand, the zeolite showed a constant positive charge in the pH range from 2 to 8.8 ( Figure 4, Profile a), confirming that AFB1 adsorption is not driven primarily by electrostatic attractions. For the non-commercial zeolitic mineral, only three bond indexes were calculated: the OH stretch, the Si-O-Si stretch and the Si-O bond ( Figure 3, Profile b). In general, the bond index of Si-O-Si was significantly higher than the others, which means that the Si-O-Si surface is strictly hydrophobic [21]. Although, the AFB 1 adsorption into clays is mediated by weak electrostatic attractions, other mechanisms such as moderate electron donor-acceptor attraction and a strong calcium-bridging linkage are also responsible for the AFB 1 adsorption [22].

Point of Zero Charge (pH pzc ) and Zeta Potential (ζ-Potential)
The net electrical charge of the sorbent surface plays an essential role during sorption. In this regard, pH pzc gives helpful information about the surface charge of the biosorbents and ensures that electrostatic interaction is one of the mechanisms that drive adsorption. In solution, at pH > pH pzc , the sorbent surface is negatively charged and could interact with positively charged sorbate species. In contrast, at pH < pH pzc , the sorbent surface becomes positively charged (the acidic water donates more protons than hydroxide groups) and starts trapping negatively charged species. Figure 4 (Profile a) shows the pH pzc of the agro-waste-based sorbents and the inorganic mycotoxin binder (zeolite). It appears from this figure that the pH pzc , that is, the pH at which the adsorption of potential positively or negatively charged species is identical, lies in the vicinity of pH 6 for both kale and lettuce biosorbents. A pH pzc near pH 9 was recorded for the zeolitic material. In this context, Nava-Ramírez et al. [4] reported a pH pzc value of 5.65 for a biosorbent prepared from lettuce. Such a difference might be due in part to differences in the maturity stage, atmospheric conditions during growing, and harvest phase. Interestingly, both agro-wastebased sorbents have high negative-charged surfaces under intestinal pH simulation; as a result, these sorptive materials have significant AFB 1 uptakes in the in vitro digestive model. On the other hand, the zeolite showed a constant positive charge in the pH range from 2 to 8.8 ( Figure 4, Profile a), confirming that AFB 1 adsorption is not driven primarily by electrostatic attractions.  To further elucidate the possible mechanism through which agro-waste-based sorbents can efficiently adsorb AFB1, a ζ-potential study was also conducted. Zeta poten tial is the electric potential in the interfacial double layer at the location of the slipping plane relative to a point in the bulk fluid away from the interface. Figure 4 (Profile b shows the relationship between ζ-potential and pH of the sorptive materials. In general ζ-potential significantly increased with increasing pH and reached the maximum at pH 11. It is important to highlight that both agro-waste-based sorbents presented high nega tive ζ-potential values in two out of three compartments of the dynamic gastrointestina tract-simulated model; as a result, the interaction type in these compartments would be electrostatic in nature. Moreover, it was determined that the isoelectric point (iep) of the agro-waste-based sorbents occurs in the range of pH 2-3, indicating that the potential a the slipping plane (that is, the ζ-potential) is zero. In other words, at this particular pH particles do not move when exposed to an electric field. Graph b in Figure 4 also show that the zeolite was negatively charged throughout the entire pH range (from 2 to 11) consequently, the iep could not be determined. In general, for agro-waste-based sorbents the iep (determined by changes in ζ-potential vs. pH) was found to be very different from the pHpzc (calculated in terms of ΔpH). These differences in the iep and pHpzc on the pH scale point to complex specific AFB1-adsorption at the interface. Interestingly, the ζ-po tential values were significantly higher in the lettuce biosorbent. Under these circum stances, it would be expected that the sorbent prepared from lettuce adsorbed AFB1 more efficiently than the kale biosorbent. This new finding pointed out that different mecha nisms could accomplish adsorption.

