Prevention of lipid loss from hair by surface and internal modification

Surfactants during routine washing have a tremendous effect on lipid loss from hair. This study aims to understand the loss of lipids from hair upon contact with surfactants and develop a way to prevent the lipid loss. The change in lipid levels depends on the relative hydrophobicity of the lipid. We herein propose that the change in lipid levels can be protected by two modifications. In the case of fatty acids and cholesterol (group A), the concentration difference between virgin hair versus surface modified hair with highly charged polymer was 22 to 32% higher after washing with surfactants while the loss of squalene and wax esters (group B) in response to surfactants still occurred even after the surface modification. In the hair treated by internal modification with the carbodiimide reaction, 52.0 to 81.3% more lipids in group B were prevented than in the untreated hair. Finally, different types of lipids were successfully protected by surface and internal modifications from the surfactant treatment. This study will be the basis for understanding the mechanisms by which surfactants damage the lipid barrier of tissues including hair and for establishing strategies to defend the barrier.


Extraction of lipids
The following procedure applies to the extraction of components covalently and noncovalently bound to hair. A 15 cm hair swatch was cut to a length of less than 1 mm with a mass of 50 mg, and the hair residue was transferred to a 15 mL vial filled with 3 mL of extraction solvents. Lipid was extracted merely by isolation with a methylation solvent. The methylation solvent was prepared by the following process. The solvent was mixed so that the ratio of benzene to methanol was 1:3, and 4 g H 2 SO 4 was then added to 100 mL of the mixture. After blending for 3 h, 10 g Na 2 SO 4 was added, and the blending was continued for another 3 h. The vial was tightly covered with a Teflon cap, and the vial was incubated at 80 °C for 4 d. The standard configuration for the same solvent was placed in the same vial to observe whether the solvent evaporated. The vial was cooled to room temperature, and 2 mL n-hexane was added to the solvent. The fraction was shaken vigorously. After several minutes for complete separation and clarification, the system was separated into two layers. The supernatant containing the extracted lipids was collected. This treatment was repeated three times, and the supernatant of three consecutive collections was evaporated in a cooler to a final volume of 2 mL.

Extraction of noncovalently bound lipids
Hair residue was prepared in the same way for covalently and noncovalently bound lipids. A 15 cm hair swatch was cut to a length of less than 1 mm with a mass of 250 mg, and the hair residue was transferred to a 50 mL vial filled with extraction solvents. Each 250 mg sample of hair was homogenized in a 30 mL solvent with a 2:1 chloroform:methanol mixture by sonication for 3 h.
The hair residues were suspended for 9 h, and the solvent containing the extracted lipids was decanted to a new vial. Subsequently, a new solvent (30 mL) of a 1:1 chloroform:methanol mixture was added to the hair residue, sonicated for 3 h and suspended for 9 h. After decanting S3 the solvent to the vial, a new solvent with a 1:2 mixture of chloroform:methanol was added to the hair residue and suspended for 9 h after sonication for 3 h. After the third solvent was collected in a vial, the final solvent containing a 18:9:1 mixture of chloroform:methanol:water was added to the hair residue, sonicated for 3 h and suspended for 18 h. The solvent was decanted to one vial, and the hair lipids were containted in the final 120 mL solvent. The supernatant of four consecutive collections was evaporated in a cooler (JeioTech, hx-03, Korea) to a final volume of 3 mL.

