Carbonylation of hair proteins: A robust biomarker of molecular and structural oxidative damage in hair fibres

Abstract

Objective

The first objective of this study is to compare two molecular markers, cysteic acid and protein carbonylation, to track the level of chemical oxidation and photochemical (UV) oxidation of human hair. The second objective is to investigate how the protein carbonylation biomarker evolution is associated with physical characteristics of the fibre. The third objective is to understand the damage localization within the hair sub‐structure.

Materials and Methods

For the chemical oxidation study European natural hair 6/0 (medium brown) is treated with non‐pigmented commercial hair colour mixture and hair bleaching in increasing chemical oxidation conditions. For the photochemical study two European natural hair colours, 4/0 (dark brown) and 9/0 (extra light blond) are gradually photo irradiated. Molecular changes are investigated through cysteic acid measurement by NIR spectroscopy and protein carbonylation measurement; Biophysical changes by differential scanning calorimetry (DSC), and tensile strength measurements. Carbonyls are labelled with a specific fluorescent probe and measured in gel electrophoresis and in situ through image analysis and densitometric quantification after protein extraction respectively.

Results

The two molecular markers, cysteic acid and protein carbonylation, increase similarly in both chemical and photochemical oxidation cases and show a good level of association across the oxidation levels. The fibre physical characteristics (DSC, Tensile Strength) decrease while the protein carbonylation and cysteic acid increase. The in situ visualization of the protein carbonylation shows a high impact on the hair cuticle and a gradual increase of photo‐oxidation through the cortex, phenomenon which is more prominent for the extra light blonde hair in the case of photochemical oxidation.

Conclusion

The protein carbonylation biomarker is validated as another key molecular marker to monitor oxidative chemical changes in the hair chemical groups. It complements the cysteic acid, and appears more suitable in the case of photochemical oxidation, where it offers clear advantages over cysteic acid by being more sensitive and accurate, and by allowing in situ distinct damage visualization. Besides cysteic acid, amino acids such as proline, threonine, arginine, lysine, and peptide bonds are targets of oxidation. Under photochemical oxidation, the photoprotective effect of melanin is confirmed.

The protein carbonylation biomarker is validated as reliable molecular damage marker to monitor hair oxidative damage (chemical and photo oxidation). It is slightly more sensitive than the cysteic acid in the case of photochemical oxidation and allows to geolocate qualitatively and quantitatively the oxidation within the hair fibre substructures.

INTRODUCTION

A human hair fibre emerges from the scalp where the hair follicle, controlled by individual's genetics and influenced by the nourishment supply provided by the blood vessels, synthesizes biologically proteins and cells. As it grows out, the hair fibre passes through the zone of keratinization where key structural chemical linkages like the isodipeptide crosslinks and the characteristic disulfide bridges are built. Eventually, the hair fibre emerges as a permanent fibre, after further dehydration and cornification processes settled the structure [1]. At this point, the structure and colour can no longer be changed by internal biochemical processes. The hair fibre is composed of several ultra‐structurally different compartments. The outer part, the cuticle, shows 5 to 10 overlapping cuticular scales [1, 2]. The main part of the hair fibre named cortex is made up of closely packed cortical cells [1]. The cortical cells further contain a crystalline fibrous protein structure named Intermediate Filaments embedded in an amorphous protein structure called matrix. From the chemical perspective, a human hair is composed primarily of proteins (65%–95%) [1]. Further components include lipids, water, trace elements and pigments [1]. The natural hair colour is determined by the presence of melanin granules [3]. They are mainly found in the matrix of the cortex [4]. The proteins are further defined by their amino acids distribution and sequence. The chemical functionality and reactivity of the amino acids are given by their side chain. Whereas the top 5 amino acids constituting a whole unaltered hair fibre in a quantitative analysis (micromoles/g dry hair) are half‐cystine (disulfide), serine (hydroxyl), glutamic acid (carboxylic acid), proline (hydrocarbon) and threonine (hydroxyl), the most frequent side chains overall are hydrocarbon, hydroxyl, primary amide and carboxylic acid, basic (arginine, lysine, histidine), sulfur and phenolic respectively [1]. Furthermore, the sub‐structures of hair described above have distinctive amino acids constitutions which are in direct relation with their morphology and properties. The cuticle is a sulfur‐rich region with the top 5 amino acids in a quantitative analysis (micromoles/g dry hair) being half‐cystine, (disulfide), serine (hydroxyl), proline (hydrocarbon), glutamic acid (carboxylic acid), valine (hydrocarbon) respectively [1]. The matrix is also a sulfur‐rich region with the top 5 amino acids in a quantitative analysis (residues/100 residues) being half‐cystine, (disulfide), proline (hydrocarbon), serine (hydroxyl), threonine (hydroxyl), arginine (basic)/valine (hydrocarbon) respectively. The alpha‐helical microfibrils are a low‐sulfur region with top 5 amino acids in a quantitative analysis (residues/100 residues) being glutamic acid (carboxylic acid), leucine (hydrocarbon), glycine (hydrocarbon), aspartic acid (acid), arginine (basic) / serine (hydroxyl) respectively [5] holding the helical structure together through hydrogen and ionic bonds. A large percentage of hair proteins are subject to chemical transformation upon weathering, oxidative, reductive processes.

