Current Views on Noninvasive in vivo Determination of Physiological Parameters of the Stratum Corneum Using Confocal Raman Microspectroscopy Free

The stratum corneum (SC), the outermost layer of the epidermis, consists of enucleated cells – corneocytes embedded into the intercellular lipid (ICL) matrix and serves as initial barrier against penetration of xenobiotics into the skin [1]. Compared to other cellular types, corneocytes are stiff cells [2] and consist mainly of keratin filaments, natural moisturizing factor (NMF) molecules, and water, surrounded by inner protein and outer lipid envelopes [3, 4]. Corneocytes contribute to the skin barrier function [2] and represent an effective barrier to xenobiotics.

Keratin is the main fibrillar protein that forms the cytoskeleton of corneocytes and is involved in providing the skin barrier function. Keratin is characterized by strength and insolubility in water and salt solutions due to the presence of a large number of disulfide bonds (covalent bonds between the sulfur atoms of the sulfur-containing amino acid cysteine) between the peptide chains in its structure. Keratin synthesis begins in the basal layer of the epidermis, where in keratinocytes, the low molecular weight prekeratin has no disulfide bonds. During the proliferation of keratinocytes, due to the action of specific enzymes, prekeratin “matures” in the lower regions of the SC and turns into mature keratin [5], characterized by the presence of disulfide bonds. The SC contains so-called soft keratin, which has a low concentration of sulfur-containing amino acids (≈2–3%). The concentration of keratin and ICL is nonhomogeneously distributed throughout the SC and increases from the bottom toward the uppermost SC depths [6].

The ICLs in the SC are released from lamellar granules of the stratum granulosum [7] and form lamellas, where lipids (mainly ceramides, free fatty acids, and cholesterol [8]) are bound to each other and form the entire membrane-like lamellar structure [9-11]. The nonpolar hydrophobic tails of lipid molecules are connected to each other forming a common hydrophobic central zone of the membrane that is not in contact with water, while polar hydrophilic lipid heads form the periphery of the membrane, which is directed toward the water environment [12, 13], forming a heterogeneous layered structure inside the lamellae. It has been shown that ICL forms a three-layer lamellar structure, consisting of alternating lipid bilayers defining “broad-narrow-broad” segments [14-18], which are known as long (≈13.4 nm) and short (≈6.4 nm) periodicity phases [15, 18, 19]. There is also an opinion that the thicknesses of all lipid layers inside the lamellae are identical and are characterized only by a short periodicity phase with a distance of ≈4.2–4.8 nm [10, 17, 20, 21]. Another model is based on the presence of an asymmetric lamellar phase within the ICL matrix with a distance of ≈11 nm, which is a combination of two phases of 4.5 and 6.5 nm [10, 22]. The presence of three-layer segments with distances of 11.5, 6.0, and 4.5 nm was also observed in vitro analyzing human SC using small-angle and wide-angle X-ray scattering [23]. The exact lamellar organization of the SC’s ICL matrix is still debated, and there is no consensus on exact understanding of how ceramides, cholesterol, free fatty acids, and other substances are linked together in an ICL membrane. Thus, lipids inside the ICL matrix are strictly structured inside lamellas and are oriented, mainly, perpendicular to the skin surface [11], which directly determines their lateral organization and, consequently, the skin barrier function [19, 24, 25]. The lateral organization of the ICL matrix contains orthorhombic (ordered, very densely packed lipids) and hexagonal (ordered, less densely packed lipids) phases, which are nonhomogeneously distributed in the SC depth [23, 26]. Using the x-ray diffraction method in in vitro and ex vivo experiments, a predominant content of the orthorhombic phase was observed at a depth of ≈35–50% of the SC thickness. The presence of orthorhombic lateral organization of ICL determines the skin barrier function provided by the SC [27]. As in vivo measurements are not possible, the skin sample preparation protocol may significantly affect the morphology of the ICL lamellas and should be considered in studies [10, 22]. Quatela et al. [28] support this statement, showing that the separation of SC leads to a decrease in the skin barrier function, i.e., the sample preparation procedure has an effect on the integrity of ICL lamellas.