Analysis of lipid concentrations
For the gas chromatography/mass spectrometry (GC/MS), samples were analyzed on an Agilent (7890A GC System, US) gas chromatograph coupled to an Agilent detector (5975C MSD System, US). 1 μL of each sample was injected into the gas chromatograph inlet. The injector temperature was maintained at 230 °C during the analysis while the ion source temperature and interface temperature were 250 and 300 °C, respectively. Separation was performed on an HP-5ms UI column (30 m × 0.25 mm × 0.25 μm) using helium as the carrier gas. It was programmed to maintain an initial column temperature of 80 °C for 1 min, then change to 320 °C at a rate of 6 °C min -1 and hold for 20 min at this temperature. Injection was performed using a single tapered 4 mm glass inlet liner packed with silanized glass wool in a programmed split mode. Before injection, the carrier gas flow was set to 9 mL min -1 with a split value of 5:1. After injection, the split value was set to 5:1, and the carrier gas flow was set to 1 mL min -1 .
The mass spectrometer was operated in an electron impact mode at an ionization voltage of 70 eV. Mass spectra in the full scan mode were recorded in a mass range of 50-500 amu. Selected (v/v) was added. Then, 1 mL of 1% acetonitrile was added. The reaction mixture was capped, mixed with a vortex mixer and stored at 80°C for 1 h. After cooling, the final solution was filtered and stored at room temperature until HPLC analysis.
Matrix-assisted laser desorption ionization-tandem time of flight mass spectrometry (MALDI-TOF/TOF MS) analysis was performed in the positive ion reflectron mode using an ultrafleXtreme system (Bruker Daltonics). The samples were analyzed as described previously 1 .

Raman mapping of cross-sectioned hair
Because melanin in black hairs can affect Raman signals, European white hair (Jehyun trading, Korea) was used for Raman analysis. White hair was placed at -120 °C and cut with an ultramicrotome system (Leica EM FC7) at a cutting speed of 0.2 mm s -1 . Hair was cut on one side because a variation in signal can be caused by variations of sample thickness. The use of tape (3M Scotch Cat. 810D) for embedding improved the quality of cross-sections. Images were S5 recorded with a MZ6 stereo microscope (Leica). Scanning electron microscope (SEM) images of hair were obtained using a Hitachi S-4800 instrument (Hitachi, Japan).
Raman imaging combining spectral and spatial information at the same time was performed with a commercial apparatus (LabRAM HR, HORIBA Jobin-Yvon Inc., Japan). A semiconductor diode near-infrared laser operating at λ = 1064 nm was used as the excitation source and delivered a laser power of 2.5 mW measured at the cross-sectioned hair. The Raman images were obtained using a Raman point mapping scan. Light from the illuminated spot was collected by an optical lens and sent through a monochromator with an optical filter to block Raleigh scattering.
The Raman signal was focused through a 500 μm wide slit and dispersed by a diffraction grating of 1800 grooves per millimeter onto a deep depletion InGaAs detector (Horiba) of a size of 30 × 30 pixels with 1 µm × 1 µm area. The exposure time per pixel was 40 s, and the scan was repeated 10 times. The distribution of Raman spectra was analyzed using a software program (Labspec 6.5.2.5).

Determination of the ζ-potential
The ζ-potential of treated hair was determined using the streaming potential (SurPASS3, Anton Paar, Germany). A 0.5 g bundle of treated hair was placed in a cylindrical cell without cutting.
Measurements were conducted in 1 mmol L −1 KCl as a supporting electrolyte. All measurements are performed with three samples for the hair bundle.
To determine the ζ-potential of the hair surface, the streaming current coefficient Δ was measured at 299 K and pH 5.8. The ζ-potential was calculated from the electrophoretic mobility measured using the Helmholtz-Smoluchowski (equ. 1).
where κ, the electrical conductivity, is 0.025 S m -1 , and η and ε×ε 0 are the viscosity and dielectric permittivity of the electrolyte solution are 8.79×10 -4 Pa·s and 78, respectively.
Since the Ag/AgCl electrode is affected by the cationic polymer, the non-zero voltage offset at zero pressure was due to the polymers in the hair sample (Fig. 2a). The streaming current coefficient of positively charged hair was higher than that observed in the control ( Fig. 2A). The measured streaming current coefficient of hair changed almost linearly at pressures ranging from 2.0 to 4.5 mbar. PQ10B, PQ10C and PQ10D were found to provide appropriate cationic properties. The averaged ζ-potential was determined to be -16.3 mV for the control using (equ. 1).
A hair has anionic surface due to its low isoelectric point, 2 and the hair treated with a cleansing solution containing general cationic polymer also had negatively charged surface even in repetitive treatments (data not shown). Surprisingly, the averaged ζ-potential was determined to be 17.1 mV, 25.1 mV and 31.6 mV for the PQ10B, PQ10C and PQ10D, respectively.