Caring, styling and transforming the hair fibre is of high importance for many people who perceive hair products and their hairstyle as cultivating, socially relevant, a way to express values and individuality, a structural element of everyday life as well as a rejuvenating action [6]. Colouring and bleaching are particularly desired hair transformations to mask the signs of time thus looking younger like a new self and to express individuality [6]. These transformations rely on chemical oxidation reactions, involving Reactive Oxygen Species (ROS), to decompose the natural melanin pigments and induce the formation of the chemical dyestuffs. Alkaline hydrogen peroxide (pH = 10) is the principal oxidizing agent used in the cosmetic colouring and bleaching procedure, and salts of persulfate are often used as “accelerators” in bleaching powders to achieve more lightening effect [1]. Because of the severe conditions employed in the oxidative bleaching process, both alkalizers and oxidizing agents not only decompose the melanin pigments to lighten the underlying colour of the hair, but also cause undesirable side reactions [7] and negative effects for the consumers' hair like cuticle fragility, fibre porosity, surface roughness, dryness, loss of colour, loss of shine, stiffness and brittleness, split ends and hair breakage. Hair weathering, in the sense of light exposure, is another inevitable oxidative process involving Reactive Oxygen Species (ROS). Both oxidation processes are unselective and can impact the hair structure in many ways at the molecular level. Chemical bonds or groups like thioester bond, disulfide bond, double bond with allylic hydrogen, amine group, hydroxyl group, peptide bond, basic group are oxidation targets [1, 8]. Their oxidation leads to damage, degradation, dissolution of protein and lipids and ultimately to damage in the overall hair structure. Considering the prominence of half‐cystine in the hair fibre, the most harmful effect of oxidative bleaching on hair is the irreversible oxidative cleavage of cystine to cysteic acid [9, 10]. Naturally, cysteic acid is a key damage marker of oxidative damage and can be quantified by normalizing the S=O band intensity (near 1040 cm⁻¹) using Fourier transform near‐infrared spectroscopy (FT‐NIR) [11]. However, other amino acids are known to be sensitive to oxidation like methionine, tyrosine, lysine, histidine [1]. Under sunlight, the hair fibre shows a high sensitivity towards cystine, proline and valine oxidation as well as the tertiary hydrogen atoms and allylic hydrogen atoms of lipid structures like 18‐MEA, cholesterol, cholesterol sulfates, oleic and palmitoleic acids [1]. Additionally, oxidation at the peptide backbone of the proteins has been shown to occur under UV‐light exposure and cleaves the protein chain in one entity with a carbonyl group and one entity with an amide group [1]. Previous studies highlighted the negative consequences of protein carbonylation on hair fibres structural damage and deterioration, underscoring the importance of protecting hair proteins from carbonylation as an efficient strategy for safeguarding against airborne pollutants and UV radiation [12]. Further damage induced by perm treatment, bleach treatment, treatment with oxidative hair dyes, combing, heat treatment combined with environmental stressors can also produce protein carbonylation. This harmful irreversible oxidative protein modification is becoming more and more of interest for hair cosmetic industries, to investigate oxidative hair damage holistically.

As a matter of fact, several oxidative mechanisms are known to introduce carbonyl groups into proteins. One mechanism is related to direct oxidation, where hydroxyl radicals induce the oxidation of specific amino acid side chains (Pro, Arg, Lys, Thr) or break the protein backbone. This process forms reactive carbonyl derivatives on proteins resulting in modified amino acid and protein structures. Another mechanism for carbonylation in proteins involves the reaction of lysine residues with reactive carbonyl species generated through glyoxidation or glycation. This reaction, involving reducing sugars, leads to the formation of advanced glycation end products (AGEs) with carbonylated groups, including carboxymethyllysine and pentosidine, for example. Additionally, protein‐bound carbonyls can arise from interactions with lipid peroxidation products (LPPs) like 4‐hydroxy‐nonenal‐protein adducts and malondialdehyde‐lysine, formed through the metal‐catalysed oxidation of polyunsaturated fatty acids (PUFAs) and reacting with side chains of lysine, histidine, and cysteine [13, 14]. It is significant to note that the F‐layer of hair shafts, present on the surface of cuticles, consists of covalently bound 18‐methyleicosanoic acid (18‐MEA) and an array of free fatty acids. The presence of carbonylated proteins often results in reduced enzymatic activity, heightened susceptibility to proteolysis and aggregation, and changes in structural properties and intra‐ or inter‐ protein interactions [13, 14]. Contrary to the cysteic acid formation which accounts for only about 9% of the amino acid composition, this process targets approximately 33% of the hair fibres' amino acid composition. An analysis of oxidative changes at the molecular level is proposed in Figure 1 [1, 15, 16]. Accumulation of damage to hair proteins at the molecular level translates to damage across higher structural levels and ultimately the holistic performance and quality of the hair such as poor hair feel and look, reduced hair strength, increased incidents of split ends and breakage [1, 10]. Over the years, the oxidative damage in hair has been studied by numerous techniques. Physical changes in hair structure, or bulk properties, have been quantified via differential scanning calorimetry (DSC) and tensile strength measurement.

FIGURE 1.

Analysis of molecular oxidative changes in the hair fibre leading to the creation of the cysteic acid (orange) and the protein carbonylation (yellow) damage markers. The hair subcomponents are oxidized by Reactive Oxygen Species (ROS). The extent of damage can be monitored by the molecular changes and end products on the degradation paths [1, 15, 16].

As opposed to protein carbonylation, cysteic acid has been well established as conventional damage marker for the hair fibre, in particular for the inner hair fibre. Indeed, the amount of cysteic acid increases with increasing oxidation levels via bleaching and increasing levels of cysteic acid correlate with the overall decrease of biophysical properties like decrease in the denaturation temperature (DSC) and decrease in Young's Modulus in wet tensile strength testing [1]. This is particularly valid for oxidation via bleaching where the cysteic acid is the main amino acid degradation, along with other derivatives like lanthionine, lysinoalanine, carboxylic salts and methionine sulfone. In the case of UV oxidation however, the variety of oxidation derivatives is slightly higher, ranging from cysteic acid, Bunte salts, to single amino acids oxidation and lipid oxidation introducing characteristic carbonyl groups [1].

Although previous studies showed that protein carbonylation is associated with hair cuticle hydrophobicity changes, increased damage, as well as with structural impairment and increased permeability of the hair fibre, the association between this damage marker and the biophysical hair fibre characteristics has not been fully studied.

In this work, the combination of different scientific techniques and experiments to investigate the inner hair structure, performed under different conditions, is used to obtain a more comprehensive indication and systematically analyse the effect of bleaching and UV irradiation on hair. Additionally, by comparing the information provided from those techniques, it will be possible to assess the suitability and the sensitivity of the newer protein carbonylation damage biomarker for hair, identify new oxidative degradation products of hair protein and correlate molecular changes like carbonylation of hair keratin with structural integrity of the hair fibres.

METHODS

Hair material

European mixed hair from Kerling International (Backnang, Germany) was used for the investigations. For chemical oxidation study, European Natural Hair 6/0 (medium brown), in form of hair swatches (length 12 cm, weight 1 ± 0.05 g) was used. For the photochemical study, European Natural Hair 4/0 (dark brown) and 9/0 (extra light blond), super fine, in form of tresses (length 15 cm, thin hair layer) was used.

