Effects of mixtures containing chlordecone and a dechlorinated by-product on hydra regeneration capacity: use of experimental design to evaluate the toxicity risk associated with remediation programs in long-lasting polluted areas

Xavier Moreau Aix-Marseille-University: Aix-Marseille Universite Magalie Claeys-Bruno Aix-Marseille-University: Aix-Marseille Universite Jean-Pascal Andraud Aix-Marseille-University: Aix-Marseille Universite Hervé Macarie IRD: Institut de recherche pour le developpement Daniel E Martínez Pomona College Maxime Robin Aix-Marseille-University: Aix-Marseille Universite Michelle Sergent Aix-Marseille-University: Aix-Marseille Universite Laetitia De Jong (  laetitia.dejong-moreau@imbe.fr ) Aix-Marseille-University: Aix-Marseille Universite https://orcid.org/0000-0001-8798-3131


Introduction
Chlordecone (C 10 Cl 10 O, CLD) is a persistent organochlorine insecticide which was formerly manufactured in the USA under the trade name of Kepone®. One of the main uses of CLD has been for controlling black banana weevil populations. For this speci c use, CLD was formulated as a very ne powder (also called dust by the pesticide industry) diluted to 5% by weight in a mineral matrix which was applied manually directly to the soil surface in circles around the banana pseudostem (Clostre et al. 2014a;Epstein 1978).
Due to the mismanagement of the production process, most of the workers making CLD in the United States were poisoned by it, which resulted in the closure of the factory and the complete ban on production, commercialization and use of CLD in USA in 1977 (Cannon et al. 1978; Dawson et al. 1979). To cope with shortage of CLD due to this closure, a consortium of French banana producers succeeded to relaunch the production of CLD in 1982 under a new brand name, Curlone®, with the objective of using it mostly in the banana plantations of French West Indies (FWI). Overall, either as Kepone® or Curlone®, it is estimated that around 300 tons or 1/6 of the CLD ever produced in the world have been used in FWI from 1972 to 1993 (Le Déaut and Procaccia 2009; Devault et al. 2016). Due to its physicochemical properties, especially a strong a nity for organic matter, CLD is highly persistent in soils. In the FWI, it has been estimated that depending on the soil composition, 60 to 700 years would be necessary before the almost complete CLD disappearance (Cabidoche et al. 2009). Although CLD exhibits weak water solubility (Dawson et al. 1979), under the action of rain, through leaching, runoff, and erosion, it is exported from the soil to surface and underground waters. Chemical Reduction (ISCR) process, which consists of adding Fe 0 to the soil, is the only one to have been tested with some success under real environmental conditions in FWI, up to the plot scale ). The ultimate fate of CLD during this process remains to be established but it is known that it will generate a series of dechlorination products which conserved the bishomocubane structure of CLD but have lost 1 to 7 chlorine atoms replaced by hydrogen atoms and known as hydrochlordecone. In absence of complete mineralization, there is always the risk that the degradation byproducts may have similar toxicity or be even more toxic than the parent compounds (Benoit et al. 2017;Dol ng et al. 2012). To address this question, the genotoxicity, mutagenicity and proangiogenic properties of 3 major CLD derivatives formed during the ISCR process, and which have lost 1, 3, and 4 chlorine atoms have been compared to that of CLD, showing that like CLD, they remained non-genotoxic and non-mutagens, while their proangiogenic properties tested both in vitro and in vivo were greatly reduced (Alibrahim et al. 2020; were tested individually but during the ISCR process, they were present simultaneously in soil although in proportions that may uctuate over time (e.g., Mouvet et al. 2020). In the same way, these byproducts will be transported to the other environmental compartments as a mixture in water (Ollivier et al. 2020a, b). Due to possible additive, synergistic or potentiation effects such mixtures containing CLD and CLD byproducts may have a greater toxic effect than CLD alone as observed for other compounds (e.g. Brown In this context, the present study has been conducted to answer a fundamental question: are mixtures containing chlordecone and dechlorination byproducts, which may be formed during remediation processes, less or more toxic when the concentration of the dechlorinated product increases in the mixtures? To answer this question and in contrast with classical methods applied in ecotoxicology, where, for example, one concentration is modi ed at a time keeping the others constant in a mixture, we proposed an experimental design that allows: (i) to vary simultaneously all concentrations of the compounds of the mixture, (ii) to limit the number of experiments, (iii) to detect the in uence of the compounds in the mixtures, (iv) to produce a predictive model of the toxicity of the mixtures and (v) to considerably reduce the amount of waste generated during the experimental procedure. The bioassays have consisted in evaluating the regeneration capacity of a freshwater hydra clone exposed, in 96 well polycarbonate microplates, to environmental concentrations of CLD in presence of corresponding concentrations of the trihydrochlordecone (CLD-3Cl) generated by the ISCR process. The CLD concentrations were in the range of those measured in FWI surface freshwaters (Lesueur-Jannoyer et al. 2016, Mottes et al. 2020). Hydra was chosen as a freshwater animal model for the study because it has been already used in a recent work to assess the effect of long-term exposure of CLD on several biological markers such as stress gene expression, oxidative stress endpoints and reproduction rate (Colpaert et al. 2020). The pertinence of this choice has been reinforced recently by the nding that Hydra species are ubiquitous in FWI surface waters (Macarie and Martínez, unpublished results). Preliminary to our investigation, the taxonomic position of the hydra clone used in the bioassays was de nitively established.

