A New Ex Vivo Model to Evaluate the Hair Protective E ﬀ ect of a Biomimetic Exopolysaccharide against Water Pollution

: As an external appendage, hair is exposed to multiple stresses of di ﬀ erent origins such as particles and gases in air, or heavy metals and chemicals in water. So far, little research has addressed the impact of water pollution on hair. The present study describes a new ex vivo model that allowed us to document the adverse e ﬀ ects of water pollutants on the structure of hair proteins, as well as the protective potential of active cosmetic ingredients derived from a biomimetic exopolysaccharide (EPS). The impact of water pollution was evaluated on hair from a Caucasian donor repeatedly immersed in heavy metal-containing water. Heavy metal retention in and on hair was then quantiﬁed using Inductively Coupled Plasma Spectrometry (ICP / MS). The adverse e ﬀ ects of heavy metals on the internal structure of hair and its prevention by the EPS were assessed through measurement of keratin birefringence. Notably, the method allows the monitoring of the organization of keratin ﬁbers and therefore the initial change on it in order to modulate the global damage in the hair. Results revealed an increasing amount of lead, cadmium and copper, following multiple exposures to polluted water. In parallel, the structure of keratin was also altered with exposures. However, heavy metal-induced keratin ﬁber damage could be prevented in the presence of the tested EPS, avoiding more drastic hair problems, such as lack of shine, or decrease in strength, due to damage accumulation.


Introduction
The condition of a human internal hair fiber, crucial for the physical properties of the hair, is traditionally determined by measuring its mechanical stress-strain characteristics [1] or by evaluating the state of its surface, using scanning electron microscopy or atomic force microscopy technologies [2,3].
Hair fiber is composed of three layers, the cuticle, the cortex and the medulla. The cortex is the main layer of hair fiber and responsible for mechanical property with macrofibrils or intermediate filaments, composed of keratin [4]. Taking into account that more than 85% of the weight of hair comes from keratin [5], it seems reasonable to assume that any alterations in the condition of hair, as documented with the techniques mentioned above, would be due to changes in its major component, keratin [6]. So, to the best of our knowledge, it is crucial to detect the first change in the protein in order to avoid more dramatic change in the hair properties. Keratins belongs to a family of fibrous

Polluted Water Model
To mimic the effect of water pollution on hair, a model of heavy metal-polluted water was developed, based on data obtained from a toxicological evaluation of the river Yamuna, in the Indian city of Agra [24]. As pollution of this river mainly results from human activities (agriculture, industrialization, urbanization), the model is therefore considered as representative of global water pollution.
Polluted water was prepared with dilution of concentrated commercial standards of the elements of interests (Pb, Zn, Cu, Cd, Cr, Ag, Hg and Fe), and addition and dissolution of the proper amount of pure solid sodium carbonate (Na 2 CO 3 ) in distilled water (see Table 1). Once the dissolution is ready, the theoretical concentrations are checked by ICP/MS analysis of the elements present (Pb, Zn, Cu, Cd, Cr, Ag, Hg, Fe and Na). When the concentration was under detection limits of the machine, "<0.01" was indicated.

Analysis of Heavy Metal Levels in Hair, Following Exposure to Polluted Water
Strands of healthy hair from a single Caucasian donor were used. Hair strands were prewashed, rinsed in either distilled water (control) or polluted water and then left to dry. Following 10 immersions into polluted water, representative of 1 or 2 weeks of daily exposure, 50 mg of hair samples was mineralized in nitric acid (sigma) for metal retention quantification using Inductively Coupled Plasma Spectrometry (ICP/MS). Analyses were completed on a XSeries II (Thermo Electron) equipped with a Meinhard nebulizer. The analysis was done with a plasma power at 1450 W; a nebulizer gas flow of 1.016 L/min, the coolant gas flow of 1.016 L/min, 50 sweeps and 10 ms per isotope of dwell time. Retention of exogenous lead, cadmium and copper was documented.