All tresses/swatches were cleaned with a 3% aqueous solution (deionized water) of SLES (pH adjusted to 6–7) for 30 min. Finally, they were air dried for at least 48 h before further treatment.

They were subsequently treated with either wet chemicals or a photochemical treatment.

Hair treatments

Chemical treatment

To investigate the evolution of the damage markers with oxidation, gradual oxidation conditions were defined: a medium oxidative damage (Tone on Tone, TT2) and a high oxidative damage (Bleaching, B6). The chemical oxidation process was performed by applying a commercial product base (Schwarzkopf Professional, Hamburg, Germany). Due to the fact that the oxidative damage is only caused by the hydrogen peroxide and alkalizing agent, the tone on tone colouring was done without pigments: Tone on Tone (TT2) permanent hair colouring cream base without pigments: Schwarzkopf Professional Igora Royal 6–0 without pigments. Tone on tone (TT2) hair colour developer: Schwarzkopf Professional Igora Royal 6%/20 Volume developer. Bleaching powder (B6): Schwarzkopf Professional Blond Me. Bleaching developer (B6): Schwarzkopf Professional Blond Me 9%/30 Volume developer. Both TT2 components were mixed in 1:1 ratio, and applied on dry hair (for DSC 4 g coloration mixture per g hair, for single hair measurements surplus of application mixture) for 30 min at 32°C. Both B6 components were mixed in 1:2 ratio, and applied on dry hair (for DSC 4 g bleaching mixture per g hair, for single hair measurements surplus of application mixture) for 45 min at 32°C.

Afterwards the strands were rinsed thoroughly with lukewarm tap water for 1 min and blow dried.

Photochemical treatments

The irradiation was conducted with a Q‐Sun XE‐3 from Q‐Lab Germany using a xenon arc lamp with a day light filter (0.66 W/m² controlled at 340 nm) to simulate an everyday exposure to UV light in Middle Europe. The assumed real irradiation duration and intensity was calculated on the base of a publication of the Splendid Research GmbH, Hamburg in combination with additional information from Germany's National Meteorological Service, Offenbach [17, 18, 19, 20]. One tress per natural hair colour (4/0 and 9/0), per method and test condition was irradiated.

To investigate the evolution of the damage markers with photochemical oxidation, gradual photochemical oxidation conditions were defined: the following periods with different durations of sun exposure were chosen to simulate the UV exposure

• 0 h representing the untreated reference.

• 9 h equivalent to 1 month exposure (equals 2.57 MJ/m²).

• 28 h equivalent to 3 months exposure (equals 7.72 MJ/m²).

• 55 h equivalent to 6 months exposure (equals 15.43 MJ/m²).

• 110 h equivalent to 12 months exposure (equals 30.88 MJ/m²).

Spectroscopy

The NIR spectra were recorded by a MPA™ FT‐NIR‐Spectrometer from Bruke Optik GmbH. The near infrared (NIR) range extends from 12 500 to 4000 cm⁻¹ and is characteristic for overtone and combination vibrations for the CH‐, OH‐ and NH‐ groups. One hair tress per test condition was analysed at six different sample positions in diffuse reflection each using the integrating sphere model. The wavelengths area from 5790 to 4020 cm⁻¹ was used for the analysis of the evaluation of cysteic acid. The content of the chemical marker cysteic acid is expressed in weight percent.

Protein carbonylation

Protein carbonylation assessment was performed in situ and upon biochemical extraction of hair proteins as described previously [12].

In situ detection

Cross‐sections of hair fibres, perpendicular to the hair axis, were obtained with a thickness of 5 μm using a LEICA Crystat (Leica Biosystem, Nussloch, Germany). Carbonyl moieties were specifically labelled in situ on hair strands cross‐sections as described previously [12]. Hair images were captured using epi‐fluorescence microscopy (EVOS M5000, Thermo Fisher Scientific). To determine carbonylation levels, images were processed with ImageJ software [21]. Carbonylation levels were quantified (n = 3 per condition) as intensity relative fluorescence units (R.F.U.) over the surface. Analyses focused on both the cuticle region and cortex, with the medulla region being excluded. Identical conditions (acquisition time, exposure, focus, and resolution) were maintained during image acquisition for all samples within the analysed series.

Carbonylated proteins analysis upon extraction

Hair strands (10 mg, n = 3 per condition) were immersed in an optimized extraction buffer tailored for this matrix [12]. Protein concentration was determined and carbonylated proteins labelled using a specific fluorescent probe reacting directly and specifically with carbonyls [12]. Subsequently, proteins were separated by high‐resolution electrophoresis. Total proteins were post‐stained with SyproRuby™ (Thermofisher) protein gel stain. Image acquisition for both carbonylated and total proteins was carried out using an iBright system (Thermo Fisher Scientific). Image processing and densitometric analysis of protein bands were conducted using ImageJ software [21]. The levels of carbonylation (absolute quantification) were determined by quantifying the densitometric signal of carbonylated proteins, normalized to the signal of total proteins for each sample (Carbonylation score in sample X = Fluorescent signal of carbonylated proteins in sample X/Fluorescent signal of total proteins in sample X). Finally, the carbonylation levels were normalized on the control reference condition.

Biophysical methods

DSC testing

For the investigations with DSC, a Perkin Elmer DSC 8000 was used. The measurement was conducted in a temperature range of 100–190°C with a heating rate of 10 Kelvin per min. Three hair swatches per test condition were prepared for the chemical oxidation study and one tress was used for the photochemical study. The hair was cut into snippets of about 1 mm in length. Twelve aliquots per test condition were placed into the large volume (60 μL) stainless steel DSC‐pans. After adding 50 μL of deionized water each pan was sealed. The denaturation temperature (peak apex [°C]) was determined.

