Metal ion removal using waste byssus from aquaculture

Byssus is a thread-like seafood waste that has a natural high efficiency in anchoring many metal ions thanks to its richness of diverse functional groups. It also has structural stability in extreme chemical, physical and mechanical conditions. The combination of these properties, absent in other waste materials, has novelty suggested its use as matrix for water remediation. Thus, pristine byssus, upon de-metalation, was studied to remove metal ions from ideal solutions at pH 4 and 7, as model chemical systems of industrial and environmental polluted waters, respectively. The byssus matrix’s uptake of metal ions was determined by ICP-OES and its surface microstructure investigated by SEM. The results showed that the byssus matrix excellently uptakes metal ions slightly reorganizing its surface micro-structure. As example of its efficiency: 50 mg of byssus absorbed 21.7 mg·g−1 of Cd2+ from a 10 mM solution at pH 7. The adsorption isotherm models of Freundlich and Langmuir were mainly used to describe the system at pH 7 and pH 4, respectively. In conclusion, we showed that the byssus, a waste material that is an environmental issue, has the potential to purify polluted industrial and environmental waters from metal ions.

In this research, we aim to exploit the unique composition and structure of the byssus, evolved to bind metal ions, as material for metal ion removal from polluted waters. It has the potential to be a very cheap, high efficient, and disposable or reusable adsorbent. Firstly, the byssus was de-metaled by chelating agents that left its macromolecular composition unaltered 34,40 and, secondly, tested for metal ion removal application using both metal ions naturally present in the native byssus (Fe 3+ , Zn 2+ , and Cu 2+ ) and some metal ions of environmental interest (Al 3+ , Cd 2+ , Co 2+ , Ni 2+ , Mn 2+ , and V 3+ ).

Materials and methods
Materials. Reagents and solvents were purchased from Sigma Aldrich. They were utilized without any further purification. For each experiment, daily fresh solutions were prepared. A mussel farm close to Fano (Italy) provided the byssus from Mytilus galloprovincialis.
Byssus pre-treatment. The byssus (1-3 wt. %) was collected by hand from the mussels and pre-treated according to the procedure reported in Montroni et al. 29 . Briefly, the collected byssus was washed using tap water and soap until clear washing water. Then the soap was eliminated by rinsing with distilled water and the byssus was stirred twice in ethanol for 30 min and washed again with distilled water for 15 min. The clean and dry byssus was stored in a desiccator under vacuum.
Byssus de-metalation. The de-metalation process was performed according to the procedure reported by Schmitt et al. 40 , and modified by Montroni et al. 29 . A 0.2 M ethylenediaminetetraacetic acid (EDTA) in a 0.1 M pH 4.3 Tris buffer solution was used as a metal ion chelating agent. The de-metaled byssus was then washed several times in milliQ water to restore the neutral pH and eliminate the EDTA. The air-dried sample was conserved in a plastic Petri dish in a desiccator under vacuum.
Metal ion removal experiments. Metal ion removal was performed inserting 50 mg of byssus in 2 mL of the buffered solution with different concentrations of metal ions. The experiments were performed in a polystyrene multi-well plate. For each experiment only one metal ion was used. Control experiments were performed to quantify the eventual desorption of metal from the byssus in the buffer solution used. The incubation time was 72 h, afterward, the metal solution was collected. The byssus was then washed twice with 100 µL of the buffer. The latter buffer and the metal ion solution were collected together and diluted to 5 mL using the experimental buffer. The concentration of metal ion in solution was measured using induced coupling plasma optical emission spectroscopy (ICP-OES) in the final 5 mL solutions and in the loading solutions. The detection limits are reported in Table S1. The metal ion loading concentrations used were 0, 0.01, 0.1, 0.5, 1, and 10 mM. Each experiment  Metal ion quantification within the byssus matrix. The byssus from the metal ion removal experiments was washed three times using 1 mL of Pre-milliQ water to remove experimental buffer traces. The metal inside the byssus was quantified digesting 50 mg of sample. The sample was set in a Teflon holder with 0.5 mL of H 2 O 2 (30% Carlo Erba, for electronic applications) and 6 mL of nitric acid (65% Honeywell). The holder was set in a microwave oven, Milestone, programmed to operate as follows: 2 min at 250 W, 2 min at 400 W, 1 min at 0 W, and 3 min at 750 W. The digested sample was quantitatively collected and diluted to 10 mL with water, filtered on paper, and analyzed as described in the ICP-OES section. This analytic procedure was verified using a certified reference material (Lagarosiphon major, CRM 60; Community Bureau of Reference, Commission of the European Communities). The measure was carried out in duplicate for each sample type.
