Eichhornia crassipes: a Powerful Bio-indicator for Water Pollution by Emerging Pollutants

Eichhornia crassipes is well known as an invasive aquatic plant. It is also used very effectively in phytoremediation, particularly for the rhizofiltration of effluents contaminated by heavy metals. In this article, we show that it is also an excellent bioindicator of water polluted by worrying organic pollutants such as endocrine disruptors and neonicotinoids. As a proof of concept, di-n-hexylphthalate, pentabromodiphenyl ether, nitenpyram, acetamiprid and bis (3-tert-butyl-4-hydroxy-6-methylphenyl) sulfide were clearly identified by UHPLC-HRMS or GC-MS in the root system of E. crassipes after a short period of exposure. These results open up new perspectives for the remediation of water polluted by alarming organic pollutants.


Results and Discussion
The bioaccumulation capacities of organic pollutants by E. crassipes was first investigated. To do so we selected 4 different types of organic pollutants to be tested: di-n-hexyl phthalate 1, pentabromodiphenyl ether 2, nitenpyram 3 and acetamiprid 4. Di-n-hexyl phthalate 1 is a monomer required for the production of plastics and pentabromodiphenyl ether 2 is a flame retardant; both are endocrine disruptors. Nitenpyram 3 and acetamiprid 4 are neonicotinoids, i.e. neurotoxic insecticides. All the pollutants studied are listed in Fig. 1. From a structural point of view, 1, 2, 3 and 4 are compounds which exhibit aromatic cycles. They present also some similarities with the phthalates identified by Shanab et al. 19 in E. crassipes.
The contaminated solutions were prepared by dissolving each organic pollutant in water or in a mixture of water/isopropanol for the compounds insoluble in aqueous medium (compounds 1 and 2). The accumulation abilities of E. crassipes were investigated in two parts: • an entire living plant of E. crassipes was placed in the contaminated solution. Shoots and roots were harvested separately and were treated with methanol. The organic compounds were extracted in the methanol layer and were analyzed by UHPLC-HRMS (for compounds 3 and 4) and GC-MS (for compounds 1 and 2). • dried roots in powder form were introduced in the contaminated solution. After stirring, the powder was separated by filtration and treated with methanol. The organic compounds were extracted in the methanol layer and were analysed by GC-MS.
The methanol layers were compared with the commercial references diluted in methanol (compounds 1 to 4). As a control sample, water hyacinth plants were not subjected to any contaminated solutions but simply grew in water and were analysed as described previously by GC-MS. study of the control plants -not subjected to contamination. Analysis of crude root extracts. No organic compounds were detected in the methanol layer of extracted shoots. However, the composition of root extracts was very intriguing and interesting. Two compounds were clearly identified after comparison with commercial samples: 4-methoxy benzaldehyde (2.61 min of retention time) and bis (3-tert-butyl-4-hydroxy-6-m ethylphenyl) sulfide 5 (6.13 min of retention time) (Fig. 2).  www.nature.com/scientificreports www.nature.com/scientificreports/ It was not surprising to find 4-Methoxy benzaldehyde in the plant root extracts. The biosynthesis of 4-Methoxy benzaldehyde is derived from the degradation products of lignin catalysed by the H 2 O 2 -activated peroxidase 21 . 4-Methoxy benzaldehyde probably corresponds to a degradation product of lignin, which is abundant in roots.
The presence of thioether 5 was more surprising. Its structure corresponds to santanox or santox, which is an antioxidant used for the manufacture of tarpaulin used to line artificial ponds. Its presence in roots rather unexpected was then investigated. A few milligrams of tarpaulin consisting of vulcanizable EPDM rubber mixtures were cut and dissolved in isopropanol at 40 °C for 24 hours and analysed by GC-MS (Fig. 3). A peak having a similar retention time (6.13 min) could be attributable to santanox from the root extracts. Thioether 5, used for the manufacture of the tarpaulin lining the water ponds used to grow water hyacinths, was thus found in their roots. This control experiment reflects the capability of E. crassipes to extract pollutants derived from aromatic structures.