Tensile strength testing

For the chemical oxidation study 50 hair fibres per test condition were mounted manually in hair clamps with plastic tabs (supplied from Dia‐Stron Ltd., UK) and liquid epoxy resin. The free fibre length between the two plastic tabs was 30 mm. Single hair fibres were treated as described earlier and stored at least 48 h before running the stress–strain measurements. For the photochemical study one hair tress was used for each test condition. Fifty hair fibres per test condition were mounted as described above. At the beginning of each test, the thickness of each hair fibre, as mean cross‐sectional area, was determined at 23°C and a relative humidity of 50%. The device used is an automated hair thickness measurement system S.E.A. HDM with Keyence 3‐CMOS Green Light LED system LS‐9006 (S.E.A.—Science & Engineering Applications Datentechnik GmbH, Germany). The cross‐sectional area is used to normalize the measured force values and thus obtain a comparable stress value (N/mm²). The hair fibres were then soaked in water for at least 20 min for equilibrium purposes. Afterwards, they were stretched with a constant speed rate of 20 mm/min within the elastic phase (0%–1.5% elongation) using a Stress–Strain‐System MTT 690 with control unit UV 1000 from Dia‐Stron Ltd., UK. The E‐Modulus (Young's Modulus) in water was calculated before product application and after product application (for the chemical oxidation study) using the UvWin 4.2.2.0 software from Dia‐Stron Ltd., UK.

Statistics

For the DSC and Tensile strength: the measurement series were compared by using the following statistical tests implemented in STATISTICA 12.0 (Stat Soft Inc., USA).

For the protein carbonylation, statistical analyses were performed by one‐way ANOVA followed by Dunnett's multiple comparisons test, performed using GraphPad Prism version 10.0.0 for Windows, GraphPad Software, Boston, Massachusetts USA, http://www.graphpad.com.

Results were considered statistically different (*) when p‐value <0.05.

RESULTS AND DISCUSSION

Evolution of the cysteic acid and protein carbonylation markers with oxidative levels

Figure 2 shows the evolution of both damage markers upon chemical oxidation side by side.

FIGURE 2.

Quantification of cysteic acid and carbonylation levels in hair upon chemical oxidation presented as bar graph. Cysteic acid is measured by NIR and expressed in weight % and carbonylated protein levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (RV; Reference Virgin). Bars (cysteic acid in plain line; carbonylation levels dotted line) represent the arithmetic means, whiskers are the 95% confidence ranges, (*) are the statistically significant differences (Cysteic acid: Tukey HSD Test, p < 0.05).

The cysteic acid content gradually increases as the chemical oxidation level increases. The differences observed are statistically significant. This is consistent with the observations in the literature with the same method [22] and with other methods like FTIR and Raman [23, 24] or amino acid analysis [25]. The protein carbonylation score of the whole fibre gradually increases as the chemical oxidation level increases. The differences observed are statistically significant. This is in good agreement, in a wider scope, with the findings of Masaki et al. [26] on the effect of bleaching on the protein carbonylation of hair surface.

Both damage markers follow the same evolution profile and can statistically differentiate between all the oxidation levels. It is worth noticing that the standard deviations are smaller for the protein carbonylation data set than for the cysteic acid data set.

Figure 3 visualizes the protein carbonylation level of the outer hair fibre. The fluorescent emission signal for the carbonylated proteins increases on the cuticle as the oxidative level increases. This corroborates the results obtained by Masaki et al. [26].

FIGURE 3.

Visualization of carbonylation levels in situ (on hair fibre surface). Carbonylated proteins were labelled in situ on hair shaft with a specific fluorescence probe and presented as maximum intensity levels in white and lower levels in deep purple/black. Left panel: Image of cuticle (on surface) carbonylation levels of non‐treated hair fibre (RV, reference virgin); central panel: Treated hair fibre (TT2, tone on tone coloured), right panel: Treated (Bleached) hair fibre.

Figures 4 and 5 show the evolution of both damage markers upon irradiation side by side for a dark brown hair (4/0) and an extra light blond (9/0) hair respectively.

FIGURE 4.

Quantification of cysteic acid and carbonylation levels in dark brown hair (4/0) upon photochemical oxidation presented as bar graph. Cysteic acid is measured by NIR and expressed in weight % and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation). Bars (cysteic acid in plain line; carbonylation levels dotted line) represent the arithmetic means, whiskers are the 95% confidence ranges, (*) are the statistically significant differences (Cysteic acid: Tukey HSD Test, p < 0.05).

FIGURE 5.

Quantification of cysteic acid and carbonylation levels in extra light blonde hair (9/0) upon photochemical oxidation presented as bar graph. Cysteic acid is measured by NIR and expressed in weight % and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation). Bars (cysteic acid in plain line; carbonylation levels dotted line) represent the median (Cysteic Acid) or arithmetic mean (Protein Carbonylation), whiskers are the 95% confidence ranges, (*) are the statistically significant differences (Cysteic acid: Siegel‐Castellan Test, p < 0.05).

Although the cysteic acid relative content tends to increase with the duration of the irradiation for both natural hair colours, there are only significant differences between the standard (0 h) and the highest irradiation duration (110 h; 12 months) for both the dark brown hair (4/0) and the extra light blonde hair (9/0). According to Robbins [1], a photo‐degradation of the disulfide bond with the creation of cysteic acid through radical reaction should have been expected. However, considering that the relative content of cystine decreases (data not shown), these data seem to be in good agreement with Hoting [27], and would support the theory of multiple cystine degradation paths. A more thorough amino acid analysis could bring more explanation. Additionally, the slight difference in cysteic acid relative content evolution between the dark brown (4/0) and extra light blonde (9/0) hair, the dark brown hair showing an overall trend towards a slower increase, can be attributed to the photoprotective effect of melanin which retards the protein oxidation as melanin and keratin compete for the absorption of photons.

The orders of magnitude between the chemical oxidation and the photochemical oxidation are different and it can be noticed that the chemical oxidation (bleach) has more impact on the cysteic acid molecular marker than 110 h UV irradiation. The protein carbonylation score increases gradually with the duration of the irradiation for both natural colours. The increases are significant already as from 28 h for dark brown hair and 9 h for extra light blonde hair. The molecular biomarker is significantly impacted between the standard (0 h) and the highest irradiation duration (110 h; 12 months) for both natural hair colours. This was expected from Robbins [1] as it has been shown in wool and hair that the oxidation at the peptide backbone under UV exposure produces carbonyl groups. A similar evolution of the same biomarker was observed under UV‐A irradiation by Cavagnino et al. [12].