Hydra clone culture conditions
The population of the hydra clone (strain IMBE1) was raised in TES buffer (0.1 mM; pH 7) (Sigma-Aldrich, Saint Quentin-Fallavier, France), at 20°C ± 0.1°C and under a 12/12h light-dark cycle according to the procedure of De Jong et al. (2016) that has been adapted from Trottier et al. (1997). The polyps were fed every three to four days with 24 h-hatched Artemia sp. nauplii ad libitum. All specimens used in this experiment raised from asexual reproduction and belong to the same clone. DNA Extraction, Sequencing and Phylogenetic analysis DNA from the IMBE1 hydra clone was extracted using the QIAGEN DNAeasy Blood & Tissue Kit following manufacturer's protocol ( nal elution was 150 µl). Ribosomal DNA (the ITS region comprising 18S partial; ITS1, 5.4 S complete, ITS2 and 28S partial) was ampli ed by PCR (94°C for 3 min; and 35 cycles of denaturing at 94°C 30 sec, annealing at 55°C 60 sec, extension at 72°C for 90 sec). The ampli ed fragment was veri ed by 1% agarose electrophoresis and cloned into a PROMEGA pGEM-T Easy vector (ligation was performed overnight at 4°C). Sequencing was carried out by Euro ns Genomics.
The IMBE1 ITS sequence was rst aligned with CLUSTAL Omega to hydra sequences from all 4 major clades of hydra: Viridissima, Braueri, Oligactis and Vulgaris (Martínez et al. 2010). Once the speci c hydra group was determined, a second alignment was performed to investigate the geographic origin of the IMBE1 strain. This second alignment included a total of 76 hydra ITS sequences, 7 of which were used as outgroups in the phylogenetic analyses. A maximum likelihood analysis of the phylogenetic relationships was implemented using Garli 2.0 (Zwickl 2006). For this analysis, the invariant, and hence uninformative, 5.4S portion of the sequence was not included in the alignment. The best likelihood model was selected

Toxicity evaluation
In exposure experiments (see below in the "Experimental design" section), healthy and budding polyps of the hydra clone of similar size were randomly collected among a dense healthy population under a stereomicroscope. One day before the beginning of the experiment, the population was fed with living 24 h-hatched Artemia sp. nauplii. After two transverse cuts of the polyps retained for exposure, the central section of the hydra body, also called gastric region, was conserved for regeneration experiments ( Figure  1).
Each section of the gastric region was carefully placed in a well of a 96 well-microplate either in 250 µL of TES buffer (0.1 mM; pH 7) for controls or in 250 µL of a mixture solution of CLD and CLD-3Cl at concentrations de ned according to an experimental design (see the following section). Then, the 96 wellmicroplates were placed into a thermo-regulated incubator at 20°C ± 0.1°C under a 12/12 h light-dark cycle for 96 h. Control conditions, performed in triplicate, were needed to ascertain the good health of the population of the hydra clone used in exposure experiments. In control conditions, the gastric region regenerates basal and oral regions following normal progressive steps according to time. The capacity of the hydra clone regeneration has been evaluated using a score scale ranging from 0 to 10. The score scale used here has been built with different score scales proposed in previous studies ( reference score -2 0 to 1 = death; 2 to 5.9 = extremely toxic; 6 to 6.9 = very toxic; 7 to 7.9 = toxic; 8 to 8.9 = low toxicity, viable polyp; 9 to 10 = no toxicity, healthy polyp.