Viscous modulus
To understand viscous modulus and ζ-potential pattern after coating the hair surface with PQs, surface tension, γ was measured by drop shape analyzer, DSA-100 (Krüss, Germany) using pendant drop method. 4 The surface viscosity, μ was obtained from equ. 2.
where A is interfacial volume dilatation, and f, the frequency for deformation of droplet of 0.5% PQs, was 0.1 Hz. All experiments were averaged with repetitive (at least three) cycles.

Dye tagging and fluorescence images of hair
To perform the fluorescence quenching experiments, carboxytetramethylrhodamine (TAMRA) dye was added to a 5% aqueous solution of polylysine. Then, the TAMRA-labeled polylysine was

Micellar size of SLES
Critical micelle concentration (cmc) of SLES was determined by a tensiometer (K100C, KRUSS, Germany), and the cmc as measured was 0.428 mM. The size of micelle in the SLES solution (c=3 mM) was monitored by dynamic light scattering method using a zetasizer (ZEN3600, Malvern Instruments, UK).
Supplementary Table 1. The amounts of lipids from the extraction of hair (μg/g hair). The ± sign indicates the standard deviation. The analyses from (a) to (e) were performed by GC/MS. The hair lipid in (a) was extracted successively with 30 mL portions of 2:1, 1:1, and 1:2 ratios of CHCl 3 to CH 3 OH, and the hair residue was homogenized with a 18:9:1 mixture of CHCl 3 :CH 3 OH:H 2 O. Those steps were performed with sonication for 3 hours each, and suspension was performed between different solvent mixtures for 9 hours. In (b), the extraction by the fourth solvent that has water was not performed. (c) was performed by shaking instead of sonication. All samples were extracted with 250 mg of hair, but (d) was performed with 900 mg of hair. (e) was performed for 4 days. In the case of (e), sonication was performed for 3 hours each, and the sample was suspended for 21 hours each. The lipid extractions of (f) and (g) were performed in the same way as was performed for (a), but the sample was analyzed by GC/FID and LC/UV, respectively. The analytical results are the average from 6 samples. This section aims to determine the method for extracting the maximum amount of lipid. Here, we present a comparison to define the best method whereby a high concentration of lipids can be extracted in a 2 day process ( Supplementary Table 1 (a)). The amounts of lipids obtained by various extraction methods are shown in Supplementary Table 1. Additionally, the lipids are divided into 2 groups based on the assumed hydrophobicity.
To determine the composition of solvents, hair mass, period of extraction, and analytical method, the amounts of lipid were determined in repeated analyses. The mixture of chloroform and methanol is conventionally used as the solvent for lipid extraction in histological studies of hair. [6][7][8][9] Different mixtures with varying ratios of chloroform to methanol, which are used to effectively extract lipids that have different polarities, are required to achieve the maximum yield for the extraction. Therefore, 250 mg of hair residue was sonicated in the chloroform:methanol series of 2:1, 1:1, and 1:2 for 3 hours each with suspension for 9 hours. The recent method of Wu et al. 7 maximized the extraction quantity via CHCl 3 /CH 3 OH = 2:1, but the study extracted only the squalene fraction from the total lipids. To increase the extraction of polar lipids, water was added in the last procedures. When water was excluded from the solvent for extraction in Supplementary Table 1 (b), we obtained dominantly less group A lipids in repeated experiments. However, the group B lipids mostly did not show a difference even when water was not added to the solvent. Finally, the results indicate that three extractions and one subsequent extraction with water were sufficient for solvent extraction of hair lipids.