Similarly to the cysteic acid evolution, the protein carbonylation score tends to increase faster in blonde hair than in brown hair. This supports the photoprotective effect of melanin.

Overall, both damage markers follow the same evolution profile and can statistically differentiate between the extreme oxidation levels. The statistical differentiation between the smaller oxidation steps is limited for the cysteic acid NIR method, which can only differentiate between 9 and 110 h, while the protein carbonylation method is more sensitive, and can detect earlier changes, as from 9 h irradiation. This can be explained by a lower standard deviation thus a higher precision of the method.

Figure 6 visualizes the protein carbonylation on the outer hair fibre and the inner hair fibre of a dark brown (4/0) hair. The fluorescent emission signal for the carbonylated proteins visually increases on the cuticle as well as slightly within the cortex as the oxidative level increases.

FIGURE 6.

Visualization of carbonylation levels in situ, on hair fibre surface (cuticle view, a) and on hair fibre cross‐section (sagittal view, b) for dark brown (4/0) hair. Carbonylated proteins were labelled in situ on hair shaft with a specific fluorescence probe and presented as maximum intensity levels in white and lower levels in deep purple/black. Left panels: Not irradiated hair fibre (0 h); central panels: Irradiated hair fibre (28 h, equivalent to 3 months irradiation); right panels: Irradiated hair fibre (110 h, equivalent to 12 months irradiation).

Figure 7 visualizes the protein carbonylation on the outer hair fibre and the inner hair fibre of an extra light blond (9/0) hair. The fluorescent emission signal for the carbonylated proteins visually increases on the cuticle as well as within the cortex as the oxidative level increases.

FIGURE 7.

Visualization of carbonylation levels in situ, on hair fibre surface (cuticle view, a) and on hair fibre cross‐section (sagittal view, b) for extra light blonde (9/0) hair. Carbonylated proteins were labelled in situ on hair shaft with a specific fluorescence probe and presented as maximum intensity levels in white and lower levels in deep purple/black. Left panels: Not irradiated hair fibre (0 h); central panels: Irradiated hair fibre (28 h, equivalent to 3 months irradiation); right panels: Irradiated hair fibre (110 h, equivalent to 12 months irradiation).

The visual comparison of the fluorescent signals for the carbonylated proteins of dark brown (4/0) and extra light blond (9/0) hair supports once again the photoprotective effect of melanin observed by Hoting et al. [12, 27] as well: for both natural hair colours, the cuticle is the sub‐component which is primarily and strongly impacted by the photochemical oxidation, and the cortex is secondarily impacted. However, the cortex sub‐component is much more impacted in the extra light blond (9/0) hair with lower level of melanin granules than in the dark brown (4/0) hair with higher level of melanin granules. The protein degradation is expected to come principally from the UV‐A and UV‐B fractions of the light spectrum [27] and the lipid degradation mostly from the visible fraction of the light spectrum [28].

This observation is confirmed when looking at the quantification of the respective fluorescent signals graphed in Figure 8. The levels of carbonylated proteins reflected by the scales are overall higher for the extra light blond (9/0) hair than for the dark brown (4/0) hair. For the dark brown hair, the amount of carbonylated proteins increases as the oxidative level increases and reaches a point of saturation for all sub‐components (cuticle, cortex) as from 28 h (3 months) irradiation. This means that there is no significant increase in carbonylation levels beyond this point, indicating that the levels reached a plateau. For the extra light blond hair, the amount of carbonylated proteins increases as the oxidative level increases. No plateau is observed for any of the sub‐components for this hair natural colour. A visual assessment and quantification of the hair strands colour change (data not shown) corroborates the evolution profiles of both the whole hair and the cortex sub‐component for both natural hair colours. This difference can be attributed to the photoprotective effect of melanin for the cortex: the extra light blond hair has less melanin granules to compete with the keratin structure for the absorption of photons and therefore the keratin structure is left unprotected and degrades gradually upon irradiation. Whereas the dark brown hair still has melanin granules which absorb photons and contain the protein carbonylation process. Interestingly, the levels of carbonylation reached a plateau for the cuticle sub‐component of the dark brown hair, which is not expected and surprising. These results cannot be associated, at first, to the photoprotective effect of melanin as the melanin granules are located in the cortex sub‐component. The expectation would have been a gradual increase of the carbonylated protein level, similar to the extra light blond hair. One hypothesis could be that the oxidation level after 28 h is already enough to oxidize all the proteins chemical target groups prone to carbonylation or related to already relatively high initial levels of carbonylation, which is still surprising because the absolute levels are higher in the Extra Light blonde hair. Nevertheless, the differences of kinetics of the protein carbonylation in the cuticle region between the two natural colour remains unexplained. Another explanation could be that the melanin granules in the cortex still play a photoprotective role by intercepting the light entering from the diametrically opposed side of the hair. Further investigations may help elucidating these dynamics.

FIGURE 8.

Quantification of carbonylation levels in hair upon photochemical oxidation presented as bar graphs. The carbonylation levels analysis is made by fluorescence intensity quantification over the evaluated surface. Three (3) images per condition were analysed. Results are reported by hair sub‐component (i) Whole hair, (ii) Cuticle; (iii) Cortex) for (a) dark brown and (b) extra light blond hair after 0 h (equivalent to no irradiation), 28 h (equivalent to 3 months irradiation), and 110 h (equivalent to 12 months irradiation). Bars represent arithmetic means, whiskers are the 95% confidence ranges, (*) are the statistically significant differences.

Association of cysteic acid and protein carbonylation

A plot of the two datasets, cysteic acid against protein carbonylation (Figure S1), reveals a good association of the two damage markers in the case of a chemical oxidation. The coefficient of determination for the regression line fitting the data points is R ² = 0.8322. Further data points could be added to refine this analysis.

Figure 9 shows a plot of the two datasets, cysteic acid against protein carbonylation for the dark brown hair. It reveals an association of the two damage markers in the case of photochemical oxidation. The coefficient of determination for the regression line fitting the data points is R ² = 0.7272. Further data points could be added to refine this analysis.

FIGURE 9.