Non-Destructive Estimation of Pigments
It has been reported that pigments fulfill several important roles. For instance: chlo rophylls can form strong non covalent complexes in vitro with AFB1 [4]; carotenoids are recognized as powerful antioxidants scavenging both singlet molecular oxygen and per oxyl radicals [23]; and anthocyanins are also potent polyphenolic antioxidants [24]. Figure  5 shows the diffuse reflectance spectra of the sorbents. For the agro-waste-based sorbents variable absorbance was observed in both the visible and near infrared regions of the spec trum ( Figure 5, Profile a). In the red region, two distinctive bands are clearly distin guished, a broad band of Chlorophyll a (Chl a) at 677 nm and a shoulder of Chlorophyll (Chl b) at about 650 nm [6]. In the blue region -where chlorophylls and carotenoids ab To further elucidate the possible mechanism through which agro-waste-based sorbents can efficiently adsorb AFB 1 , a ζ-potential study was also conducted. Zeta potential is the electric potential in the interfacial double layer at the location of the slipping plane relative to a point in the bulk fluid away from the interface. Figure 4 (Profile b) shows the relationship between ζ-potential and pH of the sorptive materials. In general, ζ-potential significantly increased with increasing pH and reached the maximum at pH 11. It is important to highlight that both agro-waste-based sorbents presented high negative ζ-potential values in two out of three compartments of the dynamic gastrointestinal tract-simulated model; as a result, the interaction type in these compartments would be electrostatic in nature. Moreover, it was determined that the isoelectric point (iep) of the agro-waste-based sorbents occurs in the range of pH 2-3, indicating that the potential at the slipping plane (that is, the ζ-potential) is zero. In other words, at this particular pH, particles do not move when exposed to an electric field. Graph b in Figure 4 also shows that the zeolite was negatively charged throughout the entire pH range (from 2 to 11); consequently, the iep could not be determined. In general, for agro-waste-based sorbents, the iep (determined by changes in ζ-potential vs. pH) was found to be very different from the pH pzc (calculated in terms of ∆pH). These differences in the iep and pH pzc on the pH scale point to complex specific AFB 1 -adsorption at the interface. Interestingly, the ζ-potential values were significantly higher in the lettuce biosorbent. Under these circumstances, it would be expected that the sorbent prepared from lettuce adsorbed AFB 1 more efficiently than the kale biosorbent. This new finding pointed out that different mechanisms could accomplish adsorption.

Non-Destructive Estimation of Pigments
It has been reported that pigments fulfill several important roles. For instance: chlorophylls can form strong non covalent complexes in vitro with AFB 1 [4]; carotenoids are recognized as powerful antioxidants scavenging both singlet molecular oxygen and peroxyl radicals [23]; and anthocyanins are also potent polyphenolic antioxidants [24]. Figure 5 shows the diffuse reflectance spectra of the sorbents. For the agro-waste-based sorbents, variable absorbance was observed in both the visible and near infrared regions of the spectrum ( Figure 5, Profile a). In the red region, two distinctive bands are clearly distinguished, a broad band of Chlorophyll a (Chl a) at 677 nm and a shoulder of Chlorophyll b (Chl b) at about 650 nm [6]. In the blue region-where chlorophylls and carotenoids absorb-some spectral details could be attributed to carotenoids (425 nm) and, to a lesser extent, to Chl b absorption [25]. Finally, in the green region, the absorbance at 550 nm can be associated with the presence of anthocyanins [26]. As shown from Figure 5 (Profile a), absorbance increases sharply with increasing pigment content (up to 0.95 A. U. in the sorbent prepared from kale). Furthermore, the zeolitic material only showed an intraconfigurational transition in the visible region near 500 nm ( Figure 5, Profile b). This minor contribution is usually associated with FeO x oligomers in the clay material [27].
oxins 2021, 13, x FOR PEER REVIEW 7 o absorbance increases sharply with increasing pigment content (up to 0.95 A. U. in sorbent prepared from kale). Furthermore, the zeolitic material only showed an intrac figurational transition in the visible region near 500 nm ( Figure 5, Profile b). This mi contribution is usually associated with FeOx oligomers in the clay material [27].

Quantitative Determination of Chlorophylls and Carotenoids
Many studies have shown that chlorophylls have significant anticarcinogenic pot tial against a wide range of human carcinogens, including AFB1 [28][29][30][31]. Therefore, dif ent mechanisms responsible for the cancer-preventative activity have been proposed, cluding antioxidant activity [32,33], modulation of detoxification pathways [34], induct of apoptosis [35], and carcinogen trapping [4,[36][37][38]. The last statement is more plausi because, in our previous work, it was demonstrated that the formation of AFB1-chlo phyll complexes improves the rate of AFB1 uptake by biosorbents containing considera amounts of chlorophylls [4]. In the present study, major photosynthetic pigments w extracted with ethanol and their content was determined spectrophotometrically us the absorption coefficients of Lichtenthaler and Wellburn [39].