Experimental design
Two quantitative factors (U 1 , U 2 ) were studied: CLD and CLD-3Cl concentrations and the aim of this study was to model the effects of mixtures of CLD and CLD-3Cl. In other words, the idea was to obtain the value of hydra regeneration score over  Table 2. The same ranges of molar concentrations have been used for CLD-3Cl (Table 2). Within the de ned experimental domain, the possibility of different mixtures represents an in nite number of experiments. Therefore, to determine the effects of all mixtures, we have chosen a mathematic tool allowing to collect the maximum of information using a minimum of experiments. This tool is an empirical model which can represent the hydra regeneration score (noted Y in the equation below) in the domain of interest i.e. for all combinations of CLD and CLD-3Cl concentrations. We have postulated a quadratic polynomial model since it allows the consideration of curvature effects and to understand the in uence of each quantitative factor (U, Table 2). This model based on two predictor variables, CLD and CLD-3Cl concentrations, with coded variables X 1 and X 2 was written as follows: To estimate the coe cients of the mathematical model, experiments were carefully chosen by the experimental design. Indeed, the quality of the coe cient estimation and the quality of the prevision only  Eighteen different experimental conditions were performed, and each experimental condition has been at least repeated in triplicate ( Figure 2). Finally, a total of 72 experiments have been carried out (Table 3).
Chemicals and reagents CLD in the form of a powder with a purity over 97% was a gift from Azur Isotope (Marseille, France). CLD-3Cl was synthesized by Alpha Chimica (Châtenay-Malabry, France) and was also provided as a powder. It had a chromatographic purity over 90% expressed as % of the total area of peaks detected ( Figure 3A). The position of the substitution of the chlorine atoms by hydrogen atoms on the CLD-3Cl carbon chain was unknown with precision. The predominant mode of fragmentation on electronic impact of the compounds with a bishomocubane structure such as CLD-3Cl is known to correspond to the cleavage in half, of their cage forming two pentacyclo-fragments (Dilling and Dilling 1967).

Results
Phylogenetic a nities and geographical origin of the IMBE1 strain  1985). The variation in substitution rates between different sites was Gamma distributed. The maximum likelihood phylogram generated implemented using Garli 2.0 (Zwickl 2006) clearly showed that strain IMBE1 is a Eurasian hydra most likely from Europe ( Figure 4). The bootstrap values calculated by three different methods indicate the topology of the tree is quite robust which adds a high degree of certainty to our conclusion.