Cysteic acid (measured by NIR; expressed in weight %) versus carbonylation levels (measured by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups; expressed as % of RV respectively), as a function of photochemical damage after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation for dark brown (4/0) hair. Trendline is added with Excel and shown as dotted line (……).

Figure 10 shows a plot of the two datasets, cysteic acid against protein carbonylation for the extra light blond (9/0) hair. It reveals a good association of the two damage markers in the case of photochemical oxidation. The coefficient of determination for the regression line fitting the data points is R ² = 0.8105. Further data points could be added to refine this analysis.

FIGURE 10.

Cysteic acid (measured by NIR; expressed in weight %) versus carbonylation levels (measured by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups; expressed as % of RV respectively), as a function of photochemical damage after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation for extra light blonde (9/0) hair. Trendline is added with Excel and shown as dotted line (……).

The two molecular damage markers cysteic acid and protein carbonylation show a good association across the two oxidation sources (chemical and photochemical).

The level of association of the two damage markers is higher in the case of the chemical oxidation than in the case of a photochemical oxidation, the lowest association level being for the photochemical oxidation of the dark brown hair. This could suggest that the degradation paths triggered by these two damage sources are slightly different and could support the theory of multiple cystine degradation paths in the case of photochemical oxidation. Adding more data points to the chemical oxidation experiment could help confirm the high level of association and repeating the photochemical experiment with the irradiation of single fibres could confirm this hypothesis.

Thus, the protein carbonylation damage biomarker appears to be more sensitive and reliable in detecting the effects of photochemical oxidation on the hair fibre. The visualization of the oxidized proteins via fluorescence emission signal adds a valuable complement of information in localizing the effect in the hair structural sub‐components, allowing to support the efficiency of substances in the hair fibre.

In terms of localization of damage within the hair substructure, the cysteic acid damage marker principally reflects the extent of the damage in the cuticle and matrix regions, with the half‐cystine being the top amino acid in the cuticle and matrix region [1, 5].

On the other hand, the residues susceptible to carbonylation, like the amino acids Proline, Arginine, Threonine, Lysine, are mostly present in the matrix with Proline, Threonine and Arginine being Top 2, 4, 5 amino acids in the matrix; and in the Intermediate Filaments with Arginine being Top 5 amino acid [1, 5]. Under the conditions of the study, the peptide bonds could be cleaved and the cleaved ends are susceptible to oxidize leading to protein carbonylation; the lipids could undergo peroxidation leading to protein carbonylation by indirect reaction as well. Both the peptide bonds and the lipids are located in the whole hair fibre and in the different cell membrane complex regions respectively. The protein carbonylation damage marker enables a more comprehensive evaluation of the oxidation state in all the substructures of the hair fibre, along with the visualization of the damage which points out differentiated damage at the level of morphological components of hair.

EVOLUTION OF THE DAMAGE MARKERS AND BIOPHYSICAL PARAMETERS

Evolution of the damage markers and denaturation temperature

The chemical damage induced by bleaching on the structure of alpha‐helix keratin of human hair was investigated by using differential scanning calorimetry. According to the two‐phase model, α‐keratins are constituted of two components which are identified as the partially helical, crystalline intermediate filaments (IFs) and the cross‐linked, amorphous matrix, made of intermediate filament associated‐proteins (IFAPs) [29]. DSC yields the denaturation temperature T _D, which is kinetically controlled by the cross‐link density and viscosity of the non‐helical amorphous matrix. When the hair fibre is damaged, the denaturation temperature of alpha‐helix keratin is expected to decrease. The DSC measurement of hair was conducted in water to avoid interference from pyrolysis effects above approx. 230°C [30]. Figure 11 summarizes the results for denaturation temperatures using DSC measurement to assess hair damage level after colouring (TT2) and bleaching (B6) treatment respectively. Oxidative chemical treatments lead to a dramatical decrease of denaturation temperature, which confirms a degree of chemical changes to the fibre. The oxidative colouring and bleaching treatment break the disulfide bonds creating cysteic acid and oxidize the protein structure leading to the creation of carbonylated protein. Both of these molecular changes lead to a weaker hair structure, which results in a decrease of thermal properties. A more important decrease in the denaturation temperature of alpha‐helix keratin of bleach‐damaged hair is observed, indicating a lower energy needed to disrupt the keratin alpha‐helix structure of bleached hair. As expected, the stronger the oxidative conditions are, the higher the level of hair damage is.

FIGURE 11.

Quantification of peak apex temperature or denaturation temperature (°C) and carbonylation levels of European Natural Hair (6/0) upon chemical oxidation presented as bar graph and individual data points. Median peak apex temperature, expressed in°C, is measured by differential scanning calorimetry and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference RV. Results are reported for RV (Reference Virgin), TT2 (Tone on Tone coloured) and B6 (Bleached) hair. Bars represent median peak apex temperature, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant differences for the median peak apex temperature (multiple comparison with post‐hoc Test after Dwass‐Steel‐Critchlow‐Fligner, p < 0.05). Statistical differences for the Protein Carbonylation Score were shown in Figure 2.

Figure S2 shows a plot of the two datasets, peak apex temperature against protein carbonylation. Figure S3 shows a plot of the two datasets, peak apex temperature against cysteic acid. They reveal a good association of the denaturation temperature with the two respective damage markers in the case of chemical oxidation. The coefficients of determination for the regression line fitting the data points are R ² = 0.9375 and R ² = 0.9983 for the protein carbonylation and the cysteic acid respectively. Further data points could be added to refine this analysis.

Figure 12 shows the evolution of the protein carbonylation damage marker and the denaturation temperature upon irradiation for a dark brown (4/0) hair. They follow an inverse evolution profile: as the photochemical oxidation level increases, the protein carbonylation level increases while the denaturation temperature decreases. The decrease of the denaturation temperature upon photochemical oxidation was expected from Giancola et al. [31] where the denaturation temperature of both a virgin unpigmented hair and a medium brown hair decreased. A similar observation was made by He [32] on medium brown hair as well.

FIGURE 12.

Quantification of peak apex temperature or denaturation temperature (°C) and carbonylation levels of dark brown hair (4/0) upon photochemical oxidation presented as bar graph and individual data points. Median peak apex temperature, expressed in°C, is measured by differential scanning calorimetry and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Bars represent median peak apex temperature, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant differences for the median peak apex temperature (multiple comparison with post‐hoc Test after Dwass‐Steel‐Critchlow‐Fligner, p < 0.05). Statistical differences for the Protein Carbonylation Score were shown in Figure 4.

