A Stratum corneum lipid model as a platform for biophysical profiling of bio-active chemical interactions at the skin level

: Stratum Corneum (SC), the outermost layer of mammalian skin, provides the first and most challenging barrier to skin permeation of cosmetic and pharmaceutical compounds. SC barrier function relies primarily on the complex structure and organization of the intercellular lipid matrix. This matrix consists mainly of ceramides (CER), free fatty acids (FFA), and cholesterol (Chol) in an equimolar ratio, forming a multilamellar structure with short-and long-periodicity phases. Along with permeation studies, it becomes paramount the molecular investigation of the interaction between the lipid matrix and the compounds, towards a comprehensive picture of their biophysical impact, including lipid packing


INTRODUCTION
The administration of compounds through the skin is a highly esteemed route from both cosmetic and pharmaceutical point of view.The stratum corneum (SC), the outermost layer of mammalian skin, provides the first and most challenging barrier for the permeation of exogenous compounds [1].The barrier properties of SC are predominantly determined by its intricate structure and composition, commonly described as a 'brick-and-mortar' wall where corneocytes are embedded in an intercellular lipid matrix.This intercellular lipid matrix is mainly constituted by ceramides (CER), free fatty acids (FFA), and cholesterol (Chol) in a nearly equimolar ratio, forming a highly ordered multilamellar structure with repeated distances of 5-6 and 12-14 nm, corresponding to short-and longperiodicity phases (SPP and LPP), respectively [2][3][4].Additionally, SC lipids exhibit primarily orthorhombic and hexagonal lateral organization, with some regions of liquid-like lipid assembly.Through the SC, it is also possible to find a non-linear pH gradient between 4.5 on its surface and 7.0 in its lower regions [5][6][7][8].The complex structural assembly of the SC can be influenced either by pathological conditions or by the permeation of exogenous compounds.Although the transdermal route of permeation is still debated, the intercellular hypothesis, in which the exogenous molecule moves through the intercellular lipid matrix of SC, is the most widely accepted.Therefore, it is not surprising that chemical models, mainly lipid-based ones, with appropriate lipid mixtures that mimic the physiological SC composition and complex organization are gaining increasing popularity for different skin-related research topics.In the realm of skin barrier topics, numerous investigations have been focused on developing SC lipid models for a better understanding of its lipid matrix organization, exploring the location and the respective roles of individual components [9][10][11][12][13][14][15][16][17].Simultaneously, other studies are employing these models to investigate the interaction of external molecules with the SC lipid matrix.These investigations aim to either elucidate the chemical nature of the interaction [18][19][20][21][22][23][24] or to evaluate the potential of certain molecules to exert a specific effect, such as permeation enhancers, which are highly used in cosmetic and pharmaceutical applications [25][26][27][28][29][30][31][32].Particularly, lipid monolayers at the air/liquid interface offer valuable insights into the biophysics behind interactions occurring at membrane surfaces [33][34][35].Monolayers serve as excellent models due to their ability to provide complete and precisely controlled lipid coverage without defects and to eliminate the influence of external charges and roughness that can be exerted by solid substrates, as in the case of supported lipid systems [34].In the existing literature, SC lipid monolayers have been used for the biophysical characterization of the SC intercellular lipid matrix organization [36,37] and to study the effect of penetration enhancers [24,[27][28][29][30]38].In these approaches, SC lipid mixtures typically involve a CER:Chol:FFA equimolar ratio.However, they fall short in accurately mimicking the complex lipid composition of the SC lipid matrix, as they often feature only one type of CER per mixture and underestimate the importance of multiple-chained FFAs content [24,29,30,[37][38][39].Indeed, both the presence of FFAs with varying acyl chain lengths and the chain matching between FFAs and CERS have been recognized as promoters of miscibility, preventing phase separation [40][41][42][43][44].Moreover, none of these SC lipid monolayers account for the presence of CER[EOS], which has been identified as pivotal for the formation and stabilization of LPP as well as for proper SC barrier function [16,[45][46][47][48].
In the present study, to address these critical factors for proper SC barrier function, a SC lipid monolayer model was developed and biophysically characterized.For an accurate choice of lipid composition, several factors were taken into consideration.In the native SC lipid matrix, FFA's chain lengths range from C16 to C30, with a predominance of C24 and C26 FFAs (33.7 and 25.2 %, respectively) [49].Additionally, for skin barrier function, it is important to ensure the coexistence of CERs with phytosphingosine or sphingosine bases, for instance, with CER[EOS] with really long fatty acid chains that are important for LPP formation [50,51] and CER[AP] with its α-hydroxylated acyl chains for their involvement in intermolecular hydrogen bonding, which contributes to the cohesion of the SPP structure [39].Therefore, a mixture of both CERs as well as four different FFAs with chain lengths ranging from C20 to C26 were used here.Furthermore, the applicability of this SC intercellular lipid matrix model as a platform for investigating the interactions of bioactive compounds at the SC level was assessed in vitro using two relatively unstudied molecules, caffeine (CAF) and testosterone (TST), which are recommended by the OECD to be used as hydrophilic and lipophilic model compounds, respectively.The chemical structures of the lipid components of the SC model and the model compounds, CAF and TST, are presented in Figure 1.