Analysis of root extract fractions after chromatography. In addition to 4-Methoxy benzaldehyde and thioether 5, the GC-MS spectrum of the crude root extracts revealed a complex mixture of compounds between 3.5 and 5.5 min of retention time. The root extracts were fractionned using flash chromatography to isolate possible organic pollutants from the complex mixture. The GC-MS analysis of the non-vegetal compound, isolated from fraction 1, showed a retention time and a mass spectrum identical to diisooctyl phthalate 6. The mass spectrum presented the fragments at 279 and 149 characteristic of phthalate structure (ions mono isooctyl phthalate and C 8 H 4 O 3 respectively) (Fig. 4). No non-vegetal compounds could be clearly isolated from the other fractions of the chromatography.
Due to the unexpected presence of diisooctyl phthalate 6, these analyses were repeated one year later (June 2018) and yielded identical results; the presence of 6 was once again detected. This result can be related to the study of Shanab et al. 19 who identified as well 4 phthalates in the plant extracts. However the water hyacinth plants grow in opposite conditions: Shanab et al. collected wild plants in River Nile (El-Zomor canal, Egypt) while we used plants that were cultivated in greenhouses under controlled conditions (Occitanie, France). Two explanations can be advanced: Shanab et al. assumed that phthalates could derive from secondary metabolites from E. crassipes. However, phthalates cannot be biosynthesized from any known biosynthetic pathways in plants. Therefore we assume that E. crassipes exhibits an unusual capacity of phytoaccumulating phthalates, that derive from plastic and are widely spread in environment.

study of the root extracts -subjected to contamination. Analysis of phytoaccumulation capabilities
of living E. crassipes by rhizofiltration. In the order to confirm the hypothesis that E. crassipes phytoaccumulates organic pollutants, E. crassipes plants were placed in contaminated solutions with the 6 emerging pollutants: di-n-hexyl phthalate 1, pentabromodiphenyl ether 2, nitenpyram 3 and acetamiprid 4. After treatment, previously described, the methanolic root extracts were analyzed by UHPLC-HRMS (and GC-MS for compounds 1 and 2).
Di-n-hexyl phthalate 1 was clearly observed by GC-MS at 4.88 min of retention time in the methanolic root extracts (Fig. 5). The characteristic fragment of phthalate structure (ion C 8 H 4 O 3 4 ) at 148.84 was observed. A co-injection of the extract and commercial di-n-hexyl phthalate was consistent with the interpretation. Compared to the control extract ( Fig. 1), the presence of phthalate 1 confirms the capacity of E. crassipes to phytoaccumulate phthalate, which is in agreement with our initial hypothesis. www.nature.com/scientificreports www.nature.com/scientificreports/ Pentabromodiphenyl ether 2 was clearly observed by GC-MS at 6.32 min of retention time in the methanolic root extracts (Fig. 6). The mass spectrum is consistent with the typical isotope distribution of five bromine atoms calculated with isotope software M, M + 2, M + 4, M + 6, M + 8, M + 10, and the pattern 1: 5: 10: 10: 5: 1. Similarly to compound 1, E. crassipes was able to bioconcentrate pentabromodiphenyl ether 2.
Structurally, 2 is close to 5 (6.13 min), since 2 is a diphenyl ether and 5 is a thiodiphenyl ether. The phytoaccumulation of 2 is not surprising since 5 is systematically present as shown in the control sample (Fig. 2).
The phytoaccumulation studies of 3 and 4 neonicotinoid derivatives could not be performed by GC-MS due to their chemical properties. The extraction of nitenpyram 3 by E. crassipes roots was studied by UHPLC-HRMS, a high performance mass spectrometer. This machine is a Q-TOF (Quadripoliare-Time Of Flight) with a High-Resolution (DeltaM/M > 48 000) and a high accuracy (<3 mDa) and allows us to determine the Elemental Composition of a Mass (see Methods).
The identification of the nitenpyram 3 in the root extracts was achieved by comparing HRMS in mode TOF MS ES + with the standard compound. 5446 structures were evaluated with 48 results within limits, and only one compound with a 99.94 percent degree of confidence could be attributed to 3 (see Methods Fig. 10).
The identification of the acetamiprid 4 in the root extracts was achieved by comparing HRMS in mode TOF MS ES + with the standard compound. 372 structures were evaluated and only one compound with a 100.00 percent degree of confidence could be attributed to 4 (see Methods section). The mass spectra is consistent with the typical isotope distribution of chlorine, its pattern 6: 1 and the parity of mass (M + 1 = 223.0750 in ESI positive) due to the four nitrogen atoms.