Figure 13 shows a plot of the two datasets, peak apex temperature against protein carbonylation for dark brown hair (4/0). It reveals a tight inverse association of the protein carbonylation damage marker and the denaturation temperature biophysical parameter in the case of photochemical oxidation. The coefficient of determination for the regression line fitting the data points is R ² = 0.8991. As a reference, for the same plot of the peak apex temperature against the cysteic acid, the coefficient of determination for the regression line fitting the data points is R ² = 0.6799 (Figure S4). This is in good agreement with the visualization of the protein carbonylation (Figure 6) and the quantification (Figure 8) which showed that the cortex was affected by carbonylation and the denaturation temperature indicates the damaging of both the matrix and the α helix (cortex).

FIGURE 13.

Median peak apex temperature (measured by differential scanning calorimetry; expressed in°C) versus carbonylation levels (measured by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups; expressed as % of reference (0 h irradiation)) for dark brown (4/0) hair, as a function of photochemical damage after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Trendline is added with Excel and shown as dotted line (……).

Figure 14 shows the evolution of the protein carbonylation damage marker and the denaturation temperature upon irradiation for an extra light blond (9/0) hair. They follow an inverse evolution profile: as the photochemical oxidation level increases, the protein carbonylation level increases while the denaturation temperature decreases. The decrease of the denaturation temperature is expected, similarly to the results of the dark brown hair and the studies of Giancola [31], particularly on unpigmented virgin hair, and He [32].

FIGURE 14.

Quantification of peak apex temperature or denaturation temperature (°C) and carbonylation levels of extra light blonde (9/0) hair upon photochemical oxidation presented as bar graph and individual data points. Mean peak apex temperature, expressed in°C, is measured by differential scanning calorimetry and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Bars represent mean peak apex temperature, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant differences for the mean peak apex temperature (multiple comparison with post‐hoc Test and HSD test for unequal N, p < 0.05). Statistical differences for the protein carbonylation score were shown in Figure 5.

Figure 15 shows a plot of the two datasets, peak apex temperature against protein carbonylation, for the extra light blond (9/0) hair. It reveals a tight inverse association of the protein carbonylation damage marker and the denaturation temperature biophysical parameter in the case of photochemical oxidation. The coefficient of determination for the regression line fitting the data points is R ² = 0.9189. As a reference, for the same plot of the peak apex temperature against the cysteic acid, the coefficient of determination for the regression line fitting the data points is R ² = 0.5723 (Figure S5). This is in good agreement with the visualization of the protein carbonylation (Figure 7) and the quantification (Figure 8) which showed an incremental increase in the cortex and the denaturation temperature indicates the damaging of both the matrix and the α helix (cortex).

FIGURE 15.

Mean peak apex temperature (measured by differential scanning calorimetry; expressed in°C) versus protein carbonylation score (measured by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups; expressed as % of reference (0 h irradiation)) for extra light blonde (9/0) hair, as a function of photochemical damage after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Trendline is added with Excel and shown as dotted line (……).

Evolution of the damage markers and tensile strength

Tensile strength measurement is another important method for the analysis of mechanical properties of human hair fibre [33]. Robbins and Crawford demonstrated that the cortex is primarily responsible for the tensile properties of human hair and that the cuticle has little involvement [34]. Therefore, the tensile properties could be primarily considered as an indicator of cortex damage. Moreover, the tensile properties are highly dependent on the condition of hair fibre and climatic conditions during tensile testing [35].

Previous studies [7, 36] suggested that the effects of oxidation are greater and a considerably large difference tends to manifest when measurements are carried out in the wet state than in the dry state. So, in this study, the tensile properties of the hair fibre were measured in water. Some physical parameters like Young's modulus, stress at 15% extension, break stress or break extension can be investigated. The most relevant parameter, the Young's modulus, also known as elastic modulus (=E‐modulus) in the elastic region of the fibre was chosen in this study to evaluate hair strength. Young's modulus is defined as the ratio of stress over strain in the Hookean region. Hooke's law of elasticity states that the longitudinal change of a material body (the strain) is linearly related to the force causing the deformation (the stress). For wet hair this region lies between a strain of approximately 0% and 2%. The higher the Young's modulus the stronger the fibre [35]. In the wet stage, the matrix is a gel and the Intermediate Filaments mainly contribute to the Young's Modulus value [37]. It was confirmed that no substantial structural change is introduced during the elastic region since the deformation is completely reversible and initial properties could be restored [36]. Therefore, its evaluation is a fast and reliable output parameter of the tensile strength experiment. The other advantage to evaluate Young's modulus is that the measurements can be performed on the exact same set of 50 replicate hair fibres. The results of Young's modulus changes after colouring and bleaching are shown in Figure 16. Oxidative colouring and bleaching treatment lead to a modification of hair fibre with a significant decrease in the values of Young's modulus in comparison to untreated hair. Tensile strength decreases as oxidation progresses. This demonstrates that oxidative damage in hair induced by colouring and bleaching results in a loss of mechanical resistance of hair fibre. Bleach treatment decreases Young's modulus even more than 1.5 times in comparison to oxidative colouring treatment.

FIGURE 16.

Mean differences of Young's modulus in the elastic region (N/m²) and carbonylation levels of European Natural Hair (6/0) upon chemical oxidation presented as bar graph and individual data points. Differences of Young's modulus are obtained by subtraction to the Young's modulus of the reference RV (Reference Virgin) and carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference RV. Results are reported for TT2 (Tone on Tone coloured) and B6 (Bleached) hair. Bars represent the mean differences of Young's Modulus in the elastic region, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant before and after treatment differences for the Young's Modulus (variable difference after treatment with Tukey HSD test, p < 0.05). Statistical differences for the carbonylation levels were shown in Figure 2.

In Figure 16, the content of carbonylated protein was considered as an indicator of hair damage induced by colouring and bleaching treatment at the molecular level. As expected, the results indicate that the level of molecular hair damage depends on the strength of oxidants used. This is consistent with the altered physico‐mechanical properties of hair associated with oxidative colouring and bleaching treatment.