Preparation of SC model monolayers
Lipid SC model solution was prepared in chloroform:methanol (3:1) at 1 mg•mL -1 for an equimolar ratio of CER:Chol:FFA.CER consisted of a mixture of 70 %(m/m) CER[EOS] and 30 %(m/m) CER [AP].FFA contained a mixture of AA:BA:LA:HA in 9:48:38.5:4.5 %(m/m).Monolayers of the SC model were obtained after spreading 80 μL of SC model solution, by using a Hamilton microsyringe, on acetate buffered (pH 5.5) subphases, either pure or with CAF or TST in solution.The concentrations of CAF and TST in the subphase (CAF at 300 μM and TST at 10 μM) were chosen to comprise with sink conditions, i.e. to assure that the concentration of compounds used does not exceed 10 to 20% of its maximum solubility following EMA guidelines [64].After spreading, the monolayers were left to equilibrate for ca.15 min for solvent evaporation prior to film compression (barrier speed of 15 cm 2 •min -1 ).

Surface pressure-area isotherm measurements
Surface pressure-area isotherms (π-A) were measured in a PTFE Langmuir trough equipped with one moving barrier (NIMA 611D) and a Wilhelmy microbalance with filter paper plate (accuracy superior to 0.1 mN/m) for measuring the surface tension of the monolayer covering surface.Before each experiment, cleanliness was confirmed by compressing the water subphase and keeping the surface pressure reading.All the measurements were performed at room temperature (≈23 °C).
The monolayer compressibility modulus (C s -1 ) is a measure of the film elasticity and compressibility and can be calculated from the π-A isotherms by equation 1: where A 0 represents the area per molecule in the condensed state, and represents the change in area per molecule with the surface pressure under isothermal conditions.Generally, lower C s -1 values are associated with lower interfacial stiffness (or higher elasticity) [65].

UV-Vis reflection spectroscopy
UV-Vis reflection spectra at normal incidence of SC monolayers under compression or after CAF and TST injection underneath a pre-formed monolayer (NIMA 611D Langmuir trough) were obtained with a Nanofilm Surface Analysis Spectrometer (RefSPEC 2 supplied by Accurion GmbH, Goettingen, Germany).The spectra were recorded with an average of 15 and integration of 100 ms and before monolayer spreading, at least 3 baselines were recorded for further adjustments when necessary.

Brewster Angle Microscopy (BAM)
Morphological images of the monolayers were obtained with a Brewster Angle Microscope (BAM), model EP4 (Nanofilm Accurion GmbH, Göttingen, Germany), settled onto the computer-interfaced Langmuir trough equipped with two moving barriers (KSV NIMA, Biolin Scientific).The angle of incidence was set at 53.2°, and BAM images were captured during monolayer compression at selected surface pressures or to monitor the effect of CAF and TST injection over time.

Fourier Transform Infrared (FTIR) Spectroscopy
Infrared measurements were recorded in an Alpha-T Transmittance Fourier Transform InfraRed spectrometer (Bruker, Massachusetts, USA).The quartz substrates were cleaned according to the following procedure prior to the deposition of any monolayer: a) immersion of the substrates in Extran soap/water solution (1: The multilayered SC model deposited on quartz substrate was analyzed through FTIR by recording 32 absorbance scans per spectrum in the spectral range of 4000 to 500 cm -1 , with a resolution of 4 cm -1 , using the OPUS software.The background spectrum was taken for the bare support of the same material.

Computer simulations
CER[EOS], CAF, and TST molecules have been used with null total charge for the complete set of calculations.The starting geometry of all molecules was optimized using the PM3 semiempirical method.A refined optimization (energy gradient < 0.001 kcal•Å −1 •mol −1 ) of all molecules was performed prior to the qualitative simulation of interactions between either CAF or TST with the CER[EOS] using molecular mechanics.The CAF and TST molecules were allowed to reach different orientations with respect to the CER[EOS] molecules.
Four CER[EOS] molecules were included, using a 1:1 molar ratio to CER/TST.HyperChem 8.0 ® was used to perform the simulations.