Water plays a decisive role in the physiological functioning of the skin. The epidermis contains ≈30% of the water in the skin. The SC is the most dehydrated epidermal layer and contains ≈15% of all water in the epidermis. Water serves as a medium for action of numerous enzymes that make the SC metabolically active. The lack of water in the SC is called dry skin and leads to the disruption of the skin barrier function and SC integrity and is often associated with skin diseases, such as atopic dermatitis or psoriasis [29-31]. The lipid loss can cause marked increases in drying of the SC [32, 33]. Thus, the SC hydration is an important physiological parameter of healthy skin, which is indirectly influencing the skin barrier function. Water is nonhomogeneously distributed throughout the SC and has minimal concentration at the superficial SC depth and highest concentration at the bottom SC depth, close to the boundary to the stratum granulosum [34]. The binding properties of water in the SC are dependent on the concentration of binding centers, such as keratin [35], NMF molecules [36], and ICLs [37], whose distribution is nonhomogeneous throughout the SC [38] and is age dependent [39, 40]. The mobility of water molecules and interaction of water with surrounding molecules are of high interest in dermatology and cosmetology. Analysis of skin hydration should be preferably performed noninvasively and in vivo due to the quick evaporation of water from skin biopsy samples.

In noninvasive investigation of the SC in vivo, imaging methods, such as confocal laser scanning microscopy, optical coherence tomography, and two-photon tomography are used mainly for determination of the SC thickness, SC’s surface appearance and to determine penetration depth of topically applied substances [41-44]. These methods are based on the measurement of fluorescence and reflectance of the SC and are not sensitive to the chemical composition of SC components. To date, Raman spectroscopy can be considered one of the most effective and informative noninvasive analytical methods for obtaining information on the chemical composition of substances, conformational order, packing, and orientation of molecules, which makes it advantageous in in vivo investigation of the SC, viable epidermis, and dermis. It was shown on skin biopsies that Raman spectroscopy is a perspective optical noninvasive method in analyzing keratin’s secondary and tertiary structures [45], ICL ordering [46-49], and water mobility states [48]. These results showed the potential for direct structural depth analysis of SC components by Raman microspectroscopy in vivo. Recent development of algorithms for analysis of Raman spectra confirms the applicability of confocal Raman microspectroscopy for in vivo determination of barrier function-related physiological parameters of the human SC, such as lamellar and lateral organization of ICL, folding of keratin, water mobility, and hydrogen bonding states, which is presented in this paper.

As shown in the literature [47, 48, 50, 51], three Raman bands at 1,060, 1,080, and 1,130 cm⁻¹ describing the C-C skeleton stretching vibrations of long-chain hydrocarbons are sensitive to the trans-gauche conformation order in lipids. The bands at 1,060 and 1,130 cm⁻¹ appear in the Raman spectrum if the presence of trans-sequences is more than 6 and the length of a long alkyl chains contain more than 18 CH₂ groups, respectively [52, 53]. With a smaller number of CH₂ groups in the trans-sequence, the position of the Raman band at 1,130 cm⁻¹ is shifted toward larger wavenumbers [53]. Thus, Raman bands at 1,060 and 1,130 cm⁻¹ characterize the lipid chains containing trans conformers (more ordered state), while the Raman band at 1,080 cm⁻¹ characterizes the lipid chains containing only gauche conformers (less ordered state). During the transition from more ordered to less ordered structure of lipids, the shape of the Raman band at 1,060 cm⁻¹ undergoes changes: its width and intensity decreases. The shape of the Raman band at 1,080 cm⁻¹ also changes: its width and intensity increase, and the position of the maximum shifts toward shorter wavenumbers [52]. The typical Raman spectrum of human’s SC measured in vivo in the 1,040–1,150 cm⁻¹ region is presented in Figure 1a, where the dotted areas show the boundaries used for calculation of lipid-related band intensities.

To describe the influence of surrounding humidity on conformational changes in human ICL ex vivo, Vyumvuhore et al. [48] used the (I_1,130 + I_1,060)/I_1,080 ratio of Raman band intensities. Since the Raman band intensity at 1,080 cm⁻¹ is significantly lower than the sum of Raman band intensities at 1,060 and 1,130 cm⁻¹ in the SC, Choe et al. [54] have used an inverse I_1,080/(I_1,130 + I_1,060) ratio to reduce the magnitude of the standard deviation. Thus, the higher value of the I_1,080/(I_1,130 + I_1,060) ratio characterizes ICL with a greater number of gauche conformers in the lipid chains, i.e., a less ordered lamellar organization of ICL. A smaller value of the I_1,080/(I_1,130 + I_1,060) ratio characterizes ICL with a greater number of trans conformers in the lipid chain, i.e., a more ordered lamellar organization of ICL.