Scanning electron microscopy (SEM). SEM images were collected using a Philips SEM 515 with a tension of 15 kV. The wet samples were glued on carbon tape, dried in a desiccator, and coated with 20 nm of gold prior image them.

Results
The pristine byssus was digested and its metal composition was analyzed (Table S2). The matrix had a 0.5 ± 0.2 wt.% metal content, the major metals were Al, Cu, Fe, and Zn. After the de-metalation process, the metal content of the byssus matrix was 0.122 ± 0.006 wt.% (80 ± 10% less than the pristine), and the main metals were Al and Fe. The concentration of other element did not change considering the standard deviation of the measures.
Five different loading concentrations were tested for each metal ion studied: 0.01, 0.1, 0.5, 1, and 10 mM. The amount of adsorbed metal ion was measured as the difference in concentration between the loading and the final solution, after incubating the byssus matrix for 72 h at room temperature (25 °C). The results of the metal ion removal at pH 7 are reported in Table 1 and Figure S1, and the discoloration of three colored metal ion solutions, i.e. Mn 2+ , Fe 3+ , and Cu 2+ , due to the byssus' metal ion adsorption are shown in Fig. 2. The system was studied in a buffered solution using bis-TRIS, a non-chelating buffer, at pH 7. For Fe 3+ a bicine buffer was used since iron (III) hydroxide precipitated in bis-TRIS. The analysis of the metal ion solution after the byssus www.nature.com/scientificreports/ treatment revealed the presence of trace metal ions released from the byssus matrix, most of them were largely below 10 ppb. Only three metal ions were released in an interfering quantity from the matrix, Al (~ 100 µM), Fe (~ 40 µM), and Zn (~ 10 µM). For this reason, reliable data for Al 3+ and for low loading concentrations of Fe 3+ and Zn 2+ were not obtained. It was also difficult to determine the uptake of Mn 2+ at a concentration lower than 0.1 mM, this was due to the partial overlap of the emission bands of Fe and Mn that made difficult to discriminate the two contributes and detect the small concentration difference between final and loading solution. The byssus matrix was also tested at pH 4 since in this condition His residues are protonated and no more able to chelate metals 41 . Four metal ions were tested at this pH: Ni 2+ and Cu 2+ ions that have a high affinity to His, V 3+ that has a high affinity to DOPA, and Mn 2+ that has no preferential binding sites. The results of the metal ion removal at pH 4 are reported in Table 2 and Figure S2. Also at pH 4 desorption of metal ions from the byssus matrix was observed, as occurred at pH 7. Al ( ~ 50 µM), Fe (~ 50 µM), and Zn (~ 5 µM) were released from the matrix. The absorption data were interpolated using three different isotherm models: Langmuir, Freundlich, and Dubinin-Radushkevich 42,43 . The statistics of the interpolations are reported in Table 3, Figure S3, and S4. The byssus matrix uptake of different metal ions (reported in mol·g -1 from the 10 mM solutions) was compared using a T-test (ν ≥ 2, p = 0.05). This test showed no significant differences among Cd 2+ , Co 2+ , Ni 2+ , Zn 2+ and Cu 2+ at pH 7. A significant difference was not detected among Mn 2+ at pH 4 and 7, and V 3+ at pH 4. The metal ions absorbed into the byssus were also quantified in few representative matrices. Before the measurements, the byssus matrices were carefully washed with deionized water and digested in acid. The data (Table S3) show that not all the metal ions removed by the solution were in the byssus, suggesting that a fraction desorbed during the washing process.
SEM images were acquired to visualize if the presence of metal ion induces morphological changes on the surface of the byssus matrix. The surfaces of byssus matrices treated with the 10 mM solution and with the 0.5 mM solution were observed. Two cases were discriminated. Case 1: no morphological differences were observed on the surfaces of the samples treated at the two concentrations examined. Case 2: the byssus thread treated using the 10 mM solution showed a smooth surface where globular structures were not visible, differently from threads treated with 0.5 mM metal ion solutions. The metal ions inducing this surface morphology alteration (case 2) at pH 7 were Co 2+ , Cu 2+ , Fe 3+ , and Mn 2+ , while Cd 2+ , Ni 2+ , V 3+ , and Zn 2+ were in case 1. At pH 4 no metal was able to induce this change (case 1). Representative images of the byssus surface were reported in Fig. 3, while the complete set of images is reported in Figures S5 and S6. After the treatment, a decrease in the color intensity of the solution is observed. The byssus exposed to Fe 3+ assumed a darker coloration. www.nature.com/scientificreports/

Discussion
The use of environmentally undesired wastes to purify polluted industrial and environmental waters from metal ions is a field of applied research of growing interest 3,[13][14][15]17 . Thus, in an era environmentally wise, several materials have been tested with a mechanistic approach, even if they do not uptake metal ions in their natural function. The byssus is a complex matrix, which makes difficult to understand the mechanism of molecular uptake of metal ions, evolved to chelate metal ions, a key element in its natural function. The main goal of this research is to test the byssus capability to absorb a wide range of metal ions in two model conditions of pH. To verify this capability, while trying to understand the mechanism of metal ion uptake, simple chemical systems were used.