Surface pressure-molecular area (π-A) measurements
The π-A isotherms of the SC model spread on different acetate buffered pH 5.5 subphases (pure, CAF at 300 μM and TST at 10 μM) are presented in Figure 2A.These isotherms provide thermodynamic and structural insights into the behavior of the SC model in the absence and presence of both bioactive molecules.The π-A isotherm of the SC model on pure buffer subphase (Figure 2A, black line) displayed a sharp transition from liquid expanded (LE) to liquid condensed (LC) phases over small areas per molecule range.These isotherms are typical of condensed and rigid film monolayers and consistent with previously reported isotherms for avian hoopoe lark SC [66] and synthetic CERs at the air/liquid interface [67].The acyl chains in an all-trans conformation contributed to this condensation level, leading to collapse at ≈44.43 Å 2 /molecule and a surface pressure of ≈43.64 mN/m.Note that the shape of the isotherms points to a miscible lipid monolayer, since commonly immiscible monolayer systems present multiple collapses, which corresponds to each individual value of collapse pressure for the pure component [68].Therefore, the presence of a single collapse point indicates a rather stable mixture even at high surface pressures [69].From the isotherm, at the most condensed state, the limit area, A lim , was estimated by extrapolating the linear part of the isotherm to the zero surface pressure [69].For the SC monolayer, an A lim of ≈53 Å 2 /molecule and an onset condensation area of ≈61 Å 2 /molecule were determined.Figure 2B plots C s -1 as a function of surface pressure for the SC monolayer in each subphase.The C s -1 values give additional insight into the compressibility of the monolayer, with the highest value corresponding to the most compressed state [68].The SC model exhibited a maximal C s -1 value of 280 mN/m at a surface pressure of π = 31.60mN/m, further confirming the LC phase of the monolayer, which presents high condensation, strong intermolecular interactions, and lower compressibility [39,68,70].
Upon the addition of CAF or TST on the buffered subphase, some deviations from the SC model behavior were observed.In the case of the SC model spread on the CAF subphase, the isotherm maintained the expected sharp transition from LE to LC while shifting it towards lower areas per molecule (Figure 2A and  2C).This decrease in SC model area per molecule in the presence of CAF can be associated with material removal from the lipid/air interface or changes in intermolecular polar headgroup interactions, which result in decreased repulsion between adjacent headgroups.These reported changes did not have a significant impact on the collapse pressure (≈44.70 mN/m at ≈40.37 Å 2 /molecule).Instead, an A lim value of around 50 Å 2 and a reduced onset condensation area of ≈58 Å 2 were determined, corresponding to a decreased C s - 1 value of ≈240 mN/m at π= 35.18 mN/m.These decreased values suggest increased compressibility of the LC phase of the SC monolayer [69][70][71].
Conversely, the SC model in the TST subphase exhibited changes in the isotherm shape and a shift towards higher areas per molecule (Figure 2A and  2C).This shift is due to increased material at the lipid/air interface, suggesting the incorporation of TST into the SC monolayer [71,72].Despite a collapse pressure that remained practically unaltered (≈44.44 mN/m and ≈48.13 Å 2 /molecule), the reduced condensation effect induced by TST was evident through an increased onset area value of 88.36 Å 2 /molecule and an increased A lim to approximately 64 Å 2 .This increased A lim , with the decreased maximum C s - 1 ( ≈217 mN/m at π≈ 38 mN/m) indicated that the interaction of TST with the SC lipids resulted in a more compressible monolayer [70].
Moreover, in Figure 2C the area per molecule at different values of surface pressure is plotted.Interestingly, while the difference in areas per molecule of the SC model in the pure buffered subphase or in the CAF subphase is constant, for the TST subphase there is a trend to decrease this difference along the compression.Furthermore, it is worth noting that while CAF isotherms occur in a similar range of molecular areas, the isotherm of TST occurs in a wider range of areas per molecule.These results highlight that CAF penetrates within the SC monolayer both in a more condensed or expanded state, whereas TST penetration within the SC monolayer is higher at lower surface pressures (i.e., at a liquid expanded state).