The depth profile of the lamellar organization of ICL is presented in Figure 1b and shows a nonhomogeneous distribution: the I_1,080/(I_1,130 + I_1,060) ratio decreases from its maximal initial value (highest content of gauche conformers in the lipid chains) on the skin surface to a minimum value (highest content of trans conformers in the lipid chains) at a depth of 20–50% of the SC thickness. Further, there is a monotonic increase of I_1,080/(I_1,130 + I_1,060) ratio at the boundary of the SC and the stratum granulosum, i.e., an increase of the content of gauche conformers in the lipid chains. Thus, the most ordered lamellar organization of the ICL in the SC is observed at a depth of ≈20–50% of SC thickness, while it is less ordered at the surface and at a boundary between SC and stratum granulosum [54].

The Fourier-transform infrared spectroscopy method can be used to directly measure the lateral organization of ICL in vivo [55], but due to the high absorption of excitation and Raman radiation by water, measurements are limited to the superficial SC depths [49]. The ratio of Raman band intensities at 2,880 cm⁻¹ (antisymmetric stretching mode of CH₂ groups) and 2,850 cm⁻¹ (symmetric stretching mode of CH₂ groups) I_2,880/I_2,850 depends on the degree of lipid organization, i.e., the transition from crystalline to amorphous states [56, 57]. Gaber and Peticolas [46] studied a mixture of hexadecan (C₁₆H₃₄) and deuterated hexadecan (C₁₆D₃₄) and showed that the I_2,880/I_2,850 ratio describes the degree of the lateral organization of lipids. A higher (lower) I_2,880/I_2,850 ratio characterizes an increased content of trans (gauche) conformers in the lipid chains – higher (lower) lateral ordering of lipids in the lamellas. The lateral organization of lipids was determined in pure lipid systems, which does not correspond to the real situation in the skin, where Raman spectra of lipids and keratin are strongly superimposed. The Raman spectra of the human SC and its components keratin and lipids in the 2,810–3,030 cm⁻¹ region are shown in Figure 2a. Since lipids in the SC are mainly in crystalline domains containing hexagonal and orthorhombic organizations [58], the I_2,880/I_2,850 ratio describes the lateral organization of ICL in the SC [51]. The possibility to determine the lateral organization of ICL in the SC using Raman spectroscopy was shown in vitro on extracted lipids [47]. For correct determination of the lateral organization of ICL in the SC in vivo, it is necessary to separate the lipid and keratin contributions from each other. Decomposition of the Raman spectrum in the 2,810–3,030 cm⁻¹ range does not provide a sufficient separation [59]. Therefore, Choe et al. [54] have developed a new algorithm, which is based on the calculation of the known contributions of lipids and keratin in the Raman spectrum in the 2,810–3,030 cm⁻¹ region, which is schematically presented in Figure 2a.

To determine the lateral organization of ICL in the SC, the mathematical algorithm shown in Equation 1 was proposed [54], where K_2,850 = JI/AD, K_2,880 = HE/AD (parameters directly measured from the skin Raman spectra), and P_K^2,850 = MI/BD = 0.150 ± 0.025, P_K^2,880= FE/BD = 0.440 ± 0.024, P_L^2,850 = KI/CD = 0.480 ± 0.031, P_L^2,880= GE/CD = 0.320 ± 0.026 (parameters measured on extracted keratin and lipids).

Further, using the developed algorithm, the distribution depth profile of the lateral organization of the ICL in the human SC was determined in vivo for the first time (Fig. 2b). A higher (lower) I_2,880/I_2,850 ratio characterizes a higher amount of trans (gauche) conformers in the ICL chains, and thus shows an increase of orthorhombic (hexagonal) packing of ICL [54]. As can be seen in Figure 2b, the ratio of I_2,880/I_2,850 reaches an intermediate value on the surface of the SC. Then, the I_2,880/I_2,850 ratio increases with increasing SC depth and reaches a maximum value (the highest content of the orthorhombic phase of the lateral ICL organization, the most dense and least permeable ICL package) at a depth of 20–40% of the SC thickness. Further, the I_2,880/I_2,850 ratio decreases and reaches a minimum value (the highest content of the hexagonal phase of the ICL organization, less dense and more permeable ICL packing) in the bottom of the SC.