Figures 17 and 18 show the evolution of the protein carbonylation biomarker and the Young's Modulus in the elastic region upon irradiation for a dark brown (4/0) hair and an extra light blond (9/0) hair. No significant change in the Young's Modulus was measured whereas the carbonylation levels increase. There is no consensus on the expected results. Past results have been compiled by Dario et al. [38] and some studies show a decrease in mechanical properties (break strength, break elongation although attributed to UVB [39, 40], Beyak et al. [40] and Robbins et al. [1] mention modifications and decrease in the 15% index) whereas other studies do not report any effect on mechanical properties [41]. References about Young's Modulus are sparse or nonexistent. The tensile strength experiment in water is thought to be mostly controlled by the alpha‐helical structure of the intermediate filaments [30]. The chemical structure of the intermediate filaments contains amino acids like glutamic acid, leucine, glycine, aspartic acid, arginine and serine [5]. The UV irradiation is oxidizing the proteins photochemically through direct, mostly UVB, absorption by the hair's most significant chromophores (amino acids tryptophan, tyrosine, phenylalanine and cysteine/cystine) and through reactive oxygen species (ROS) created by UVA and melanin photoreactive characteristic [38] which can oxidize other chemical groups. Beyond the oxidation of the peptide bond, the oxidation of certain top amino acids of the alpha‐helical structure of the intermediate filaments like arginine [8] should have led to alteration of the mechanical properties of the alpha‐helix by destabilization of the ionic bonds, considering that UVA has the ability to penetrated cuticle and influence the cortex through its longer wavelength. The stable Young's Modulus across the hair natural colours and the different levels of UVA irradiation could be explained by a non‐homogeneous irradiation of the hair strands. The irradiation might not have reached all the single fibres necessary for the measurement. The repetition of the experiment by irradiating the single fibres and not the hair strands might provide the confirmation of this hypothesis. Additionally, the experiment could be extended to other test outputs such as Break stress, break extension.

FIGURE 17.

Young's modulus (N/m²) and carbonylation levels of dark brown (4/0) hair upon photochemical oxidation presented as bar graph and individual data points. Carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Bars represent the mean Young's modulus, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant differences for the mean Young's Modulus (univariance significant tests; no statistical difference). Statistical differences for the protein carbonylation score were shown in Figure 4.

FIGURE 18.

Young's modulus (N/m²) and carbonylation levels of extra light blonde (9/0) hair upon photochemical oxidation presented as bar graph and individual data points. Carbonylation levels are obtained on each sample by in gel densitometric analysis of specific signal of fluorescence labelled carbonyl groups and normalized by the total protein signal of the reference (0 h of exposure; equivalent to no irradiation). Results are reported after 0 h (equivalent to no irradiation), 9 h (equivalent to 1 month irradiation), 28 h (equivalent to 3 months irradiation), 55 h (equivalent to 6 months irradiation) and 110 h (equivalent to 12 months irradiation) irradiation. Bars represent the mean Young's modulus, the plain dots (●) represent the carbonylation levels, the line (…) does not indicate a formal connection between data points but is rather meant as a guide for the eye. Whiskers are the 95% confidence ranges, (*) are the statistically significant differences for the mean Young's Modulus (Tukey HSD test; no statistical difference). Statistical differences for the protein carbonylation score were shown in Figure 5.

CONCLUSION

The effects of increasing chemical oxidation of one hair type and photochemical oxidation of two natural hair colours have been investigated using two molecular markers: the well‐established cysteic acid, an oxidative product of cystine, an amino acid characteristic of hair, and protein carbonylation, through a protein carbonylation score aggregating all carbonyl groups characteristic of oxidized protein side chains, oxidized ends from cleaved peptide bonds, oxidized amino acids deriving from lipid peroxidation.

The two markers had similar evolution profiles in both chemical and photochemical oxidation cases and showed a good level of association across the oxidation levels. Furthermore, the protein carbonylation score showed a good level of association with the fibre physical characteristics such as Differential Scanning Calorimetry for all the oxidative conditions, and Tensile Testing for chemical oxidative conditions. The level of association is even better than cysteic acid, in the case of the photochemical oxidation, as shown by the higher coefficients of determination of the regression lines. Thus, the protein carbonylation biomarker is validated as another sensible and reliable molecular damage marker to monitor hair oxidative damage in hair fibres. It complements very well the cysteic acid and offers the clear additional advantages over cysteic acid of being slightly more sensitive and accurate (associated to lower standard deviation) in the case of photochemical oxidation to reliably detect smaller changes as well as the possibility to geolocate qualitatively and quantitatively the oxidation within the hair fibre structural subcomponents by measuring the fluorescence emission signal of carbonylated proteins.

The study of the protein carbonylation biomarker delivered the confirmation that chemical oxidation, especially bleaching, leads to a more extensive oxidative damage than photochemical oxidation. Additionally, it confirmed that the cuticle is the most impacted subcomponent of the hair fibre upon photo oxidation, that the melanin has a photoprotective effect against photo oxidation. It confirmed the unselective oxidation processes the hair fibre is subjected to when colouring, bleaching or under weathering, and enabled to list at least proline, arginine, threonine, lysine, the peptide bonds and lipids, on top of cysteic acid, as direct or indirect irreversible oxidative victims. The biophysical results from DSC and tensile strength testing showing a weaker hair structure are the consequence of a holistic non‐selective oxidation.

This study highlights the importance of adopting preventive and protective strategies to minimize the oxidative stress impact on the hair fibre.

CONFLICT OF INTEREST STATEMENT

Authors Sabine Babiel, Camille Grosjacques, Jing Hodes are employees of Henkel AG. & Co. KGaA. This does not alter the authors' adherence to the journal's policies on sharing data and materials. Authors Anaïs Bobier, Andrea Cavagnino and Martin Baraibar have no conflict of interest to declare.

Supporting information

Figures S1–S8

Grosjacques C, Babiel S, Hodes J, Bobier A, Cavagnino A, Baraibar M. Carbonylation of hair proteins: A robust biomarker of molecular and structural oxidative damage in hair fibres. Int J Cosmet Sci. 2025;47:604–625. 10.1111/ics.13052