S9
Extraction was performed by shaking, as indicated in Supplementary Table 1 (c). Fewer lipids were extractable, and the standard deviation (SD) was larger for the group A lipids. Considering the empirical results, we decided to use sonication during extraction.
Additionally, the lipid was extracted from 50 mg and 900 mg of hair. In the case of 50 mg, wax esters had a significantly higher SD due to its lower concentration. The amounts of extracted lipid increased for 900 mg of hair, as indicated in Supplementary Table 1 (d). However, the amounts of lipid obtained from 250 mg and 900 mg of hair mostly show no significant difference. This finding shows that an excess of hair does not guarantee a higher concentration (per gram of hair) of extracted lipids.
Very similar results were found for 2-day extraction and 4-day extraction, as indicated in Supplementary Table 1 (e). The period of extraction slightly increased the extraction efficiency of lipids such as myristic acid and palmitic acid, but the amount of stearic acid and squalene slightly decreased in the 4-day extraction. Finally, in this study, we chose the short extraction time (2 days) to minimize oxidative decomposition and to perform experiments economically. Many studies performed lipid extraction for different periods. The method of Duvel 8 required too many exchanges for dissolving solvent (total of 11 times). The method of Takahashi 9 was rapid but used methylation, which cannot distinguish the noncovalently bound part from the total lipids. The method of Masukawa 6 was performed for 4 days. Here, a 2-day process has been established with the optimal conditions for lipid extraction.
Next, we selected the analytical tool to determine the lipid concentration. GC/FID had a good ability to quantify the amount of lipids and provide successful analysis with low interference from other substances. [10][11] For this study, GC/FID was inadequate because it was not sensitive enough to detect small amounts of lipids like stearic acid (Supplementary Fig. 3). Additionally, GC/FID did not distinguish tetracosane from 18-MEA due to the peak overlap ( Supplementary Fig. 3). Many studies used liquid chromatography (LC) to analyze hair lipids 6,8,10,[12][13] ; however, HPLC has a limitation when used to analyze cholesterol and 18-MEA. 14 LC/UV cannot analyze hydrocarbons and wax esters, and it is difficult to prepare samples due to derivatization. Finally, GC/MS was employed for analyzing the concentration of lipids. At the end, the concentrations of squalene and cholesterol determined in this study were larger than the 700 μg/g hair and the 410 μg/g hair values reported in other studies. 6,12 Supplementary An XPS study confirmed that even after shampooing one time, an appreciable amount of noncovalently bound lipid was removed to a depth of 3 nm. 15 One may think that the lipid loss determined by GC/MS is attributable to the peripheral lipid on the hair surface. To remove the lipids on the surface, hair was washed with SLES and hexane. To remove lipids contaminating the surface, both SLES and the n-hexane were used for washing. 7,12,16 The data for comparable washing of virgin hair with SLES and hexane are shown in Supplementary Table 4. To remove peripheral lipids on the surface of hair, 1 g of as received hair was homogenized with 100 mL of 10% SLES for 1 min and washed for 2 min. Then, the hair was incubated in 100 mL of hexane and successively washed for 2 min (Supplementary Table  4(d)). The samples in Supplementary Table 4(b) and (c) were washed with only SLES and hexane, respectively. GC/MS analysis was conducted on an extracted sample from each of the washing subjects. All data in Supplementary Table 4 are the concentrations determined from the hair, but the hexane 5 min b data point (e) is the concentration from the solvent extraction that used hexane for 5 min a (d).
The amount of squalene was 943 μg/g hair in as received virgin hair (Supplementary Table 4(a)) and 874 μg/g hair in hair washed with SLES. However, the amounts of C14 and C16 did not decrease from Supplementary Table 4(a) to (b). Washing with SLES one time (b) conserves more lipids than washing with hexane for 5 min (c). When the hair was incubated in hexane for 5 min, the concentrations of lipids were slightly decreased. In spite of the variations in the conditions, there was virtually no change in the concentrations of hair constituents between (c) and (d). These findings indicate that we do not need to use the two together. The amount of lipid decreased with hexane incubation time ((e), (f), and (g)). This finding indicates that the internal lipids might be lost by diffusion from the inside to the outside with longer exposure to hexane. Considering these results, we concluded that the hexane incubation is not needed to protect the internal lipids of hair. Incubation of hexane may be extracting a tiny amount