Quantitative Determination of Chlorophylls and Carotenoids
Many studies have shown that chlorophylls have significant anticarcinogenic potential against a wide range of human carcinogens, including AFB 1 [28][29][30][31]. Therefore, different mechanisms responsible for the cancer-preventative activity have been proposed, including antioxidant activity [32,33], modulation of detoxification pathways [34], induction of apoptosis [35], and carcinogen trapping [4,[36][37][38]. The last statement is more plausible because, in our previous work, it was demonstrated that the formation of AFB 1 -chlorophyll complexes improves the rate of AFB 1 uptake by biosorbents containing considerable amounts of chlorophylls [4]. In the present study, major photosynthetic pigments were extracted with ethanol and their content was determined spectrophotometrically using the absorption coefficients of Lichtenthaler and Wellburn [39]. Figure 6 (Profile a) shows the absorption spectra of Chl a, Chl b, and total carotenoid in ethanol.
In general, the UV-Vis spectrum was dominated by the absorption of Chl a at 432 nm (blue region) and 665 nm (red region). Carotenoids presented a broad absorption-with three maxima-in the blue spectral range (418, 432, and 467 nm). The results obtained by UV-Vis spectroscopy indicate that the biosorbent prepared from kale had higher amounts of photosynthetic pigments in comparison with its counterpart. In an attempt to confirm these results, chlorophylls were further characterized spectrofluorometrically. Figure 6 (Profile b) shows the chlorophyll fluorescence spectra of the agro-waste-based sorbents. The spectra of both biosorbents exhibit two maxima near 690 nm and 735 nm, respectively. However, the shape of the fluorescence spectra differs considerably depending on the chlorophyll content. With increasing chlorophyll content in the biosorbent (Table 1), the fluorescence yield significantly decreased due to reabsorption of the shorter wavelength fluorescence at around 690 nm [40]. Furthermore, the ratio of chlorophyll fluorescence at the two maxima (F 690 /F 735 ) as determined from the spectra of Figure 6 (Profile b) decreased considerably with increasing chlorophyll content from 0.113 (lettuce) to 0.104 (kale). These results confirmed that the biosorbent prepared from kale undoubtedly presented more chlorophyll content than lettuce biosorbent. Thus, the biosorbent prepared from kale contained approximately 30% more Chl a and up to 57% more total carotenoids than lettuce biosorbent (Table 1). Therefore, the hydrophobicity of chlorophylls and carotenoids leads to an improved trapping efficiency of the kale biosorbent to immobilize almost all AFB 1 molecules in the dynamic gastrointestinal tract-simulated model. ent mechanisms responsible for the cancer-preventative activity have been proposed, including antioxidant activity [32,33], modulation of detoxification pathways [34], induction of apoptosis [35], and carcinogen trapping [4,[36][37][38]. The last statement is more plausible because, in our previous work, it was demonstrated that the formation of AFB1-chlorophyll complexes improves the rate of AFB1 uptake by biosorbents containing considerable amounts of chlorophylls [4]. In the present study, major photosynthetic pigments were extracted with ethanol and their content was determined spectrophotometrically using the absorption coefficients of Lichtenthaler and Wellburn [39]. Figure 6 (Profile a) shows the absorption spectra of Chl a, Chl b, and total carotenoid in ethanol.

The Mechanism for AFB 1 Adsorption onto Agro-Waste-Based Sorbents
The adsorption of AFB 1 onto agro-waste-based sorbents can be a combination of electrostatic and non-electrostatic interactions (Figure 7). In general, electrostatic interactions are dependent upon the pH of the solution. At the experimental pH values of the gastrointestinal tract-simulated model (pH 2-7), the AFB 1 molecule is neither protonated nor deprotonated. However, the adsorption of AFB 1 was significantly increased above pH 6, suggesting that the surface of the agro-waste-based sorbents became extensively deprotonated, thereby increasing the adsorption of AFB 1 molecules under the intestinal pH simulation (pH 7). Moreover, the agro-waste-based sorbents contain several functional groups such as hydroxyl, amino, carboxyl, and ester that can efficiently establish hydrogen bonds with the oxygen atoms of the ether, carbonyl, and methoxy groups in the AFB 1 molecule. In particular, the pK a -acid dissociation constant-of a carboxylic acid is~5; thus, it is impossible to be found in the protonated form under the experimental conditions of the intestinal section (pH 7). Consequently, the resulting carboxylate ion will not form hydrogen bonds with the oxygen atoms of the AFB 1 molecule in this gastrointestinal tract-simulated section. Furthermore, the high number of bond indexes related to hydrophobic groups such as CH 2 , CH 3 , and C=C in the sorptive materials resulted in a highly hydrophobic surface, which was favorable to the adsorption of AFB 1 molecules via dipole-dipole or hydrophobic interactions. Finally, the agro-waste-based sorbents also contain considerable amounts of photosynthetic pigments, among them Chl a, which can form strong noncovalent complexes with the AFB 1 molecule independent of pH [4].
Consequently, the combination of these governing mechanisms resulted in materials with enhanced AFB 1 -adsorption performance.