Exposure of Hair to Polluted Water
Strands of healthy hair from the same single Caucasian donor were used to document the impact of polluted water exposure on hair structure. Hair strands were prewashed, then three groups of 41 hair fibers were selected. One group was first immersed for 2 min in a solution containing the EPS (0.1 g/L), massaged for 1 min, then exposed to polluted water for 2 min. The second group was only exposed to polluted water for 2 min. The third group was immersed for 2 min in distilled water but received no further treatment and served as control. All hair strands were next massaged for 30 s. According to the preliminary experiment, we decided to repeat this protocol 30 times, which is necessary to detect the significant impact of polluted water.
For hair structure analysis, for each of the 41 hair fibers, middle sections were cut into 2-cm lengths and mounted on different microscope slides, in Euparal mounting media (n = 1535). For each hair sample, keratin birefringence was measured at 3 distinct regions with a polarimetric scanner (K-PROBE ® polarimetric scanner) and a mean keratin birefringence value was calculated.

Polarization Study of Hair Fibers
Theoretical Background on the Birefringence Birefringent materials have the particularity to have two different refractive indexes, no (along the ordinary axis) and ne (along the extraordinary axis), the birefringence being the dimensionless quantity ∆n = ne -no. When light passes through a birefringent sample, a phase difference θ (rad.) called retardance is produced between orthogonal projections of the light electric field [26], which is proportional to the birefringence ∆n and birefringent thickness e crossed by the light upon reflection.
Retardance θ and birefringence ∆n are linked with the equation: with λ the wavelength. We should emphasize that the "thickness" of hair at a given pixel impacts the retardance value (which is measured) but not the keratin birefringence (that we want to determine). In order to determine the birefringence value, we use a parametrized ovoid hair section model: its birefringence, diameter and shape are determined by a non-linear least square regression of the retardance signal. Thus, a potential hair asymmetry can be handled by the model.
Birefringence is an intrinsic parameter of the sample, depending on its material and structure. Retardance, however, also depends on the sample thickness.
The technique used in this study is based on the emission of a laser light on the sample; its state of polarization can take any value on the Poincaré sphere. The light reflected by a mirror under the sample is collected and the intensities polarized along two orthogonal states of polarization are measured: I and I. The Kmax parameter is then computed The Kmax value is linked to the sample retardance with [27]: The instrument gives a measure of the Kmax value for each pixel of the field of view. Thus, it gives a value of the retardance applied to the light detected by each pixel. Contrarily to crossed polarizers microscopy [28], the retardance value measured by this technique is independent of rotation of the sample optic axis in the image plane, which allows quantitative characterization based on the sample retardance [29].

Theoretical Background on Hair Birefringence Measurement
In hair samples, the sample thickness e is unknown, and is spatially variant in the field of view. As it can be seen in Equation (3), the birefringence ∆n cannot be distinguished from the thickness e using a Kmax measurement, if these two parameters are unknown. Thus, in order to estimate the birefringence ∆n, we must make an assumption about the sample thickness. For that purpose, we use a hair thickness model, based on an ovoid approximation of the hair section.
Let (x,y) be a plane perpendicular to the hair growth direction (z) with its origin on the hair center, and y be the direction of propagation of light. At a given position z along the hair, let R be the hair radius and r the medulla radius, r < R, which describes an area at the center of the hair where the material has a low birefringence. We also introduce an "ovoid" factor k, describing a potential asymmetry of the hair section. Then the birefringent thickness e(x) crossed by the light upon reflection is:

Hair Fiber Birefringence Measurement
Hair fiber birefringence was measure with the K-PROBE ® POLARIMETRIC SCANNER using the XPolar ® technology, from Kamax Innovative System. The measurement is based on a contactless technique and the signal does not require any staining, which avoids any sample degradation. The wavelength of the source is 780 nm, the magnification is 400×, the image size is 800 × 600 pixels, each pixel covers an object area of about 0.25 µm 2 .
An example of raw data measured by XPolar ® is represented in Figure 1a, giving the Kmax values measured. The periodicity of the Kmax signal can be understood by looking at Equation (3), where one can notice the sin 2 (x) function, which has a kπ periodicity (k integer). The physical interpretation of this effect is that the retardance between orthogonal projections of the light electric field is evolving periodically between its minimum (for πe∆n/λ = kπ) and its maximum (for πe∆n/λ = kπ/2) depending on the material thickness e (and birefringence ∆n) crossed by the light at the considered pixel.
= kπ/2) depending on the material thickness e (and birefringence Δn) crossed by the light at the considered pixel.
In order to obtain the estimation of the hair birefringence, we compute the averaged transverse variations of the measured Kmax parameter on 500 lines of the Kmax image (called in the following Km(x)), as plotted in Figure 1c, black). The Km(x) curve will be the input data used for the estimation of the global hair birefringence. By substitution of e in Equation (3) with its expression defined in Equation (4), one can notice that we have a theoretical expression of the Kmax parameter, in the form of four input parameters (hair radius R, medulla radius r, ovoid index k and keratin birefringence Δn), in equation (5).
The theoretical Kmax variations will be designated by Kth(x,p), with p a vector containing the unknown parameters (R, r, k and Δn) (an example is plotted in Figure 1c, red curve). Estimation is performed by iterative assumptions of the four values of the parameters of the model Kth(x,p), in order to decrease the difference (least mean square difference) between the model Kth(x,p) and the experimental data Km(x). When the difference is minimized, the modelized Kmax values are the closest to the experimental Kmax values and thus we have an estimation of the four parameters, including the birefringence Δn of the considered hair. An example of fitted Kth(x,p) is represented in Figure 1c, red curve). Besides, by replication of Kth(x,p) in the z direction, we construct a simulated image (Figure 1b) which can be compared with the raw Kmax image (Figure 1a).  In order to obtain the estimation of the hair birefringence, we compute the averaged transverse variations of the measured Kmax parameter on 500 lines of the Kmax image (called in the following Km(x)), as plotted in Figure 1c, black). The Km(x) curve will be the input data used for the estimation of the global hair birefringence. By substitution of e in Equation (3) with its expression defined in Equation (4), one can notice that we have a theoretical expression of the Kmax parameter, in the form of four input parameters (hair radius R, medulla radius r, ovoid index k and keratin birefringence ∆n), in Equation (5).
The theoretical Kmax variations will be designated by Kth(x,p), with p a vector containing the unknown parameters (R, r, k and ∆n) (an example is plotted in Figure 1c, red curve). Estimation is performed by iterative assumptions of the four values of the parameters of the model Kth(x,p), Cosmetics 2020, 7, 78 6 of 12 in order to decrease the difference (least mean square difference) between the model Kth(x,p) and the experimental data Km(x). When the difference is minimized, the modelized Kmax values are the closest to the experimental Kmax values and thus we have an estimation of the four parameters, including the birefringence ∆n of the considered hair. An example of fitted Kth(x,p) is represented in Figure 1c, red curve). Besides, by replication of Kth(x,p) in the z direction, we construct a simulated image (Figure 1b) which can be compared with the raw Kmax image (Figure 1a).

Statistical Analysis
To overcome effects of potential uncontrolled variables, we randomly build 3 sets of hairs coming from the same hair strand, which will experience the 3 different conditions. To search for differences between populations that may be influenced by 1 or more uncontrolled variables, the obtained data and percentage variations between two conditions were submitted to a one-sided student t-test for independent samples with unequal variances. The statistical significance value was considered at p < 0.05.

Heavy Metals Retention in Hair
Following 10 cycles of exposure to water, loaded or not with heavy metals, metal retention on the surface and inside hair fibers was quantified using Inductively Coupled Plasma Spectrometry (ICP/MS). Exposure to polluted water significantly increased hair concentrations of copper, cadmium and lead by 120%, 846% and 214% respectively, compared to control (Figure 2). Such significant increases validate our polluted water model to study the impact of water pollution on the structure of hair. The presence of zinc and calcium was also recorded, however, substantial quantities of these minerals are naturally found in hair fibers.

Statistical Analysis
To overcome effects of potential uncontrolled variables, we randomly build 3 sets of hairs coming from the same hair strand, which will experience the 3 different conditions. To search for differences between populations that may be influenced by 1 or more uncontrolled variables, the obtained data and percentage variations between two conditions were submitted to a one-sided student t-test for independent samples with unequal variances. The statistical significance value was considered at p < 0.05.

