Altered structure indicating reduced barrier function of lesional compared to non‐lesional psoriatic skin—A non‐invasive in vivo study of the human stratum corneum with confocal Raman micro‐spectroscopy

Psoriasis, one of the most common skin diseases affecting roughly 2%–3% of the world population, is associated with a reduced skin barrier function (SBF) that might play an important role in its pathophysiology. The SBF is provided primarily by the stratum corneum (SC) of the skin. Previous studies have revealed a higher trans‐epidermal water loss, lower hydration, abnormal concentration and composition of intercellular lipids, as well as alterations in secondary keratin structure in the psoriatic SC. We compared on molecular level lesional psoriatic skin (LPS) with non‐lesional psoriatic skin (nLPS) from 19 patients non‐invasively in vivo, using confocal Raman micro‐spectroscopy. By analysing the corresponding Raman spectra, we determined SBF‐defining parameters of the SC depth‐dependently. Our results revealed a lower total lipid concentration, a shift of lamellar lipid organisation towards more gauche‐conformers and an increase of the less dense hexagonal lateral packing of the intercellular lipids in LPS. Furthermore, we observed lower natural moisturising factor concentration, lower total water as well as a strong tendency towards less strongly bound and more weakly bound water molecules in LPS. Finally, we detected a less stable secondary keratin structure with increased β‐sheets, in contrast to the tertiary structure, showing a higher degree of folded keratin in LPS. These findings clearly suggest structural differences indicating a reduced SBF in LPS, and are discussed in juxtaposition to preceding outcomes for psoriatic and healthy skin. Understanding the alterations of the psoriatic SC provides insights into the exact pathophysiology of psoriasis and paves the way for optimal future treatments.