of lipids from the inside of hair. Finally, all hairs were incubated with a 10% SLES solution for 1 min. On the assumption that prewashing with SLES does not extract an appreciable amount of internal lipids, prewashing was carried out. S13 Supplementary  Fig. 4. TOF-MALDI data for ceramides. Five ceramides were used as model compounds for ceramides. (A) Two kinds of sphingosine. Those sphingosines are ceramides, but hydrocarbons were not used: DL-threo-dihydrosphingosine, and DL-erythro-dihydrosphingosine. (B) Three ceramides: N-hexadecanoyl-D-erythro-dihydrosphingosine, N-octadecanoyl-D-erythrodihydrosphingosine, and N-(R,S)-alpha-hydroxyhexadecanoyl-D-erythro-dihydrosphingosine. For glycerides, it was difficult to detect each component due to the low concentrations. This finding indicates that either the ceramides exist in hair at concentrations below 10 μM (limit of detection) or ceramides do not exist in our hair sample. Supplementary Fig. 6. GC/MS SIM chromatogram of standard materials after methylation for Supplementary Fig. 7. These results involve not only fatty acids but also wax esters and glycerides. Peak: 1. Myristic acid, 2. Palmitic acid, 3. Stearic acid, 4. Oleic acid, 5. 18-MEA.
Supplementary Fig. 7. Normalized concentration of lipids. Numbers indicate the average concentration (μg/g hair) of the lipids extracted from virgin hair. The normalized concentrations were calculated by dividing these concentrations of the control. Error bars represent the standard deviation (SD) (n = 6).
Supplementary Fig. 7 exhibits the total amounts of covalently and noncovalently bound components extracted after methylation. Since the hair was pre-rinsed to remove peripheral lipids (Supplementary Table 4), all lipids found by GC/MS were originally embedded in the hair. The amount of 18-MEA located at the outermost side of the cuticle and covalently bound to underlying proteins through a thioester linkage did not show a significant difference between before and after washing. However, the concentration of fatty acids decreased compared to that of the virgin hair. These findings mean that cuticles layers were not peeled off and lipids were lost from the inside of the hair. Since these fatty acids include those derived from wax esters and triglycerides, it is difficult to characterize the individual behavior of lipids in response to surfactants. Therefore, further experiments ( Fig. 1b) were carried out for noncovalently bound components.
Supplementary Fig. 8. SEM images of samples embedded in paraffin (A), epoxy (B), and conventional adhesive tape (C).
Quantitative data acquisition for spectral mapping demands cross-sectioned hair. Since the conventional methods for sectioning involve a process of removing the embedding material by solvent, the solvent may definitely induce lipid loss from hair. A better method is to perform simple infiltration, which does not require the removal of embedding material. We tried to embed hair in paraffin (A), epoxy resin (B) and conventional adhesive tape (3M Scotch Cat. 810D) (C) without removing the embedding material. After a rubbing wash, hair was embedded in paraffin or epoxy resin (Struers, Denmark) to allow either longitudinal or transversal sectioning. The isolated hair was subjected to an ultramicrotome to obtain a flattened preparation. To obtain a consistent focus in each spot, the sectioned surface had to be flat to generate correct Raman mapping. The flatness of cross-sectioned hair was confirmed by monitoring scanning electron microscope (SEM) images. The method of using adhesive tape was the best because internal cracks were detected for the other methods, as shown in Supplementary Fig. S8(A) and (B). Finally, the hair was isolated between conventional adhesive tape, as shown in Supplementary Fig. S8(C). Supplementary Fig. 10. Raman spectra of hair soaked in pure water, in the solution of 5% polylysine, and in the solution of 5% polylysine and 5% PCI. The Raman spectra of the up and bottom panel, which indicates from different hair swatch, are similar. The spectral intensity in cuticle is higher than that in cortex. Intensity of the amide III changes only after internal modification. Supplementary Fig. 11. Concentration of lipids after immersion washing and rubbing washing. The x-axis in A and B indicates the period when hair contacted the surfactant. The duration of the washing cycles for the rubbing wash equals the time for immersion washing (1 cycle = 1 min). C16 (A) and C16-C16 (B) are shown as one of the lipids in groups A and B, respectively. In (A), the difference between immersion and rubbing washing increased with the number of washes. In (B), the difference between immersion and rubbing wash was maximized at 10 washing times. The patterns of C14, C18, C18=1, and cholesterol were same as those in (A), and the patterns of squalene, C14-C16, and C18-C16 were same as those in (B).