Heavy Metals Retention in Hair
Following 10 cycles of exposure to water, loaded or not with heavy metals, metal retention on the surface and inside hair fibers was quantified using Inductively Coupled Plasma Spectrometry (ICP/MS). Exposure to polluted water significantly increased hair concentrations of copper, cadmium and lead by 120%, 846% and 214% respectively, compared to control (Figure 2). Such significant increases validate our polluted water model to study the impact of water pollution on the structure of hair. The presence of zinc and calcium was also recorded, however, substantial quantities of these minerals are naturally found in hair fibers. Hair is essentially made of keratin, a fibrous structural protein of major importance for the mechanical properties and appearance of hair. Within hair, keratin fibers are held together through various types of bonds (hydrogen, salt and disulfide bonds). Besides their structural role, these bonds also provide soft spots through which the environment can affect the hair. For instance, human hair has a strongly negative surface charge mainly due to the aniomic molecules in their composition, such as glutamic acid and aspartic acid. Taking into account that glutamic acid and aspartic acid make up for 25.9% of the total amino acid composition of keratin [30], and that keratin is the main protein responsible of hair physical properties [6], it is clear that any interaction with this components will have an impact in the hair properties. As water pollutants are essentially mono-or di-valent cations, they have a strong attraction for the negative charge of hair, and thanks to their size can penetrate inside the fiber to interact with the internal bonds responsible for the hair structure.
Here we show that hair repeatedly exposed to polluted water significantly retains cationic heavy metals such as copper, cadmium and lead. The retention can be explained by interactions forming Hair is essentially made of keratin, a fibrous structural protein of major importance for the mechanical properties and appearance of hair. Within hair, keratin fibers are held together through various types of bonds (hydrogen, salt and disulfide bonds). Besides their structural role, these bonds also provide soft spots through which the environment can affect the hair. For instance, human hair has a strongly negative surface charge mainly due to the aniomic molecules in their composition, such as glutamic acid and aspartic acid. Taking into account that glutamic acid and aspartic acid make up for 25.9% of the total amino acid composition of keratin [30], and that keratin is the main protein responsible of hair physical properties [6], it is clear that any interaction with this components will have an impact in the hair properties. As water pollutants are essentially mono-or di-valent cations, they have a strong attraction for the negative charge of hair, and thanks to their size can penetrate inside the fiber to interact with the internal bonds responsible for the hair structure.
Here we show that hair repeatedly exposed to polluted water significantly retains cationic heavy metals such as copper, cadmium and lead. The retention can be explained by interactions forming ionic bonds between hair and pollutants. Once heavy metals have penetrated within the hair, they bind to keratin with the potential to affect its structure and therefore their functions.

Impact of Polluted Water and Protective Potential of the Biomimetic EPS
The exopolysaccharide used in this study was chosen for its chemical property. Indeed, it presents high molecular weight which confers him filmogenic property. Furthermore, with its composition, rich in uronic acid groups and its negative potential zeta (−16.4 mV), the product has proven its efficiency to chelate heavy metals ( Table 2). Table 2 shows the decrease in heavy metals in an aqueous solution after contact with the EPS. The impact of heavy metal retention on the structure of hair and the protective potential of EPS were documented at the molecular level through the measurement of changes in keratin birefringence, using K-Probe ® polarimetric scanner technology associated with the XPolar ® imagery system. The determination at the molecular level allow us to determine the initial change in the keratin structure. Figure 3 displays the representative experimental images resulting from birefringence analysis, representative of the average retardance of their respective hair set (bottom panels) and the corresponding fitted theoretical composition retardance of hairs obtained with the theoretical model (top panels) in various conditions. A typical result for the control group is shown in the left panels, for the polluted water-challenged group in the middle panels, and for the EPS-pre-treated group in the right panels. Clearly, immersion in polluted water changes the hair birefringence response compared to control, revealing that the internal structure of hair has been modified by the challenge. However, pre-treating the hair with the biomimetic EPS (0.1 g/L) negates the effect of water pollution and restores hair response to light seen in control, attesting of the protective effect of the EPS on global keratin fiber structure.
Despite the major visual impact of these images, they are only tools through which hair birefringence values can be obtained, in order to measure the impact of polluted water on proteins structure. The average birefringence values calculated from these images for each hair fiber is shown in Table 3 (average of at least three points measurement of measure from the same hair fiber) and reviewed in Figure 4 with standard deviation.
(top panels) in various conditions. A typical result for the control group is shown in the left panels, for the polluted water-challenged group in the middle panels, and for the EPS-pre-treated group in the right panels. Clearly, immersion in polluted water changes the hair birefringence response compared to control, revealing that the internal structure of hair has been modified by the challenge. However, pre-treating the hair with the biomimetic EPS (0.1 g/L) negates the effect of water pollution and restores hair response to light seen in control, attesting of the protective effect of the EPS on global keratin fiber structure.