| INTRODUC TI ON
Psoriasis is a chronic inflammatory skin disease, affecting approximately 2%-3% of the worldwide population of various age and sex. 1,2The pathophysiology of the disease involves environmental, genetic and immunologic aspects, 3,4 with recent research also revealing an altered microbiome 5 and epigenetic dysregulation. 6In the most common clinical manifestation, psoriatic lesions appear as well demarcated erythematous plaques with coarse scales, which are characterised by an abnormal and incomplete proliferation and differentiation of keratinocytes resulting in increased thickness of the epidermis, corresponding to the histological findings of parakeratosis, acanthosis and hyperkeratosis. 2,7,8oriasis is associated with a disrupted skin barrier function (SBF), however, it is a matter of debate [9][10][11] whether this dysfunction is of primary or secondary pathogenetic nature, with recent evidence suggesting rather a combination of both. 12Thus, inflammation is one cause of barrier dysfunction, for example, by disrupting the physiological production and maintenance of SBF-related constituents.
On the other hand, genetic variations in psoriasis are also known to affect the composition and structure of the epidermis, such as cholesterol and ceramide production, which in turn lead to inflammation, higher susceptibility to environmental influences and potentially induce the psoriatic phenotype. 11,13Therefore, to understand the exact pathophysiology of psoriasis, it is instructive to examine the structure and SBF of psoriatic skin on molecular level.
The physiologic SBF is primarily provided by the stratum corneum (SC) and the interactions of its structural components. 9,14,15ong the most important components are the intercellular lipids (ICL), which consist of a nearly equal ratio of free fatty acids (FFA), cholesterol and ceramides 16 and are stacked together in form of bilayers inside lamellae.Depending on the amount of trans-and gauche-conformers, the lateral organisation of lipids is a combination of orthorhombic (ordered, very densely packed lipids) and hexagonal (ordered, less densely packed lipids) phases, which are inhomogeneous in the SC depth 17,18 and essential in maintaining the integrity of the SC and SBF. 10,19A further important SBF-related component are the corneocytes, bound together by corneodesmosomes and located in a water-lipid extracellular matrix (brick-and-mortar model of the SC). 9,20The corneocytes entail stabilising keratin filaments, mostly of the types K1/10, 21 which can vary between less or more folded conformations, as well as various amino acids with strong hygroscopic ability, forming the natural moisturising factor (NMF).
Both the folding of keratin and the NMF concentration are primarily responsible for binding water within the corneocytes in a depthdependent manner, maintaining the physiological SC hydration. 22ter diffusion into and out of the epidermis is controlled by the SC. 23One of the most commonly examined parameters describing the SBF in vivo is the trans-epidermal water loss (TEWL), which correlates with the orthorhombic organisation of the ICL. 24vertheless, TEWL might not be in every case a reliable parameter for the inside-outside barrier, especially in diseased and topically treated skin. 9Further relevant parameters are the SC hydration and the different water mobility states.[28] Several methods have been used to examine the psoriatic SC, like TEWL, corneometry, 'attenuated total reflectance-Fouriertransform infrared spectroscopy', 29,30 tape-stripping, 31 multiphoton CARS tomography, 32 optical coherence tomography, 33 diffuse reflectance spectroscopy, 34 scanning electron microscopy 35 and laser scanning microscopy. 36,37Previous studies have revealed a less ordered SC structure regarding lipid organisation in psoriatic lesions, 10,36,38 which is thought to be partly due to differences in the composition of ceramides. 38,39The FFA concentration of lesional psoriatic skin (LPS) has been found to be lower than of non-lesional psoriatic skin (nLPS). 29Furthermore, scanning electron microscopy has shown an abnormal lamellar ICL structure and appearance of lipid droplets in the psoriatic SC. 35 The NMF concentration in LPS is known to be lower than in nLPS, with the latter being reported similar to healthy skin (HS). 29Further differences are suggested for the secondary keratin structure, with LPS having a higher ratio of β-sheet/α-helix compared to nLPS or HS. 29 Additionally, LPS shows a higher TEWL in comparison to nLPS, 29,40 with the extent possibly correlating with the disease severity. 41Finally, SC hydration, which also correlates with SBF, [42][43][44] is reported to be lower in LPS than nLPS. 29,45In contrast to LPS, it is still unclear whether nLPS can be thought of as HS regarding its structure and SBF.Some studies indicate so, regarding lipid composition or TEWL and SC hydration, 29,46,47 but others [48][49][50] suggest an intermediate state of nLPS between LPS and HS.
Confocal Raman micro-spectroscopy (CRM) has become an increasingly valuable method for the non-invasive examination of the composition and SBF-related parameters of the SC. 19,23,51,52garding psoriasis, different versions of CRM and near-infrared spectroscopy have been applied in nLPS, 53 LPS areas of deceased psoriasis patients, 54 psoriatic scales in vitro, 55 an in vitro psoriatic skin model, 56 as well as in a pilot in vivo study 57 focusing on specific SC components.Raman spectroscopy has also been used in order to monitor the therapy progress on psoriatic skin. 58 far, to the best of the authors' knowledge, no in vivo studies have examined SBF-related structural parameters of the psoriatic SC depth-dependently on molecular level.Therefore, we have applied CRM and analytical algorithms established before 17,22,26 to carotenoids, keratin, lipid organisation, psoriasis, water binding compare the SC of LPS and nLPS in an explorative analysis with regard to lipid concentration and organisation, carotenoids, NMF, keratin secondary and tertiary structure, water concentration and mobility states as depth-dependent molecular parameters.
Examining the fine SC structure in contrast to integrated parameters over the entire SC allows a better understanding of the SBF and its alteration in LPS.

| Patients
Nineteen patients (8 female and 11 male) between 19-59 years old (mean age: 37 ± 12 years) of Fitzpatrick skin type I-III, 59 diagnosed with psoriasis vulgaris, were recruited from the department of Dermatology, Venereology and Allergology of the Charité-Universitätsmedizin Berlin.All patients were required to have at least one psoriatic lesion, chosen with a preferably low desquamation level (Table 1 and Table S1) on the elbows and/ or forearms, and instructed not to apply any topical skincare or pharmaceutical products on the examined body parts for at least 24 h prior to the CRM examination.All patients gave their written informed consent for their participation in the study.The study was conducted after approval by the ethics committee of the Charité-Universitätsmedizin Berlin (EA1/228/19) according to the Declaration of Helsinki.