Comparison with the control not subjected to contamination. The rhizofiltration studies on the six pollutants found in the root extracts of E. crassipes allowed us to develop specific and accurate analytical tools for their  www.nature.com/scientificreports www.nature.com/scientificreports/ detection in non-enriched environment. We revisited the composition of the root extracts of our plant control, E. crassipes that was not subjected to any contaminated solutions.
Surprisingly all pollutants (1, 2, 3 and 4) were identified in the control root extracts after fractionation by liquid chromatography. If the phytoaccumulation of the phthalate 1 and thioether 5 in the control samples has been explained, the presence of neonicotinoids in E. crassipes not subjected to a solution enriched in neonicotinoids was unexpected.
Investigations near the production site revealed that neonicotinoids have been used for non-aquatic crops located close to the ponds. This observation can explain the presence of 3 and 4 in untreated roots. This demonstrates how E. crassipes is able to concentrate neonicotinoids and behaves as a powerful bioindicator. However, there was no direct local explanation for the presence of compound 2, the flame retardant. It was probably due to an indirect contamination of the aquatic plant culture.

Analysis of phytoaccumulation capabilities of dried roots of E. crassipes by biosorption.
Encouraged by all these results, we decided to study the biosorption ability of E. crassipes roots in dehydrated powder form. Dihexyl phthalate 1 and pentabromodiphenyl ether 2 were selected as model pollutants for the study. Dried roots in powder form were introduced in the contaminated solutions of 1 and 2. After stirring, the powder was separated by filtration and treated with methanol. The organic compounds were extracted in the methanol layer and were analyzed by GC-MS. Both dihexyl phthalate 1 (4.88 min) and pentabromodiphenyl ether 2 (6.32 min) were clearly identified by GC-MS: 1 was identified by its characteristic fragment at 148.84 (ion C 8 H 4 O 3 ) in MS spectrum and 2 by its isotopic signature characteristic of pentabrominated products presenting mono and dibrominated fragments in MS spectrum (Fig. 7).
The kinetic monitoring of dihexyl phthalate 1 presence (4.87 min) was demonstrative of the rapid biosorption by the dried roots. The follow-up was performed by SIM (Selected ion monitoring) scanning mode such that only the characteristic fragment of phthalate structure (ion C 8 H 4 O 3 ) at 148.84 was detected (Fig. 8).
Considering the capacities of root powder to biosorb phthalates, an accurate quantification of biosorption was carried out. The quantity of phthalate extracted was assessed by subtracting the residual amount in the solution from the initial amount. Due to the hydrophobic nature of phthalates, the adsorption of phthalates on the walls of the reactor was controlled. No phthalate adsorption on the walls of the reactor could be detected, which demonstrated that the decrease of phthalates in solution was attributable to the biosorption by root powder.
The factor of biosorption, the measurement range, accuracy, relative standard deviation RSD were considered to quantify the biosorption (See experimental section). After validation of the calibration using commercial solutions and dodecane as the internal standard, it was possible to establish that 28% of the phthalates (mean of four replicates) were extracted from the concentrated solution (10 mg/L), which corresponded to 0.7 mg of phthalates per 39 mg of dried roots.
Discussion on the mechanism of biosorption by E. crassipes. We showed that living E. crassipes is able to phytoaccumulate the pollutants 1, 2, 3 and 4 in natural conditions and that E. crassipes roots in dehydrated powder form presented biosorption capacities for 1 and 2. Compounds 1, 2, 3 and 4 share a common structural feature, aromatic cycles. Considering the lipophilic character of the aromatic structure, we hypothesized an affinity of the organic pollutants for phenylphenalen structures to phospholipids naturally present in the plant.
We carried out a total lipid extraction of roots, following the procedure described by Folch et al. 22 , to examine the role of lipophilic structures in the biosorption procedure. A new biosorption of di-n-hexylphthalate 1 of roots lacking phospholipids was performed, but the results were not modified. www.nature.com/scientificreports www.nature.com/scientificreports/ The organic pollutants must have an affinity to another component of water hyacinth roots. FT-IR is an ideal procedure to analyze biomass, entirely exploring the structural information of samples. ATR-FT-IR was used to analyze the E. crassipes roots powder components (Fig. 9).