21, 13, x FOR PEER REVIEW 9 of 15
Consequently, the combination of these governing mechanisms resulted in materials with enhanced AFB1-adsorption performance. Figure 7. The proposed mechanism for the adsorption of AFB1 by the unmodified agro-wastebased materials.

Conclusions
In this research, two agro-waste-based sorbents were eco-friendly prepared from kale and lettuce residues and both exhibited competitive adsorption capacities. The biosorbent prepared from kale has significant potential for removing AFB1 in the dynamic gastrointestinal tract-simulated model (up to 93.6%). Thus, kale residues can be considered a promising agro-waste material for developing novel AFB1 binders. To our knowledge, this is the first report showing that unmodified agro-waste-based sorbents containing considerable amounts of photosynthetic pigments have significant ability to remove AFB1 in a dynamic gastrointestinal model. Although this model seems to be completer and more realistic of those available in the literature, it also has some limitations, as there are no mucosal cells, microbiota, and immune system inside the model. Consequently, biosorbents must be tested in in vivo models to demonstrate their efficacy in counteracting the toxic effects of AFB1 and other occurring mycotoxins such as fumonisins, ochratoxin A (OTA), and zearalenone (ZEA). Research in this direction is in progress in our laboratories.

Materials
Kale (Brassica oleracea L.) and lettuce (Lactuca sativa L.) agro-wastes were kindly donated by a local horticultural producer group (Tenango del Valle, State of Mexico, Mexico). Wastes were collected at the end of the winter season. A non-commercial zeolitic mineral from Taxco-Guerrero, Mexico, was used as reference material. The zeolite was powdered and sieved, and particles with diameter < 250 µm were selected to carry out the experiments.

Conclusions
In this research, two agro-waste-based sorbents were eco-friendly prepared from kale and lettuce residues and both exhibited competitive adsorption capacities. The biosorbent prepared from kale has significant potential for removing AFB 1 in the dynamic gastrointestinal tract-simulated model (up to 93.6%). Thus, kale residues can be considered a promising agro-waste material for developing novel AFB 1 binders. To our knowledge, this is the first report showing that unmodified agro-waste-based sorbents containing considerable amounts of photosynthetic pigments have significant ability to remove AFB 1 in a dynamic gastrointestinal model. Although this model seems to be completer and more realistic of those available in the literature, it also has some limitations, as there are no mucosal cells, microbiota, and immune system inside the model. Consequently, biosorbents must be tested in in vivo models to demonstrate their efficacy in counteracting the toxic effects of AFB 1 and other occurring mycotoxins such as fumonisins, ochratoxin A (OTA), and zearalenone (ZEA). Research in this direction is in progress in our laboratories.

Materials
Kale (Brassica oleracea L.) and lettuce (Lactuca sativa L.) agro-wastes were kindly donated by a local horticultural producer group (Tenango del Valle, Mexico City, Mexico). Wastes were collected at the end of the winter season. A non-commercial zeolitic mineral from Taxco-Guerrero, Mexico, was used as reference material. The zeolite was powdered and sieved, and particles with diameter < 250 µm were selected to carry out the experiments.

Unmodified Biosorbent Preparation
Fresh kale and lettuce leaves were thoroughly cleaned with a brush under tap water to remove all traces of sand and soil, and then washed with distilled water. After cleaning, leaves were cut into small pieces (10 cm 2 ) and separately dehydrated in a solar drying system designed and manufactured by UNAM-FES-C. As shown in Figure 8, the experimental set-up consists of a natural conventional solar dryer with a flat plate solar collector and a cabinet dryer. Drying experiments were carried out during March 2021 at the Research in Renewable Energies Laboratory (UNAM-FES-C). Kale and lettuce samples (batches of 1 kg) were uniformly distributed as a single layer on the trays and dehydrated for 10 h using a drying air temperature of 60 • C and a drying airflow rate of 0.017 kg/s. The global radiation reached the maximum value of 1800 W/m 2 at 13:00 (local time). Weight losses of leaves were measured repeatedly until the average moisture content reached a constant value (approximately 7%). Dehydrated samples were ground and sieved (60 mesh) to obtain materials with particle sizes of <250 µm. The unmodified agro-waste-sorbent materials were stored in vacuum-sealed plastic containers at −20 • C until further analysis. Weight losses of leaves were measured repeatedly until the average moisture content reached a constant value (approximately 7%). Dehydrated samples were ground and sieved (60 mesh) to obtain materials with particle sizes of <250 µm. The unmodified agrowaste-sorbent materials were stored in vacuum-sealed plastic containers at −20 °C until further analysis.