In them only one metal ion was present, allowing us to discriminate the byssus capability uptake, even if this is an important limitation compared to real cases.
Although is reasonable to attribute the metal ions uptake to the high content of DOPA and His residues, our results clearly suggest that many other binding sites are present, which could be no specific. The choice of the metal ions to be used in this research was addressed by a double objective, to prove the capability to uptake toxic and hazardous ions (e.g. Cd 2+ ) and to understand the uptake mechanism due to specific interactions with functional groups (e.g. V 3+ with DOPA 40 ). Moreover, to simplify the chemical system, only metal chloride salts dissolved in buffers that avoided the formation of precipitates were used. All the solutions at the start and the end Table 3. Calculated parameters from the fitting of the adsorption isotherms with different models. The R 2 values from the isotherm model(s) that indicated the best fitting are highlighted in bold. (-) the values of K L and q m were not reported since negative and with no chemical-physical meaning.

Langmuir
Freundlich Dubinin-Radushkevich R 2 K L (mg -1 ) q m (mg g -1 ) R 2 K F (mg g -1 ) n R 2 K D (mol 2 kJ -2 ) q D (mg g -1 )  www.nature.com/scientificreports/ of each adsorption experiment were clear and no insoluble phase was observed. The presence of soluble species was confirmed by a tentative of metal ion speciation performed using the software Visual MINTEQ 44 . It revealed that the metal ions at pH 7 were present as soluble aqueous species, complex ions with chloride 45 , bis-TRIS 46 and bicine 47 . At pH 4, Cu 2+ , Mn 2+ , and Ni 2+ were mainly present as aqueous ions. V 3+ has a complex chemistry 48 and formed oxyanion complexes at both pH values 49 . However, this ion was still used since it is a proxy for DOPA chelation 40 . Despite the importance of the metal ion speciation, the complexity of the sorption matrix due to its richness of functional groups and different potential adsorption sites makes difficult to understand what metal ion species was adsorbed. This is a weakness from the point of view of understanding the adsorption mechanism, but it represents the strength of this matrix in terms of versatility and efficiency. The utilization of byssus threads for metal ion uptake required a starting de-metalation process to produce metal ion free binding sites. This process, which reduced of 80 ± 10% of the metal ion content, did not affect the molecular composition, unless at least not in the case of side reactions involving DOPA cross-linking, as already reported 34,40 . After this simple procedure a low amounts of Al 3+ from the aluminum rich sediments 50 glued to the plaques, and Fe 3+ and Zn 2+ not completely removed from the byssus, desorbed during the uptake experiments, as reported in the result section.
Despite this apparent limitation, the capability of metal ion uptake by the byssus matrix (e.g.: 22 mg·g −1 for Cd 2+ , equal to 2 wt.% at pH 7) was higher than many other waste biomasses 3,17 . Similar efficiencies were reported using the waste of jatropha fruit 51 , and the natural biomass collected from an irrigation pond 52 , but these biomasses did not show the versatility of the byssus in terms of pH conditions and metal ion variety.