UV-Vis reflection spectroscopy studies
The UV-vis reflection under normal incident light from the SC monolayer on the different subphases at different surface areas was monitored simultaneously with the π-A isotherm.In Figure 3, the integral values of reflection are displayed as a function of the area per SC molecule on buffered CAF or TST subphases (Figure 3A or 3B, respectively).The integral ΔR values presented in Figure 3 have been normalized by the area per molecule to eliminate the influence of surface concentration on the reflection spectrum [73,74].Therefore, variations in the normalized values of integral with the compression of the monolayer are indicative of relative changes in the surface concentration of the chromophore (CAF or TST) per lipid molecule.In both graphs (Figure 3A and 3B) an increased ΔR signal with compression is highlighted, suggesting increasing amounts of either CAF or TST at the lipid/air interface.This presence of the chromophores at the interface points to the ability of both compounds to permeate through the SC monolayer and their interaction with polar headgroups either by dipole-dipole or hydrogen bond interactions.The evidence of CAF presence at the interface supports the hypothesis raised before that CAF is promoting changes at intermolecular polar headgroup interactions.On the other hand, the detection of TST through UV-Vis reflection supports the previously proposed hypothesis of TST incorporation into the SC monolayer.

Compression-relaxation cycles
The degree of reversibility of the π-A isotherms for a given monolayer can be investigated using compression-relaxation π-A isotherm cycles [75].In Figure 4, three consecutive compression-relaxation (full or dashed lines, respectively) isotherms of SC monolayers in buffered subphase at pH 5.5 are presented in the absence (black and grey shades) or in the presence of CAF (purple shades) or TST (blue shades).The three cycles of compressing and relaxing the SC monolayer on buffered subphase (Figure 4, black and grey shades) demonstrated almost complete reversibility, except for a shoulder located at ≈52 Å 2 , 20 mN/m.This shoulder is characteristic of traces of a mixture of LE+LC phase coexistence during the LE to LC transition.It is also interesting to note that this mixture of LE+LC phases tends to disappear with consecutive compression-relaxation cycles.
The isotherms obtained from the compression-relaxation cycles of the SC monolayer in the CAF subphase appeared at lower areas per molecule, with complete reversibility, which suggests that CAF is interacting only at the polar headgroup level of the SC model.As hypothesized above, the isotherm shift to lower areas per molecule indicates that CAF interaction with the SC model might be inducing material loss from the lipid/air interface or reducing intermolecular dipole-dipole repulsion of the lipid headgroups.From the complete reversibility of the compression-relaxation isotherm, it becomes clear that CAF interaction at the headgroup region of SC models is more likely to result in changes at the intermolecular dipole-dipole interaction than in lipid loss at the interface.Indeed, if CAF was promoting the withdrawal of lipid material from the air/water interface, it would be expected that through compression-relaxation cycles, the isotherm would be shifted to lower values of area per molecule.However, the consecutive isotherms are completely overlapping.Moreover, the absence of the LE+LC phase mixture induced by CAF, as suggested by the disappearance of the shoulder, supports the hypothesis that CAF interacts at the headgroup region, resulting in a screening of dipole-dipole repulsion.
In the cycled isotherms of the SC model monolayer on the TST subphase, it is possible to observe a significant difference between the first compression (Figure 4, darker blue full line) and all the other isotherms completing the compressionrelaxation cycles.Additionally, the shoulder correspondent to the LE+LC phase coexistence during the LE to LC transition is, in this case, more prominent than for the SC model in the absence of any bioactive in the subphase.These observations suggest that TST is somehow participating in SC monolayer and provoking structural changes.In the second and third cycles, the reversibility is not complete, suggesting that TST is interacting not exclusively with the headgroups of the SC monolayer but also with their hydrocarbon chains.
The occurrence of hysteresis under monolayer compression-relaxation is a common feature and is related to the stability of closely packed states formed under compression and their tendency to slowly undergo reversible reorganization to their initial state under expansion [76,77].Therefore, it is also interesting to note that the SC model in the absence of any bioactive showed the smallest hysteresis area, while it slightly increased on the CAF subphase.This slight difference can be associated with the CAF promoting increased lipid packing, which hinders the SC model reorganization to the initial state.Contrarywise, on the TST subphase, the hysteresis differs between cycles, and it can be observed that there is a significantly higher hysteresis area (compared to the SC monolayer) in the first cycle and almost no hysteresis in the following cycles.The TST large hysteresis area in the first cycle might be explained by the bioactive penetration within the monolayer and its bulky nucleus, which assumes a vertical orientation between adjacent SC model lipids.The subsequent decreased area in the following cycles is in line with the rearrangement of TST into the monolayer.