In older humans, these parameters are changed at exemplary SC depths toward increase of orthorhombic organization at 20–30% SC depth (p < 0.1 between 29 y.o. and 50 y.o. age-groups) [39], which was recently confirmed by Rigal et al. [60]. The lipid organization is changed toward decreasing of skin barrier function in the xerosis and atopic skin in vivo [32, 61]. Porcine skin, which is widely used as an ex vivo model of human skin in dermatological research, has more hexagonal organization of ICL at 10–50% SC depth (p < 0.01) in comparison with human SC in vivo [62]. The two depth profiles presented in Figures 1b and2b correlate and show an increased content of trans conformers in the ICL chains at the depth of 20–40% SC thickness, i.e., an increase of orthorhombic organization of ICL and enhanced skin barrier function at these exemplary SC depths.

The valence vibrations of the OH group of water originate from the broad and intense Raman band at ≈3,000–3,700 cm⁻¹. The water concentration in the SC is nonhomogeneously distributed, increases from the surface (minimal concentration) toward the bottom of the SC (maximal concentration), and is typically calculated by the intensity ratio of the 3,350–3,550 cm⁻¹ water-related and the 2,910–2,965 cm⁻¹ keratin-related Raman bands [34]. This method has the main assumption that the keratin concentration in the SC is homogenously distributed. However, it has recently been shown that the concentration of keratin decreases toward the bottom of the SC [63]. The correction of the nonhomogeneous distribution of keratin results in a slightly decreased water concentration at 50–100% SC depth (p < 0.05), which reaches its maximum of ≈61.5% at the bottom of the SC [38].

In the SC water exists in different mobility states, which are determined by the strength of hydrogen bonds of water with surrounding molecules, according to the principle: strong (weak) hydrogen bonds – strongly (weakly) bound water, absence of hydrogen bonds – unbound water (UBW) [35]. The concentration of bound water in the skin can be measured using differential scanning calorimetry [37, 64], proton magnetic resonance [65, 66], and Raman spectroscopy [45]. All measurements presented in these studies were performed on skin biopsies ex vivo. In particular, water associated with ICL and keratin was a research subject in determining the limit of physiological moisturizing of the isolated SC ex vivo using Raman spectroscopy. The Raman spectrum of the SC in the region ≈3,100–3,700 cm⁻¹ was decomposed using 4 Gaussian functions to determine the concentration of water mobility states [48]. The contribution of keratin in the Raman spectra was not taken into consideration.

To determine water mobility states in the SC in vivo, the contribution of keratin at ≈3,063 and ≈3,330 cm⁻¹ should be taken into consideration. For this reason, the decomposition in the ≈3,000–3,700 cm⁻¹ region was performed using six Gaussian functions, which is presented in Figure 3a. Two Gaussian functions describe keratin (at ≈3,063 and ≈3,330 cm⁻¹) and four remaining Gaussian functions describe water in different mobility states (at ≈3,005 cm⁻¹ – tightly-bound water [TBW]; ≈3,277 cm⁻¹ – strongly-bound water [SBW]; ≈3,458 cm⁻¹ – weakly-bound water [WBW]; and ≈3,604 cm⁻¹ – UBW) [59], according to Sun, who performed statistical interpretation of temperature-dependent structural transitions in water [67] and showed the change of hydrogen bonding of water depending on surrounding solvents [68] using Raman spectroscopy.

Calculating the area under the corresponding Gaussian function curve for the entire SC depth provides information about the distribution of water concentrations remaining in different mobility states, depending on the strength of the hydrogen bonds in the SC (Fig. 3b). Figure 3c shows the depth profile of the percent of TBW, SBW, WBW, and UBW normalized by total water content in the SC, which was calculated here as a sum of all water mobility states (TBW + SBW + WBW + UBW). As can be seen, SBW and WBW together represent ≥90%, while TBW and UBW represent the remaining ≤10% of the total water in the SC. Figure 3с shows that the summation of the percentages of all water states differs from 100%. This is explained by the normalization of the SC thickness to 100%, which included an interpolation procedure in 10% steps, as well as intermediate averaging for all measured SC thickness values and data for all investigated areas on the skin.