Regeneration scores and toxicity scale
In control conditions, values of regeneration scores recorded after 96 hours were of 10 suggesting that the polyp population was healthy at the beginning of exposures. In exposure conditions, among the malformations and damages observed, stunted regeneration, lack of mouth, abnormal tentacles (bulbed, short...) and loss of physiological functions (i.e., osmoregulation) are concerning because they impair nutrition (catching and ingestion of living preys) and lead to polyp death. Therefore, using the mean regeneration scores, the following toxicity scale can be proposed: score 0 to 1 = death; score 2 to 5.9 = extremely toxic; score 6 to 6.9 = very toxic; score 7 to 7.9 = toxic; score 8 to 8.9 = low toxicity, viable polyp; score 9 to 10 = no toxicity, healthy polyp. Only mean regeneration scores below 8 are concerning ( Table   1). The experimental results into the domain of interest are summarized in gure 5.
In In all mixtures with CLD concentrations above 2.04.10 −2 µM (right part of the experimental domain in Figure 5), regeneration scores reaching 8.9 or more were not observed. One mixture has shown a low toxicity and all other ones led to disturbing conditions. Two of them were extremely toxic (mean regeneration scores 5.5 ± 2; 5.0 ± 0.9, experiments 13 and 4) and revealed that these severe toxic effects could be attributed to both compounds at high concentrations.
For the modeling step, we considered the nine points of the composite design (rectangles in gure 2) to calculate the model coe cients using multilinear regression on the coded variables (X 1 and X 2 ). This model was then validated using the four "validation points" (ovals in gure 2) which experimental values have been further compared to the previously calculated ones. Data showed a non-signi cative difference (Student Test) with signi cance values largely above 5% indicating that the model and the experimental values were close. These four "validation points" were thus integrated for the calculation of the coe cients. The model can be written as follows: This mathematical model was used to predict, whatever the proportions of CLD and CLD-3Cl in the mixtures, all the regeneration scores (Y) inside the experimental domain ( Figure 5). However, the mathematical model was not able to predict the stochastic "endocrine disruptor effect" observed at CLD concentration of 2.88.10 −3 µM (experiments 16, 17 and 18; at these experimental conditions low regeneration scores re ecting extremely toxic conditions were observed). Therefore, the 6 points of the matrix that have been added to the experimental design to study the possible stochastic "endocrine disruptor" effect were not used in the calculation of the model (experiments 14, 15 16, 17 and 18; stars in gure 2).
The modeling shows isoscore lines, i.e. different mixtures of CLD and CLD-3Cl leading to the same score of regeneration of H. vulgaris ( Figure 6). The generally vertical direction of the isoscore lines points out the preponderant effect of CLD in the mixtures. The presence of CLD-3Cl in the mixtures has no particular in uence on the regeneration scores i.e. on the toxicity. Furthermore, for the minimal CLD concentrations tested, higher CLD-3Cl concentrations in the mixtures led to better regeneration scores, isoscores 8.5 and 9, re ecting low or no toxicity, respectively. At the highest concentrations of both compounds, the model predicts very disturbing regeneration scores re ecting very toxic conditions. These very toxic conditions were explained by the presence of both compounds in the mixtures.