Brewster Angle Microscopy
BAM provides valuable insights into the in-plane structure of SC monolayers by acquiring images at different surface pressures along with the isotherms [29].
Figure 5A presents BAM images of SC model monolayers at skin relevant pH of 5.5 in the absence and presence of CAF or TST at different surface pressures (≈0, 1, 5, and 30 mN/m).Phase contrast in BAM images results from differences in the reflectance of ppolarized light between the monolayer and the subphase.Brighter regions are typically associated with a higher density of film components, e.g., more condensed phases, whereas the darker background represents less condensed phases [39,65].As a result, condensation level histograms based on the imaged pixel greyscale were generated for each discussed surface pressure (Figure 5B).
In the first column of Figure 5A, the morphology of the SC monolayer in acetate buffered subphase (pH 5. associated their more densely packed and more ordered region with a FFAenriched phase, and the darker more compressible domains with a less ordered CER and Chol phase [65].As the monolayer is compressed, it is observed a gradual homogeneity between both phases, consistent with a uniform chain orientation, until at ≈30 mN/m brighter spots appeared. The presence of CAF on the subphase did not dramatically modified monolayer morphology, but at ≈0 mN/m an increase in condensed domains, even before the compression, is visible (Figure 5A).As can be seen by the histogram in Figure 5B, at larger areas (i.e., ≈0 mN/m) the sharp peak at lower condensation levels of SC gives rise, upon interaction with CAF, to a double broad peak of more condensed and less condensed phases.The observed formation of condensed phases in the SC corroborates the CAF effect, decreasing repulsion between adjacent SC headgroups as suggested by the isotherms.Then, as the monolayer is compressed to about 1 mN/m, the condensed domains increase at the expense of less condensed ones (Figure S2, phase 1 and 2, respectively, supplementary materials), as expected.At about 5 mN/m, the changes in condensation level resultant from the interaction of SC lipids with CAF are less visible, and at about 30 mN/m it is notorious the existence of a smaller quantity of brighter spots.These observations suggest an increased homogeneity of the SC model monolayer in the presence of CAF.
TST presence did not change the morphological picture of both more and less condensed phases at larger molecular areas (about 0 and 1 mN/m) (Figure 5A).However, when the monolayer is compressed to about 5 mN/m, morphological differences in the more condensed phases are notorious (Figure S3, phase 1', supplementary materials), although these cannot be discriminated through condensation level histograms (Figure 5B).These morphological differences can arise either from a phase segregation promoted by TST in the more condensed phases or even from a miscible effect in some component enriched phases.Whichever, at 30 mN/m brighter branched-like structures appeared that can be due to the presence of TST clusters within the lipids or TST-rich phases resultant from a TST effect of phase segregation.