Preparation of the AFB1-Contaminated Diet
As a primary standard solution, aflatoxin B1 (100 µg AFB1/mL) was prepared in dimethyl sulfoxide. Afterwards, the concentrated solution was diluted to 1 µg AFB1/mL using distilled water. A typical maize-soybean meal diet containing 19.5% protein (13 MJ/kg metabolizable energy) was prepared. The compositional analysis of the diet is presented in Table 2. Diet samples were artificially contaminated to reach an aflatoxin content of 100 µg AFB1/kg. Finally, five samples were randomly taken, and the presence of AFB1 was confirmed using the immunoaffinity column clean-up and liquid chromatography with fluorescence detection methodology.

Preparation of the AFB 1 -Contaminated Diet
As a primary standard solution, aflatoxin B 1 (100 µg AFB 1 /mL) was prepared in dimethyl sulfoxide. Afterwards, the concentrated solution was diluted to 1 µg AFB 1 /mL using distilled water. A typical maize-soybean meal diet containing 19.5% protein (13 MJ/kg metabolizable energy) was prepared. The compositional analysis of the diet is presented in Table 2. Diet samples were artificially contaminated to reach an aflatoxin content of 100 µg AFB 1 /kg. Finally, five samples were randomly taken, and the presence of AFB 1 was confirmed using the immunoaffinity column clean-up and liquid chromatography with fluorescence detection methodology.

Adsorption Performance
The adsorptive capacity of the different materials was evaluated in a dynamic gastrointestinal tract-simulated model reported by Hernandez-Patlán et al. [41] with minimal modifications. The assay was performed with one control (zeolite) and two different treatments (agro-waste-based sorbents). The experimental set-up consists of a biochemical oxygen demand incubator set at 40 • C accessorized with an orbital shaker. Tubes were held at 30 • angle inclination to enable proper blending. Briefly, 5 g of the aflatoxin-contaminated feed (100 µg AFB 1 /kg) plus 0.5% (w/w) sorbent were placed in 50 mL polypropylene centrifuge tubes and mixed vigorously. To simulate the enlarged part of the esophagus environment, tubes were added with 10 mL of 0.03 M HCl, reaching values around pH 5. All tubes were incubated for 30 min at 40 • C and shaken at 19 rpm. After the incubation, in each tube, 2.5 mL of 1.5 M HCl and 3000 U of pepsin (Merck KGaA, Darmstadt, Germany) per gram of diet were added to reach a pH of around 2. These conditions simulated the beginning of digestion, and tubes were incubated again 45 min. The third step was intended to mimic the intestinal section (pH~7). For that, 6.84 mg of 8× pancreatin (Merck KGaA, Darmstadt, Germany) in 6.5 mL of 1 M NaHCO 3 were added and tubes were incubated for another 120 min. Under these circumstances, the complete in vitro digestion procedure took 195 min. At the end of the incubation, tubes were centrifuged at 7000× g for 30 min and the supernatant was collected for AFB 1 quantification. Control samples (without adsorbents) were used to know the real concentration of AFB 1 per tube under the gastrointestinal tract-simulated conditions. All determinations were carried out in quintuplicate.

Aflatoxin Assay
The supernatant was cleaned up with monoclonal antibody-based immunoaffinity columns (Vicam, Watertown, MA, USA), and the eluate obtained was used for ultraperformance liquid chromatography (UPLC) analysis. A modified method previously described by Hernández-Ramírez et al. [42] was employed. An ultra-performance liquid chromatograph system (Waters ACQUITY H-class) equipped with a quaternary solvent manager and a reverse phase column (2.1 mm × 100 mm, 1.7 µm particles) was used. Methanolic extracts collected from the immunoaffinity columns (1 µL) were injected and eluted with a mobile phase of water:methanol:acetonitrile (64:18:18) at a flow rate of 0.7 mL/min. Detection was accomplished with a fluorescence detector set at 365 nm excitation and 429 nm emission. The AFB 1 concentration was calculated using a standard reference (AFB 1 , Merck KGaA, Darmstadt, Germany) with a calibration curve. The detection limit of AFB 1 was found to be 0.002 µg/L.