Unprotected Protected (EPS 0.1g/L)
Cosmetics 2020, 7, x FOR PEER REVIEW 8 of 12 Figure 3. Representative photography of hair section analyzed with K-PROBE ® polarimetric scanner. Kmax experimental images (top panels) and the corresponding fitted image (bottom panels) obtained for a hair diameter of approximately 60 µm, following immersion in heavy metal-containing water without (middle panels) and with EPS pre-treatment (right panels), compared to control (left panels).
Despite the major visual impact of these images, they are only tools through which hair birefringence values can be obtained, in order to measure the impact of polluted water on proteins structure. The average birefringence values calculated from these images for each hair fiber is shown in Table 3 (average of at least three points measurement of measure from the same hair fiber) and reviewed in Figure 4 with standard deviation. What does a change in birefringence really mean? In order to understand the meaning of our experiment, we can provide an example taking into account that birefringence is defined as the difference between the two refractive indices of an optically anisotropic material. Imagine that we have two elements with the same diameter and composition but with a very different density or structure. The difference of their two refractive indices will be quite distinct and therefore so would be the difference in their birefringence. On the other hand, if these two elements have the same density and structure, light will be refracted in the same way and thus the difference of their refractive indices and their birefringence would be quite identical. Now, if we were to submit one of the twin materials to an attack capable of affecting its density and structure, then the refractive indices difference and birefringence of the two would become divergent.   Since the structure of hair in the control group (with no heavy metal challenge) is considered intact, the birefringence values obtained for this group can thus be taken as the basal levels for all hair samples. Following 30 cycles of exposure to polluted water without protection, the mean birefringence values decreased by 9% compared to control, with a p-value < 0.05, meaning that the impact of polluted water on hair fiber is statistically significant. However, with EPS (0.1 g/L) pre-treatment of hair fibers birefringence is not significantly different from the control but significantly different to placebo, by 11% (p-value < 0.05). The decrease in birefringence seen when hair fibers are attacked by water pollutants (compared to control) reflects a loss of structural integrity in keratin. If this damage is not stopped, it should have definite impact on the macromolecular properties of hair, for example dry, damaged, and less stretchy hair. Fortunately, pre-treating hair with the biomimetic EPS (0.1 g/L) restores birefringence values back to basal levels, avoiding initial damage of keratin structure. EPS effectively and fully protects hair keratin from water pollution-induced structural changes.
After confirming in our model, the attraction of heavy metal by hair fiber, we demonstrate the impact of heavy metals content in hair on keratin structure, the main protein of hair fiber. To further study heavy metals penetration and EPS effect, an analysis of heavy metals inside the hair fiber could give supplemented information, that we could not have with ICP/MS analysis. However, to the best of our knowledge, some inherent characteristics of the biomimetic EPS may play a role in hair protection. For one, the high molecular weight and the composition of the EPS would confer protective filmogenic properties that create a physical barrier on hair against pollutants. Furthermore, our chelating property of the EPS may help by blocking adhesion of heavy metals to hair. Both properties synergize to prevent heavy metals from damaging the structure of keratin and should help to maintain the physical properties of hair.

Conclusions
As pollution levels rise around the world, consumers are increasingly concerned about the negative effects it may have on their body and hair. Several scientific studies have already documented harmful effects of air pollution on skin and hair but little research has been done so far on the impact of water pollution on these structures.
This present study was aimed at developing a platform to assess the effects of water pollution on the internal structure of hair and test the protective potential of a cosmetic active ingredient derived from a biomimetic EPS. Results confirmed that water pollutants can indeed penetrate the hair, causing damages to keratin, its fibrous structural protein. Such detrimental effects of water pollution were prevented by protective filmogenic and chelating properties of EPS. Because keratin is the main structural protein of the hair, it is reasonable to suggest that this protection would also enable the preservation of the physical properties of the hair such as strength and elasticity against repetitive and daily detrimental exposure.