Discussion
In present study, the classi cation of our hydra IMBE1 clone has been reevaluated to dispel doubt concerning its species name. At the time of writing, this clone has been in breeding for at least the past four decades. In the early 1980's, it was used at the University of Philadelphia (USA) under the name Hydra attenuata in a prescreening test for the detection of substances with teratogenic potential in mammals (e.g. Johnson et al. 1982). The same clone was later used also by Environment Canada, and in Due to the confusion created by Schulze in 1917 around the name « attenuata » which has been used both to describe specimens belonging to the Hydra vulgaris and Hydra circumcincta species (Campbell 1989), the clones used in the previously cited works were either named as Hydra attenuata or Hydra circumcincta. The phylogenetic tree generated for this study clearly showed that the closest relatives of our IMBE1 clone, which had travelled between North America and Europe, belonged to the Hydra vulgaris species and were of European origin suggesting that it had itself the same geographic origin.
The outstanding regeneration capacity of species of Hydra genus has been successfully used in ecotoxicology by several authors to detect the teratogenic potential of numerous chemicals. To our knowledge, the pioneer was Johnson (1980)  all these compounds, Johnson and collaborators found a good correlation between the A/D ratios observed for hydra and mammals where A represents the toxic concentration for adults whereas D represents the toxic concentration affecting development. The hydra assay was sensitive enough to identify among these compounds those capable to pose hazards to development and could be used as predictive of a putative teratogen's hazard potential (A/D ratio) before performing standard mammal assays (Johnson and Gabel 1983). Later, 'gastric sections' have been used as 'arti cial embryo' by Wilby and Tesh (1990). As the assay based on 'gastric sections' was easier to perform than the one using The present study con rms the sensitivity of this bioassay, and therefore rea rmed the need of such investigations for risk assessment with environmental concentrations of xenobiotics. The simple diploblastic organization of these animals only consists in two differentiated tissues. Therefore, cnidarians lack complex endocrine system that can be found in vertebrates and several invertebrates (e.g. Arthropoda). Even if physiological regulation and potential disruption are poorly understood in cnidarians, common vertebrate hormones (e.g., steroids, iodinated organic compounds, neuropeptides, and indolamines) have been identi ed in their tissues and chemical stressors could also impact the physiology of these invertebrates (Tarrant 2005). Keeping in mind that, in contrast to most metazoans, cnidarian cells are not generally organized into organs or systems, with careful consideration, the hydra regeneration assay can be a useful predictor of the potential risk to a developing vertebrate embryo (Bowden et al. 1995). But of course, as pointed out by Tarrant (2005), "care must be taken not to assume processes will be identical in all organisms". In other words, teratogenic effects on 'hydra arti cial embryo' exposed to toxicants could re ect teratogenic effect of mammal embryos but no conclusions could be formulated concerning the involved physiological mechanisms. In the present study, in which we have used 'gastric sections' as 'arti cial embryo', teratogenic effects could be observed in several mixtures. The presence of CLD-3Cl had no particular in uence on the toxicity of the mixtures, except when both CLD and CLD-3Cl were at their highest levels. For these latter conditions, both compounds could explain the high toxicity of the mixture. Such conditions are however unlikely to occur in the environment as the result of a soil remediation process such as the ISCR one. In rst approach, the dechlorination of CLD to generate CLD-3Cl will result in a decrease of CLD concentration in the soil while that of the dechlorination product will increase and so a high concentration of both compounds cannot be present at the same time. In any case, since it will be only possible to apply the remediation processes over a few tens of centimeters of the upper part of the soil, the CLD stock in the lower parts will remain intact. Under such conditions, the work of Ollivier et al (2019, 2020) with column of soils treated in surface by ISCR has clearly shown that the concentrations of CLD-3Cl, but also those of the other dechlorination products will always be lower than CLD in the soil leachates that contaminate the surface and ground waters even if these products are more soluble and so mobile than CLD. Since the ban of CLD eld utilization, CLD concentrations in freshwaters could increase solely if CLD stored in soil matrix is released. In a recent study conducted in FWI, Sabatier and colleagues (Sabatier et al. 2021) demonstrate CLD resurgence due to the widespread use of herbicides containing glyphosate since the late 1990s'. This still current agricultural practice is considered to be responsible of a hitherto unseen rise in soil erosion and downstream of a major release of the stable CLD stored in polluted soils since their ban (Sabatier et al. 2021). The severe toxic effects observed at high CLD and CLD-3Cl concentrations on hydra development supports a special warning of agricultural practices that could remobilize CLD and lead to increasing CLD concentrations in freshwaters. In present study, teratogenic effects on 'hydra arti cial embryo' exposed to the most environmental probable mixtures can be explained by the presence of CLD in the mixture. Indeed, for a given CLD concentration, regeneration scores did not differ when CLD-3Cl concentration increased while scores uctuate between bad to good when CLD concentrations vary for a given CLD-3Cl concentration. Thus, it seems that the presence of CLD-3Cl in the mixtures, at concentrations expected after the application of a remediation process such as the ISCR, has no supplementary deleterious effect on hydra regeneration capacity. Regeneration scores re ecting no sign of toxicity (score up to 9) can only be observed in the mixtures with CLD concentrations equal or below to 2) has led to a return to satisfactory regeneration. Hence, our experimental results demonstrate that the toxic effects could not be linked to a progressive gradient of CLD concentration in the mixtures as impairments of regeneration have been observed at both low and high critical concentrations of CLD. The present results clearly demonstrated that the nonmonotonic concentration-response occurred with these organochlorine mixtures because (1) low or high concentrations could lead to the same deleterious damages and (2) because exposition to low and close to low concentrations could lead to good or bad regeneration capacities. Thus, whereas the teratogenic effect could be explained by the presence of CLD in the mixtures, it was not dependent on an increase in CLD concentrations. Such stochastic phenomenon has been also recently described concerning several biological endpoints in hydra entire polyps under exposure to CLD: reproductive rates, morphological changes, and expression of target stress genes (Colpaert et al. 2020). Therefore, our results con rm the previous observations of Colpaert et al.
(2020) and lead to the same conclusion: biological effects observed after exposure to CLD follow nonmonotonic dose-response curves. The nonmonotonic dose-response effects are di cult to model, and this phenomenon is a challenge in ecotoxicology because, in absence of a suitable mathematical model to estimate the risk, the presence of such compounds in the environment, whatever their concentrations, represent a threat for exposed populations. Our study pointed out the di culty to propose a predictive mathematical model when studying compounds belonging to EDs. Without considering the unpredictable stochastic effects of EDs, our experimental design allows to propose an empirical model that can determine the hydra regeneration score (noted Y in the equation of the model) in the domain of interest i.e. for all combinations of concentrations ( Figure 6). The approach can also provide information on the most active compounds in the mixtures. From the present data, it can be suggested that our approach is useful to study the mixtures containing compounds having concentration-dependent effects.