ATR-FTIR Spectroscopy
In this study, ATR-FTIR spectroscopy was used to analyze the vibrational features of the sorbents. The spectra were acquired using a Frontier SP8000 FTIR spectrophotometer (Perkin Elmer, Waltham, MA, USA) in the wavenumber range 4000-400 cm −1 at a resolution of 4 cm −1 . The spectra were subsequently analyzed with the Spectrum 10.4.2 software. The bond indexes (BI) of the principal chemical functional groups were calculated by using the following equations:

•
For the agro-waste-based sorbents: where BA x = the band area around the corresponding wavenumber (cm −1 ), and Σ BA = the total area of all bands in the corresponding IR spectrum.

Point of Zero Charge (pH pzc ) and Zeta Potential (ζ-Potential) Measurements
The pH pzc methodology was performed by adding identical amounts of sorbents (125 mg) to a set of flasks containing distilled water at different pH values (2, 5, 7, 9, and 11). The pH was adjusted with HCl (0.1 M) or NaOH (0.1 M) as needed. The pH values of the distilled water were denoted as the initial pH (pH i ). Subsequently, samples were shaken for 195 min using an orbital shaker at 200 rpm. After settling, the pH value of the supernatant was measured using a combination glass electrode and denoted as the final pH (pH f ). The pH pzc was obtained from the plot of ∆pH (pH f -pH i ) against pH i . Furthermore, ζ-potential measurements were performed with a ZetaSizer Pro (Malvern Instruments, Worcestershire, UK). All determinations were performed at room temperature by diluting 500 µL of the sorbent suspension (0.5% w/v) in 5 mL deionized water. Samples were evaluated at five different pH values (2, 5, 7, 9, and 11), including those that simulate the multicompartmentalized gastrointestinal model. The isoelectric point (iep) defined as the pH at which the ζ-potential was zero was obtained by plotting the curve of ζpotential against pH. Each set of experiments (pH pzc and ζ-potential) was performed in quintuplicate.

Determination of Chlorophylls and Carotenoids in the Agro-Waste-Based Sorbents Spectral Reflectance Measurements
Diffuse reflectance spectra were recorded in a range of 400-800 nm with a Lambda 365 UV-Vis spectrophotometer (Perkin Elmer, Waltham, MA, USA) equipped with a 100 mm integrating sphere attachment to capture diffusely reflected light. To provide 100% reflectance, barium sulfate (BaSO 4 ) was used as reference material. Spectral data were interfaced to a personal computer for further processing using the UV Lab software (Perkin Elmer, Waltham, MA, USA).

Photosynthetic Pigment Analysis
The pigments were quantitatively determined in the samples used for collecting the diffuse reflectance spectra. Chlorophylls and carotenoids were extracted with ethanol and determined spectrophotometrically using a Cary 8454 UV-Vis Diode Array System spectrophotometer (Agilent Technologies, Santa Clara, CA, USA). The absorption coefficients reported by Lichtenthaler and Wellburn [39] were used for Chl a, Chl b, and total carotenoid (C x + c) estimation. The following equations were utilized to determine their contents: Chl a = 13.95 A 665 − 6.88 A 649 (9) Chl b = 24.96 A 649 − 7.32 A 665 (10) C x + c = 1000 A 470 − 2.05 Chl a − 114.8 Chl b 245 (11) To further characterize chlorophylls in the biosorbents, the ethanolic extracts were also subjected to spectrofluorometry. For this purpose, fluorescence measurements were performed using a fluorescence LS-55 spectrophotometer (Perkin Elmer, Waltham, MA, USA). Spectra were recorded in the wavelength range of 575-800 nm using a 1 cm path quartz cell. The fluorescence spectra were collected at an excitation wavelength of 440 nm using equally wide excitation and emission slits (5 nm). The ratio of chlorophyll fluorescence at the two maxima (F 690 /F 735 ) was used to indicate the potential photosynthetic activity.

Experimental Design and Statistical Analysis
The experiment was conducted as a completely randomized design with five replicates. Experimental data were subjected to one-way analysis of variance (ANOVA), and means were separated using the Tukey procedure with the SAS software [43]. A significance value of α = 0.05 was used to distinguish significant differences between treatments.