Using Raman spectroscopy Maeda and Kitano [69] determined the hydrogen bonding properties of water molecules in polymer systems analyzing a I_3,400/I_3,250 ratio. Thus, the WBW/SBW ratio represents a hydrogen bonding state of water molecules in the SC (Fig. 3d) – an important physiological parameter, which shows the nonhomogeneous bonding properties of water with surrounding molecules in the SC. A maximal efficacy to bind water with surrounding molecules was observed at ≈20–40% SC depth [59].

In older humans, these parameters are changed at exemplary SC depths toward enhancement of hydrogen bonding state of water molecules at ≈10–30% SC depth (p < 0.05 between 29 y.o. and 50 y.o. age-groups) [39]. Increase of tightly bound, unbound, and total water content in elderly skin was also reported [60]. The reduction of bound water in the SC was reported for atopic skin in vivo [61]. Porcine skin, which is widely used as an ex vivo model of human skin in dermatological research, has a lower hydrogen bonding state of water molecules at ≈10–30% SC depth (p < 0.01) in comparison with human SC in vivo [62].

The secondary and tertiary structures of keratin are important for the regulation of water binding in the SC. The β-sheet/α-helix and (β-sheet + turns and random coils)/α-helix ratios measured by the I₉₆₀/I₉₃₈ (area under the curve, I_952–966/I_924–946) and (I_1,670 + I_1,685)/I_1,655 (here, the Raman band intensities I are determined as area under the corresponding Gaussian curve after decomposition of the Amide I Raman band) ratios, contain information about the secondary structure of keratin. The α-helix is a stable form of keratin, containing a double helix and less exposed side chains, which causes a weak ability to interact with water molecules. The β-sheets contain a relatively large number of open side chains, which allow water molecules to freely penetrate between them and bind to keratin by hydrogen bonds. As shown in Figure 4a, the β-sheet/α-helix ratio is nonhomogeneous in the SC – decreases from the bottom (increase of the β-sheet form) toward the surface (increase of the α-helix form) of the SC. This behavior is correlated with the conformational changes assessed using the shape of the Amide III band at 1,200–1,250 cm⁻¹ [70] using nonnegative matrix factorization of Raman spectra.

Figure 4b–d show the depth distribution for the parameters describing the tertiary structure of keratin, such as stability of disulfide bonds (calculated by the I_474–508/I_474–578 ratio, Fig. 4b), buried/exposed configuration of tyrosine residues (calculated by the I_816–838/I_838–874 ratio, Fig. 4c), and the folding state of keratin (position of the Gaussian function at ≈2,930 cm⁻¹ after decomposition, Fig. 4d) in the SC [71]. The results presented in Figure 4b–d show the nonhomogeneous distribution of the parameters describing the tertiary structure of keratin, which has a direct relation to binding of water molecules within the corneocytes.

The results shown in Figure 4 for the secondary and tertiary structures of keratin indicate that keratin is in the folded state at the superficial region (≈0–30% SC depth) and, thus, unable to bind water molecules efficiently. Here, the concentration of NMF molecules is maximal [72], and they are mainly responsible for bonding water at the superficial SC region [71]. At the intermediate region (≈30–70% SC depth), concentration of NMF molecules is reduced, keratin fibers are in the unfolded state (most unfolded state is at a depth of 50–80% SC depth (Fig. 4d)) and, thus, keratin is mainly responsible for binding water molecules. In the bottom region (≈80–100% SC depth), the concentration of water is maximal, while the concentration of NMF molecules is minimal and bonding properties of keratin are reduced compared to the intermediate SC region (Fig. 4c, d). Thus, at the bottom SC region, less water is bound to keratin and NMF molecules, which is reflected in an increased concentration of UBW and WBW (Fig. 3c) as well as in a decreased hydrogen bonding state of water molecules (Fig. 3d). It appears that NMF molecules, keratin fibers as well as ICL have almost no free states and are completely saturated in the bottom region of the SC. In the case of skin occlusion, the binding of the water molecules is due to the free states of keratin, therefore an intermediate SC region determines the swelling of the SC, whereas other SC regions do not absorb water efficiently and do not swell [71]. An increased α-helix/β-sheet ratio (less exposed keratin) in atopic skin in vivo [61] may explain the lower water-binding properties compared to healthy skin.