Conclusion
Our work con rms that hydra regeneration assay is a useful, inexpensive and rapid screen for the evaluation of mixture toxicity. The use of empirical mathematical model offers perspectives in environmental toxicology investigations, as this tool allows to study complex mixtures (with more than two products). Moreover, the quantitative factors (U1, U2, U3, …) could either be only compound concentrations, or compound concentrations and abiotic factors such as temperature or exposure time.
For example, using a similar experimental design, De Jong et al. (1994) have studied the toxicity of methyl mercury and mercury (II) chloride to a brown alga in mixtures containing calcium according to exposure time. In the present study, the values calculated by the model are close to the experimental values and therefore increase con dence in such predictive approach for a better knowledge of mixtures effects. Toxic effects of organochlorine pesticides and byproducts are di cult to analyze with classical ecotoxicity tools and the present experimental design offers possibility to detect the stochastic effects of such compounds even if no adequate modelling could be proposed to predict them. Our experimental design associated with a mathematical model allows determining the most in uential compound in the mixtures (here, CLD). The present work, even if it is limited to a mixture of two products, suggests that the hydrochlordecones generated by remediation processes such as the ISCR do not appear to be more toxic than the CLD itself, and especially that their presence next to the CLD, at least for the concentrations expected to be found in the environment, does not increase the noxiousness compared to the baseline situation where CLD is alone. This is good news for the application of remediation processes unable to mineralize CLD and which will just lead to its transformation into dechlorinated products having conserved the bishomocubane structure. Further studies with more complex mixtures, representing more adequately those generated by such processes, remain however necessary to con rm their harmlessness.

Declarations Ethical Approval
The present manuscript is an original study and was not submitted elsewhere in any form or language but some results were presented at the congress of the Hervé Macarie contributed to the selection of (1) the CLD dechlorinated product to be used for the study and (2) the range of concentrations of CLD and CLD-3Cl which are representative of the French West Indies environments. He also contributed to literature search, reconstruction of IMBE1 hydra clone lab history and to the analysis of the CLD-3Cl mass spectra to identify the possible isomers. Maxime Robin synthesized the CLD standard used in the study. All authors contributed to the analysis of the results. The rst draft of the manuscript was written by Laetitia de Jong and Xavier Moreau and all authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript.

Competing Interests
The authors have no relevant nancial or non-nancial interests to disclose.

Funding
Partial funding for this work was provided by the European Regional Development Fund (FEDER) through the RIVAGE project (MQ0003772-CIRAD) and by an internal grant of IMBE research unit for stimulating cooperation between several of its constitutive teams. JPA was supported by a PhD grant from the ED251 doctoral school on Environmental Sciences of Aix Marseille University.

Availability of data and materials
Not applicable Acknowledgments Partial funding for this work was provided by the European Regional Development Fund (FEDER) through the RIVAGE project (MQ0003772-CIRAD) and by an internal grant of IMBE research unit for stimulating cooperation between several of its constitutive teams. JPA was supported by a PhD grant from the ED251 doctoral school on Environmental Sciences of Aix Marseille University. We thank Stéphane Greff from IMBE, Marseille, France, for his help in the analysis of the CLD-3Cl electronic impact mass spectra and Christian Blaise (Environment Canada), Sophie Pachura (Agestra, France) and Paule Vasseur   regeneration scores could be interpreted as follow: 0 to 1 = death; 2 to 5.9 = extremely toxic (black symbols); 6 to 6.9 =very toxic (black symbols); 7 to 7.9 = toxic (grey symbols); 8 to 8.9 = low toxicity, viable polyp (white symbols); 9 to 10 = no toxicity, healthy polyp (white symbols). The disturbing scores are in white letters. For symbol signi cance, see gure 2 legend Modeling of hydra regeneration scores exposed to mixtures of CLD and CLD-3Cl within the experimental domain showing isoscore lines. Limits of experimental domain are 2.10 -4 µM and 4.10 -2 µM for both compounds. Mean regeneration scores could be interpreted as follow: 0 to 1 = death; 2 to 5.9 = extremely toxic; 6 to 6.9 =very toxic; 7 to 7.9 = toxic; 8 to 8.9 = low toxicity, viable polyp; 9 to 10 = no toxicity, healthy polyp