FTIR analysis
FTIR analysis was conducted to investigate the lateral packing and conformational characteristics of the SC model in the absence and presence of CAF or TST. Figure 6 displays the most relevant IR ranges selected for analysis.In the IR range spanning from 2800 to 2980 cm -1 distinct peaks corresponding to the symmetric and asymmetric CH 2 stretching vibration (ν s (CH 2 ) and ν as (CH 2 ), respectively) of SC lipids are observed, typically occurring around ≈2920 and ≈2850 cm -1 , respectively.In the IR spectra of SC model, these peaks are found at ≈2917 and ≈2849 cm -1 , respectively.The positions of CH 2 stretching vibrations furnish valuable insights into the conformational ordering of acyl chains.Specifically, symmetric positioning below ≈2850 cm -1 is indicative of highly ordered hydrocarbon chains, primarily composed of all-trans conformers [78,79].Conversely, when these vibrations are symmetrically located between ≈2916-2917 cm -1 , it signifies a lipid gel phase [80].The presence of either CAF or TST on the subphase did not alter the positioning of either of these peaks, indicating that the chain packing and the all-trans conformation is not influenced by their presence on the subphase [29].
Methylene scissoring and rocking bands (δ(CH 2 ) and ρ(CH 2 ), respectively) are sensitive to lateral chain packing [78,81].Concerning the ρ(CH 2 ) frequencies between ≈730 and ≈719 cm -1 , while the presence of a singlet around ≈721 cm -1 is characteristic of a hexagonal packing; the presence of a doublet at ≈720 and ≈730 cm -1 is characteristic of an orthorhombic lateral organization.The split in a doublet occurs due to the short-range vibrational coupling of the lipid chains, and since hexagonal chain lattice is less densely packed, no short-range coupling occurs [81,82].On the other hand, δ(CH 2 ) is usually affected by lipid packing assemblies in ordered phases and while broadening or splitting contours suggest orthorhombic lateral packing (≈1462 and ≈1473 cm -1 ), a single peak is usually assigned to hexagonal packing (≈1467 cm -1 ) [83].Both δ(CH 2 ) and ρ(CH 2 ) bands are presented in Figure 6 for SC model in absence (Figure 6A) or presence of CAF (Figure 6B) or TST (Figure 6C).SC model in acetate buffer subphase displayed a single δ(CH 2 ) peak around 1467 cm -1 , indicating the presence of lipid domains hexagonally packed.However, by the overlapped ρ(CH 2 ) singlet at 723 cm -1 with ρ(CH 2 ) doublet at 716 and 731 cm -1 , the coexistence of domains orthorhombically and hexagonally packed is irrefutable.With the addition of CAF, both broad single peak at 1467 cm -1 , as the presence of ρ(CH 2 ) doublets at 720 and 731 cm -1 remain, suggesting the coexistence of hexagonal and orthorhombic lateral packing.Conversely, the presence of TST on the subphase has split the δ(CH 2 ) into two peaks at 1464 and 1472 cm -1 , which are assigned to orthorhombic lateral packing, while maintained traces of coexistence of both lateral arrangements in the SC system through the presence of a broader doublet ρ(CH 2 ) at around 730 cm -1 .The traces of coexistence of both lateral packing can be responsible for the ρ(CH 2 ) displacement to lower wavelengths.
The regions of IR spectra related to amide I and II bands (from ≈1550 to ≈1650 cm -1 ), are important to investigate the hydrogen bonding at the headgroup level [84].Amide I band, which is related to C=O stretching vibration and N−H in-plane bend, appeared in IR spectra of the SC model as an asymmetric peak with a shoulder deconvoluted to three components positioned at ≈1612, ≈1620 and ≈1630 cm -1 (Figure 6A).In general, low wavelengths are related to the presence of strong hydrogen bond, while high wavelengths are assigned to weaker hydrogen bonds [85].From the three components it is clear that the amide I band is dominated by the positioned at ≈1630 cm -1 , that was previously identified in other hydroxy fatty acids CER [86,87] (as CER[AP] here).Moreover, recently an amide I band of a SC mixture was found to be positioned at 1624 cm -1 [84], which is close to at least two components identified here.On the other hand, amide II band is primarily assigned to N−H bending vibration and C−N stretching vibration [84].In this wavelength range, the IR spectra of the SC model presented one asymmetric peak followed by another peak, which resulted in three components at ≈1564, ≈1571 and ≈1582 cm -1 .From a general point of view, the relative lower amide I with higher frequencies of amide II indicates strong hydrogen bonds at the headgroup level.When SC model is in the presence of CAF (Figure 6B), modifications at either positioning and contour occur: amide I band results in a broad asymmetric peak centered at ≈1656 cm -1 with four components located at ≈1633, ≈1644, ≈1652 and ≈1659 cm -1 ; and amide II band presented two consecutive peaks, with one of them presenting two shoulders leading to a total of 4 components at ≈1549, ≈1564, ≈1572 and ≈1580 cm -1 .Therefore, CAF provoked an upward shift in the amide I band and a downward shift in the amide II band, which can be translated by weakening the SC headgroup hydrogen bonds [84,85].On the other hand, the presence of TST on the subphase of the SC model (Figure 6C) resulted in three components of amide I located at ≈1606, ≈1615, and ≈1624 cm -1 and of the amide II located at ≈1556, ≈1568, and ≈1575 cm -1 .The resultant downward shift on both amide bands and the fact that amide I showed less intensity than amide II demonstrates that although TST affects the headgroup interactions, the hydrogen bonds are still strong [84].In the present study, a lipid model resembling the composition of the intercellular lipid matrix of SC was developed and characterized in the form of a monolayer at the air/buffer interface.The complex composition of the SC model comprised two different types of CERs (CER[EOS] and CER[AP]), Chol and a mixture of FFA at an equimolar ratio.The SC mimetic mixture resulted in a highly condensed monolayer with acyl chains arranged in an all-trans conformation, with traces of a LE+LC to LC transition that disappear after the first compression-relaxation cycle.BAM acquired images presented domains with different condensation levels, which can be related with the coexistence of orthorhombic and hexagonal packing also detected by FTIR analysis.Typically, in SC model systems, FFAs were found to be an orthorhombic packing promotor, in contrast to hexagonal packing often observed in CER:Chol domains [45,88].The monolayer in the lipid gel phase presented a high value of C s -1 , which is intimately related to a low compressible monolayer due to a strong hydrogen bonding network involving the lipid headgroups [89], as determined by the amide I and II bands IR vibrations.The hydrogen bonding network within the SC lipid matrix is fundamental for the stabilization and formation of dense lateral packing within the lipid matrix [90][91][92][93].
All the SC lipid matrix components participate in this hydrogen bonding network.