In the following discussion, the uptake of the different metal ions at the two pH conditions is analyzed. This discussion is necessary mainly mechanistic, due to the complexity of the matrix that does not allow determining uniquely the chemistry of the binding sites for the different metal ions. Cd 2+ , Co 2+ , and Zn 2+ were adsorbed by the byssus (as mol·g −1 ) from solution at pH 7 with a similar slope of adsorbed metal ion vs. loading concentration, such slope was dissimilar and higher for Ni 2+ (Fig. 4). Their adsorption isotherms fitted slightly better a Freundlich isotherm model (Table 3), which describes heterogeneous binding sites, than a Langmuir one, which requires homogeneous mono binding sites. In Freundlich model K F is a constant indicating the adsorption capacity (mg·L −1 ) and the empirical parameter n represents the adsorption energy of the sites 53 . The Langmuir isotherm can be parametrized by dimensionless constant called separation factor (R L ) that can be calculated by the equation: R L = 1/(1 + K L ·C 0 ), where C 0 (mg·L −1 ) is the initial metal concentration and K L (L·mg −1 ) is the Langmuir constant. The value of R L indicates if the adsorption is unfavourable (R L > 1), linear and totally reversible (R L = 1), favourable (0 < R L < 1), or irreversible (R L = 0) 54,55 . In the Freundlich isotherm n value for the Cd 2+ , Co 2+ , Ni 2+ , and Zn 2+ was below 1. This indicated an increasing adsorption with the concentration in solution and was associated with low values of K F . The fitting with the Langmiur model produced a realistic value of R L only for Co 2+ , while for Cd 2+ , Ni 2+ , and Zn 2+ R L indicated an unreal desorption process. We can infer that the complex byssus matrix could have diverse binding sites only for Co 2+29,56 , which have to be mono and energetically equivalent at pH 7. The R L value of Co 2+ was 0.98 and 0.06 in the 0.01 mM and 10 mM solution, respectively, indicating a change of www.nature.com/scientificreports/ ion adsorption type from linear to irreversible. In addition, the low R L value at high initial Co 2+ concentrations indicated a favorable adsorption in these conditions, as deducted from the Freundlich isotherm 54 . It was also observed that increasing Co 2+ concentration the byssus surface morphology changed, suggesting an effect on the byssus surface structure. At pH 4 His residues were completely protonated and no more able to bind metal ions 41 . In this condition, Cu 2+ and Ni 2+ were used as model metal ions with high His affinity and diverse chemistry 57 . V 3+ was tested as a metal ion representative of those with a high affinity for DOPA residues 40 , and naturally absent in the byssus. Mn 2+ as metal ion with no specific interaction with both binding sites.
The adsorption data showed that for metal ion concentrations lower than 10 mM the byssus matrix has a higher uptake at pH 4 than pH 7. In those conditions, the lowest percentage of adsorption was 75%, despite the fact that His-based binding sites were protonated. This higher uptake might be due to the speciation of metal ions (i.e. presence of oxyanions) and to the effect of protonation of the physical properties of the byssus matrix. Indeed, collagen-like molecules of the byssus thread can protonate and swell in acid solutions 58 . This physical state should favor metal ion diffusion and made accessible specific binding sites. This agreed with the observation that adsorption isotherms were described better by the Langmuir model than the Freundlich one for Cu 2+ , Ni 2+ , and Mn 2+ . As discussed for Co 2+ , Cu 2+ and Ni 2+ showed R L values close to unity in 0.01 mM solutions (0.92 and 0.89, respectively) and close to zero in 10 mM solutions (0.008 and 0.024, respectively). These data indicated a more favorable adsorption for Cu 2+ than Ni 2+ . Anyway, both the metal ions showed a shift from a linear to an almost irreversible adsorption with an increasing of the initial metal concentration. We could speculate that these metal ions have a stronger binding to the matrix at higher initial concentrations. On the other hand, the Freundlich isotherm fitting gave n > 1 ( Table 3), meaning that a lower adsorption was observed by increasing the concentration of the metal solution. This might be explained by a saturation of the binding sites.
Cu 2+ , Cd 2+ , Co 2+ , Ni 2+ , and Zn 2+ showed many similarities, as the same metal adsorption (in mol·g −1 ) at pH 7. Despite that at pH 4 the byssus showed a higher uptake for Cu 2+ than the Ni 2+ and that at pH 7 it showed a slope of adsorbed metal ion (mol·g −1 ) vs. loading concentration (mM) (Fig. 4) higher than that of Ni 2+ . These similarities could indicate that Cu 2+ have access to the same binding sites of Ni 2+ , Cd 2+ , Co 2+ , and Zn 2+ , with a higher affinity or a higher number of non-pH dependent binding sites. The latter could be amino and sulphur ligands or, considering the high amount of Cu 2+ adsorbed, non-specific binding sites.