In older humans, keratin is in more unfolded state compared to younger humans [39], which was recently confirmed by Rigal et al. [60]. Porcine skin, which is widely used as an ex vivo model of human skin in dermatological research, has more exposed secondary structure of keratin at ≈10–90% SC depth (p < 0.01), and a more folded tertiary structure of keratin at ≈30–100% SC depth (p < 0.05) in comparison with human SC in vivo, which determines binding of water inside the corneocytes [62].

Additionally to the composition, the noninvasive evaluation of the influence of topically applied cosmetic or medical formulations on the parameters of the SC is up to date and was previously only realized ex vivo using Raman spectroscopy [73-75]. The development of the above described algorithms allows the investigations of the influence of topically applied cosmetic or medical formulations on the components and physiological parameters of the human SC in vivo and noninvasively using confocal Raman microspectroscopy [76-79], which is in great demand for cosmetic and pharmaceutical industries.

Among most concentrated lipids, keratin, and water, characterized with strongly intense Raman bands, analysis of Raman spectra makes it possible to determine depth profiles of additional molecular components of the SC, such as NMF molecules, urocanic acid, urea, lactate [70, 72, 80, 81], as well as melanin and carotenoids [70, 82, 83]. Here, the components are evaluated using known spectral libraries [72], or without knowing their spectra using multivariate curve resolution [6] or nonnegative matrix factorization [70] statistical approaches. Carotenoids can be evaluated by their prominent Raman band at around 1,524 cm⁻¹ [70, 84]; melanin can be evaluated by its Raman bands at around 1,380 and 1,570 cm⁻¹ and by a strong fluorescence [83]. Tracking the DNA-related Raman band intensity at around 785 cm⁻¹, it is possible to evaluate the location of the skin microbiome [85], which is observed in the superficial SC region at a depth of 0–2 μm and is not detected at depths exceeding 4 μm. All these components can be determined noninvasively and in vivo in human SC.

Different algorithms in analysis of Raman spectra are proposed by many scientific groups worldwide to determine the penetration depth of topically applied formulations into the skin [70, 86-96]. These algorithms can be efficiently used to determine penetration profiles into the SC in vivo and ex vivo.

The results presented in this paper show that physiological parameters of the SC, such as lamellar and lateral organization of ICL, folding of keratin, water mobility, and hydrogen bonding states are nonhomogeneously distributed in depth and confirm that confocal Raman microspectroscopy can be successfully used for determination and analysis of physiological parameters of the human SC in vivo.

It has been shown that the skin barrier function maintained by the SC is nonhomogeneous in depth and is characterized by the strongest barrier function at ≈20–40% SC depth. In this SC region, the ICL has the highest concentration of trans conformers and orthorhombic organization, which determine skin barrier function. Analysis of the secondary and tertiary structures of keratin indicate that folding of keratin is nonhomogeneous throughout the SC: at the superficial region (≈0–30% SC depth, keratin is in folded state, while at intermediate and bottom regions (≈40–100% SC depth) keratin is in more unfolded state. This behavior determines the ability of keratin to bind water molecules – water is efficiently bound by unfolded keratin. Most efficiently, keratin binds water only at the intermediate region (≈30–70% SC depth), which determines swelling of the entire SC. Water in the SC is nonhomogeneously bound with SC components: the maximal bonding state of water molecules is observed at ≈20–40% SC depth. SBW and WBW represent together ≥90% of the total water in the SC. The remaining ≤10% of the total water in the SC is TBW and UBW. The depth-dependent inhomogeneity of the physiological parameters in the SC should be taken into consideration in planning measurements on the skin in vivo and on skin biopsies ex vivo.

The authors have no conflicts of interest to declare.

This research was supported by the Foundation for Skin Physiology of the Donor Association for German Science and Humanities. Funding for C.S.C. was provided by the German Academic Exchange Service (DAAD).

M.E.D., J.L., and C.S.C. planned and designed the study. J.S. and C.S.C. conceived and performed the experiments and analyzed the data. M.E.D. wrote the main manuscript text. All authors reviewed the manuscript.