CERs are sphingolipids with two (CER[EOS]) to four hydroxyl (CER[AP]) groups and a mono-substituted amide group, serving as both hydrogen bond donors and acceptors [94].Phytosphingosines, such as Cer[AP], tend to exhibit hexagonally ordered chains with strong headgroup hydrogen bonding, while sphingosines, like CER[EOS], are more likely packed in an orthorhombic chain lattice with weaker hydrogen bonding [95].Since water molecules are preferentially hydrogen bonded to the carbonyl oxygen of CERs, the hydroxyl groups present in CERs headgroup preferentially form hydrogen bonds with the headgroups of neighbor lipid molecules [89].Oxygen atoms of FFAs are available for hydrogen bonding with adjacent lipids, particularly with dominant CER molecules, as FFAs are not able to form hydrogen bonds among themselves [96].Even Chol molecules at the ester bond level of CERs contribute to the hydrogen bonding network via the hydroxyl group of Chol and the carbonyl group of neighboring CERs [97].
In the current work, the results point to CAF location in the headgroup region, at the air/lipid interface, where it significantly modifies the intermolecular interactions among adjacent SC lipid headgroups.These results agree with other studies demonstrating CAF partitioning into lipid model membranes (but not in SC models) with a location at the headgroup region and at the headgroup-acyl chains interface [56][57][58].The non-planar structure of CAF, due to methyl groups, leads to the impossibility of CAF acting as a typical hydrogen bond donor.Owing to its high polarity, CAF predominantly interacts with hydrophilic headgroups, potentially increasing their packing [98], which is consistent with the decreased area per SC lipid observed in the π-A isotherms.CAF enhances the strength of headgroup intermolecular interactions [99], leading to a condensing effect on the monolayer with a reduced area per SC lipid.It has been reported that CAF influences membrane hydration by attracting water molecules to its vicinity, leading to the formation of water pockets and overall membrane dehydration and thickening [56].CAF can only act as a hydrogen bond acceptor and typically interacts with water through two oxygen atoms and a nitrogen atom [100], promoting membrane dehydration by maybe displacing the water molecules from the lipid headgroups.Therefore, the inverse relationship between headgroupsubphase hydrogen bonding and headgroup-headgroup hydrogen bonding [101] and consequently the lack of tightly bond water molecules results in a tight packing of CERs with neighboring lipids, and in the formation of domains with reduced mobility [102].The restriction of CAF molecules to the lipid/water interface, as exemplified by the molecular simulation sketch in Figure S4 (supplementary materials), decreases the repulsion between polar headgroups and leads to decreased intermolecular distances, allowing hydrophobic chains to spread.
Existing data indicates that in a lipid membrane environment, steroid hormones such as TST exhibit a preference for a parallel orientation to the membrane normal.They tend to position both the carbonyl and hydrophobic core within the hydrophobic environment, with hydroxyl group protruding into the water phase [103,104].This ability of TST to integrate between lipids and behave as part of the lipid system aligns with the findings of the present study.The presence of TST on the subphase resulted in a monolayer with increased values of area per lipid, and UV-Vis detected TST at the air/lipid interface.Furthermore, the incomplete reversibility of the compression-relaxation cycles suggested interactions of TST at both the headgroup and acyl chain levels.At the headgroup region, the presence of the hydroxyl group from TST strengthen the hydrogen bonding network, as observed by the FTIR results.While the hydrophobic core of TST at the acyl chains level did not impact the all-trans conformation, TST promoted phase separation, possibly overcoming packing constraints and maintaining the coexistence of hexagonal and orthorhombic packing.However, it remains challenging to discern whether the seemingly more homogeneous phase in BAM images represents the TST-rich phase, with improved miscibility, or the non-TST phase.TST molecules' ability to penetrate deeply into the monolayer and interact with SC lipid molecules, as exemplified in molecular simulations (Figure S4, supplementary materials), results in higher values of area per molecule observed in the isotherms.
The ultimate objective in developing SC lipid models is to create a simple yet descriptive platform resembling the primary permeation route for most chemical compounds administrated through the skin.The passage through SC, following the intercellular route of its lipid matrix, involves the partitioning of the compound into the lipid environment, diffusion through this lipophilic environment, and subsequent partitioning from the SC into the viable epidermis [105].
Consequently, the ideal compound should possess balanced lipophilichydrophilic properties to facilitate partitioning into the SC and from the SC into the viable epidermis.In a multi-center comparative in vitro absorption study of CAF and TST through human and rat skin, higher maximum absorption rates were observed for CAF compared to TST, with a greater presence of TST at the skin membrane after 24 h [61].Similarly, a multiparametric in vitro permeation study through Minipig ® skin membranes found a trend over all the tested concentrations of CAF and TST in liquid solution [106].After permeation studies, the bioactive content present on SC, skin membrane, and receptor fluid was quantified.The authors reported that a lower percentage of applied TST was detected in the receptor fluid at the expense of a higher rate of applied TST at both SC and skin membrane compared to CAF content distribution [106].All these reported findings are aligned with the results reported here, including with the injection studies of CAF or TST beneath a preformed SC lipid monolayer in acetate buffer at 30 mN/m (resembling the molecular packing of biological membranes [24]).
Both CAF and TST demonstrated the ability to be detected at the air/lipid interface without causing visible effects on the SC model system (Figure S5, supplementary materials).UV-Vis spectroscopy monitoring over a 7-h period post-injection revealed distinctive patterns for CAF and TST.TST exhibited rapid detection, followed by stabilization throughout the entire monitored duration.In contrast, CAF detection gradually increased over the 7-hour observation period.This observation suggests differential behavior in the passage of these compounds through the SC lipid matrix, highlighting the distinct kinetics of TST and CAF in this experimental context.Because CAF is a hydrophilic molecule capable of permeating through the lipid matrix by inducing changes in the hydrogen bonding network, it is more likely to partition from the SC lipid environment to the viable epidermis, resulting in enhanced skin permeability.In contrast, TST, as a lipophilic compound that incorporates into the lipid structure, exhibits greater resistance to leave the lipid environment, leading to its prolonged entrapment within the lipid matrix of the SC compared to CAF and resulting in lower skin permeability.