As already discussed Cu 2+ , Cd 2+ , Co 2+ , Ni 2+ , and Zn 2+ adsorption isotherms were described better by a Freundlich model at pH 7 and by a Langmuir model at pH 4. These metals showed a positive correlation between adsorption and initial metal ion concentration at pH 7 and an opposite trend at pH 4 (probably due to the saturation of the binding sites). For Cu 2+ the latter indication of energetically equivalent mono-binding site (Langmiur model) available at pH 4 was associated to a conservation in the final morphology of the byssus. In fact, Cu 2+ treated byssus showed a surface morphology related to loading concentration as in case 2 (changed) at pH 7 and as in case 1 (unchanged) at pH 4 ( Fig. 3, Figure S5 and S6). This microscopy observation apparently does not fit with the hypothesis of the collagen swelling at pH 4, with the consequent change of morphology upon metal ion uptake. However, it has to be considered that the collagen-like structures of the byssus are located in the bulk of the byssal threads and their swelling should not affect the surface texture and morphology of the byssus.
Ion manganese (II), instead, showed no difference in the uptake using the 10 mM solution in both pHs investigated, relying probably on non-pH-dependent binding sites. Moreover, as observed for other metal ions tested, an increase in the Mn 2+ uptake percentage at lower concentrations was observed at pH 4. Interestingly, only at pH 7 the 10 mM Mn 2+ solution induced a change on the morphology of the surface of the matrix. Moreover, at pH 4 the adsorption of Mn 2+ could be described with both Langmuir and Freundlich isotherm models. This suggested the anchoring on different binding sites, probably made accessible by the collagen swelling, with inversion of concentration dependent behavior as observed for Cu 2+ and Ni 2+ .
The V 3+ uptake was higher at pH 4 than pH 7 (16 mg·g −1 and 1 mg·g −1 , respectively) as expected from its speciation. The adsorption isotherms could be described by both Dubinin-Radushkevich model (DR) and Freundlich model. At pH 7, n was close to 1 showing an almost linearity of adsorption with the starting metal concentration. At pH 4, the fitting with the DR model allowed to calculate the average free energy of adsorption [E = (2·K D ) −1/2 ], 0.828 kJ·mol −159 . This fitting implies an adsorption mechanism with a Gaussian distribution of energy onto a heterogeneous surface. The higher uptake at pH 4 can be due, among other factors, to a better availability of the DOPA residues. Dopamine, the de-carboxylated form of DOPA, is known to undergo oxidation at pH > 5 60 , preventing the anchoring of metal ions. We suggest that this occurred during the byssus matrix preparation, where the pH is neutral at the end, and the 72 h of incubation at pH 7. That oxidation would explain the relatively low V 3+ and Fe 3+ uptake compared to other metal ions. On the contrary, at a pH below 5 dopamine converts to trihydroxylate form, which coordinates metal ions 61 . This change in the redox state of the DOPA can explain why the matrix was able to chelate over 10 times more V 3+ at pH 4 than pH 7. Differently to the other metals tested, V 3+ did not fit the Langmuir model at pH 4. Thus, we could assume that this metal ion did not rely on the same non-pH dependent binding sites of the other metal ions tested. The fit of a DR model might be another clue of the DOPA affinity for the vanadium. In fact, the Gaussian distribution of energy on heterogeneous binding sites might arise from the diverse chemical structures that DOPA assume from its oxidation/reduction cycle, when different crosslinking and hydroxylation reactions might take place creating a range of different catechol-based binding sites 59 .
An applicative question could be whether to reuse the byssus after metal adsorption or to dispose it. The byssus can be regenerated by chemical treatment that chelate metals associated to the protonation of its metal ion Scientific Reports | (2020) 10:22222 | https://doi.org/10.1038/s41598-020-79253-7 www.nature.com/scientificreports/ chelating functional groups. On the other hand, the byssus is a proteic material that can be pyrolyzed, leaving a residue rich of metal ions. Thus, it can represent also a tool to recover metal ions from polluted waters.

Conclusion
In this study we explored the potential of byssus, a waste material from mussels, to uptake metal ions using simple model chemical systems. This allowed us to evaluate the capability of the byssus matrix to uptake different metal ions in two pH conditions, even if the chemical systems do not represent the complexity of environmental matrices. The byssus is an excellent metal ion-anchoring matrix, especially at pH 4, being suitable and promising for applications, as both disposable and reusable cheap matrix. The results gained in this research, combined with the reported adsorbing capability for aromatic dyes 29,62 , make the byssus a very promising cheap matrix to treat wastewaters from tanneries and paper industries, which usually combine dyes and metal ions to get colored compounds.
Received: 2 April 2020; Accepted: 21 October 2020 Scientific Reports | (2020) 10:22222 | https://doi.org/10.1038/s41598-020-79253-7 www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.