The presence of aromatic compounds and carboxylic acids was unambiguously confirmed by Infrared Spectroscopy (IR) analysis using Fourier Transform Infrared Spectrometry (FT-IR).
After a mathematical fitting, (see Supporting information), we could observe aromatic compounds, which show absorptions in the region around 1607, 1594 cm −1 due to carbon-carbon stretching vibrations in the aromatic ring. In turn, carboxylate units are characterized by four intense bands at 1657, 1653, 1650 and 1631 cm −1 corresponding the carbonyl stretch C=O from the COO − group. We determined the area ratio of the C=O bond vibration band and the aromatic ring vibration bands.
The fitting of the ATR-FT-IR spectrum clearly shows that root powder from E. crassipes is rich in carboxylate functions. This material is suitable for the biosorption of metallic elements based on the formation of metal carboxylate salts 8 . The metal salts can interact with the Lewis base sites of the pollutants, which are di-carboxylic ester, 1,2-dibromophenyl, pyridine-vinylidenediamine, pyridine-acetamidine, phenol for 1, 2, 3, 4 and 5 respectively, and can participate to its biosorption.
The fitting of the ATR-FT-IR spectrum also shows the presence of aromatic compounds that can derive from lignin, a well-known component of roots. The aromatic structure of lignin possesses a lipophilic part, which could interact and retain the aromatic pollutants.
We can anticipate that the association of lignin and metals in roots favour the biosorption of aromatic organic pollutants.

Conclusion
In this article, we showed for the first time that E. crassipes can bioconcentrate emerging organic pollutants such as endocrine disruptors and neonicotinoids. The aquatic plant can be used as a bioindicator of pollution, but also as a natural remediation tool provided it is introduced in a controlled environment and in powder form. Using the root system in powder form presents three interesting solutions to important environmental problems: • it can prevent the dispersion of emerging organic pollutants into the environment.

Methods
General. All chemicals were purchased from Alfa Aesar or Sigma-Aldrich.
Gas chromatography-mass spectrometry (GC-MS) analyses were carried out by using a TRACE TM 1300 apparatus equipped with an ISQ TM QD mass spectrometer detector (Thermo Fisher Scientific). The following chromatographic system was used for analytical experiments: GC-MS (Thermo TG-5SilMS, 20 m, 0.18 µm) with hydrogen as carrier gas [80 °C (0.5 min), followed by linear gradient from 80 to 280 °C (5 min)] at a flow rate of 0.8 mL/min. Calibration curve. The identification of the samples was based on comparison of its retention time with those of the standard compound.
A straight line of the calibration was determined with standard samples. The calibration curve was constructed by plotting peak areas of 6 corresponding concentrations 0.045, 0.083, 0.197, 0.354, 0.451, 0.522 μgmol.ml −1 .
Three injections of each solution were used to obtain the curve. The calibration curve was linear (R 2 = 0.9969). The limits of detection and quantification were established at 0.028 μgmol.ml −1 and 0.045 μgmol.ml −1 respectively. The relative standard deviations (RSD) of the results obtained were satisfactory (0.6/0.9/1.1/2.8/8.2/16.8%).
The calibration curve was performed two times and the experiment of quantification was tested four times.
High resolution electrospray ionization mass spectrometry (HR-ESI-MS) were acquired in positive or negative ion mode. The conditions were the following: capillary voltage 3000 V; cone voltage 20 V, dry gas temperature 140 °C, desolvatation temperature = 450 °C, dry gas flow, 1000 L.h-1 and nitrogen as nebulizer gas, Pressure = 6.5 bars.
Analytic UHPLC was performed using a column Kinetex C18 (100 mm × 2.1 mm, 1.7 µm), at 30 °C. 1 ng/µl Leucine Enkephalin was used as standard for internal calibration. The elution gradient is described in the Table 1. www.nature.com/scientificreports www.nature.com/scientificreports/      Table 2. Results of mathematical fitting of ATR-FT-IR. p = Peak Position; a = coefficient; w = width at half height; g = contribution of Gaussian function (between Min = 0 and Max = 1); Formula = Gaussian and Lorentzian functions combinaison; Area = Area of the fit peak.