CONCLUSIONS
In the present study, we developed and characterized a lipid-based model mimicking the barrier properties of the SC intercellular lipid matrix.The aim was to propose this model as an innovative in vitro platform to explore the permeation and interactions of bioactives at the skin level.The data from biophysical, chemical, and topographic/morphological characterization proved that the SC model can resemble some barrier properties of the SC lipid matrix.Additionally, CAF and TST were used as model compounds, following the OECD guidelines.Despite its hydrophilic nature, CAF reached the air/lipid interface, and its presence promoted enhanced intermolecular headgroup interactions and increased the condensed state of the monolayer.TST, as a lipophilic model, integrated into the lipid system, impacting both headgroup and acyl chains.Therefore, the SC model monolayer has proven to be a promising approach for in vitro investigation of skin permeation and the inherent interactions of bioactive compounds or nanoparticles for administration through the skin.This approach holds particular promise owing to its ability to be compositionally modulated to investigate these interactions in different skin conditions where the barrier properties of the lipid matrix are altered, making it a valuable tool for the pharmaceutical and cosmetic industries.

Figure 2 .
Figure 2. (A) π-A isotherms of SC model monolayers in acetate buffer (pH 5.5) in the absence (black) and presence of CAF (purple) or TST (blue) on the subphase; (B) Compressibility modulus, Cs -1 , as function of surface pressure; and (C) plot of the area per molecule as function of surface pressure.The arrows are indicative of the phase transition to liquid condensed phase.

Figure 3 .
Figure 3. Values of the integral of the UV-Vis reflection spectra at different values of molecular area of SC model monolayer on buffered CAF subphase (A) and TST subphase (B), obtained simultaneously with π-A isotherm.

Figure 4 .
Figure 4.The compression-relaxation isotherms of SC model in acetate buffered subphase at pH 5.5 in the absence (black/grey shades) or in presence of CAF (purple shades) or TST (blue shades).Full or dashed lines represent the compression or relaxation, respectively.

Figure 5 .
Figure 5. BAM images acquired along the SC monolayer compression on the different subphases (A), and respective condensation level histograms based on the pixel greyscale (B).

Figure 6 .
Figure 6.FTIR regions from the spectra of SC model in buffered subphase in the absence (A) or presence of CAF (B) or TST (C): From the left to right, are presented the IR region of ν s (CH 2 ) and ν as (CH 2 ) modes, the δ(CH 2 ) mode, ρ(CH 2 ) mode and CER amides I and II bands.
[65]presence of domains constituted by irregular branch-like structures in bright grey dispersed in a darker continuous domain (FigureS1, phases 1 and 2, respectively, supplementary materials) is in line with what was previously described by Mao et al. for a different SC biomimetic mixture[65].Unfortunately, BAM images lack chemical identification, and thus, it is not possible to differentiate the chemical composition of the domains.Mao et al.
5) at ≈0 mN/m is displayed prior to compression.At larger values of molecular area, the acquired images clearly show the coexistence of multiple phases.At early stages of compression and surface pressure of ≈1 mN/m, the entire field of view was covered by the SC model monolayer with different condensation level regions, represented by different bright contrasts.