| Data collection
The measurements were performed under standardised laboratory conditions (room temperature 20-21°C).After an acclimatisation time of at least 15 min, Raman depth profiles were recorded in vivo and non-invasively from the epidermis of the patients in the LPS and nLPS areas of the elbow and/or forearm, which were photographically documented and evaluated by a local PASI (Psoriasis Area and Severity Index) score, 60 regarding erythema, induration and desquamation of the lesions (Table 1 and Table S1).The lesions were not manipulated in any way before the measurements.It should further be noted that no profile was recorded from the actual tip of the elbow as this would not have been possible due to an inadequate contact between skin and glass surface of the microscope and those profiles recorded from the forearm, were recorded from the dorsal forearm in proximity to the elbow.
In order to record the Raman spectra, the skin composition analyser designed for in vivo/ex vivo measurements (RiverD International B.V., Model 3510, Rotterdam, The Netherlands) was utilised with a 785 nm laser (20 mW on skin surface, exposure time 5 s) for the fingerprint region: 400-2000 cm −1 ; and a 671 nm laser (17 mW on skin surface, exposure time 1 s) for the high wavenumber region (HWN): 2000-4000 cm −1 .The Raman profiles were recorded at 2 μm increments up to a total depth of 40 μm with an axial depth resolution of ≤5 μm and a spectral resolution of approximately 2 cm −1 .The used CRM device has been described in detail before. 61,62Exemplary Raman Spectra of LPS and nLPS from the FP and HWN range are shown in Figure S1.
Different profiles recorded from the same lesion were never in the same position.For all fingerprint spectra the corresponding HWN spectra were recorded in the identical position.In total, CRM profiles from 24 lesional and 16 non-lesional sites (Tables S2 and   S3) were chosen for statistical evaluation after excluding low quality spectra, that is, in cases of documented movement during measurement or failure to comply with the instructions.

| Data analysis
The Raman spectra were pre-processed according to the established methods. 22,63The skin surface (0% of SC depth) was defined as the point of half maximal intensity of the keratin-related band at 1655 cm −1 seen from outside the skin. 64The transition from the SC to the stratum granulosum (100% of SC depth) was determined as the point where the first derivative of the water concentration profile curve reaches 0.5, according to Crowther et al. 65 and as confirmed by the findings of Ri et al. 66 comparing different analytical procedures.
For the correct comparison of the SBF-related parameters, the SC thickness was normalised to 100% and all parameters were interpolated from 0% to 100% at 10% increments.The exact determination of each parameter regarding lipid concentration and structure, water mobility states and concentration, as well as keratin secondary and tertiary structure can be found in the corresponding results section.
The NMF profiles were calculated as in the study of Choe et al., 27 in form of a mathematical adaptation of the non-restricted multiple least squares fitting method proposed by Caspers et al. 67 The data analysis was performed using Skin Tools 2.0 (RiverD International B.V.), Matlab R2019b (The MathWorks Inc.), Origin 2020b (OriginLab Corporation) and Excel 2016 (Microsoft Corporation).For the examination of statistical significance between the mean values of LPS versus nLPS parameters at specific SC depths, t-tests were used assuming the normal distribution for the parameters on the basis of the large sample sizes and in accordance with the literature on CRM. 22,27,63The significance level was predefined as p < 0.05 and denoted with the symbol * in the graphs and a highly significant result was predefined with p < 0.01 and symbolised by **.

TA B L E 1
The median, maximum and minimum values of the psoriasis area severity index (PASI) parameters of each psoriatic lesion used for the measurements and statistical analysis (compare also Tables S1 and S3).

| Stratum corneum thickness
The SC thickness could be determined with the method described above in 61 out of 89 Raman spectra profiles of nLPS (≈69%) and 23 out of 85 profiles of LPS (≈27%), as can also be seen in Table S3.In the remaining profiles, either the skin surface could not be correctly determined, or the water-related gradient of 0.5 was not observed.
The mean SC thickness of nLPS was 24 ± 6 μm, while LPS had a mean SC thickness of 26 ± 8 μm.The difference between the means was not statistically significant (p = 0.21).Raman profiles with an undetermined SC thickness were excluded from further analysis.

| Intercellular lipids
The ICL concentration was calculated as the sum of the decomposed lipid-related Raman band intensities at 2880 and 2850 cm −1 , normalised by the keratin-related band intensity at 2930 cm −1 . 63,68e resulting comparison is summarised in Figure 1A, which clearly demonstrates that LPS has a lower mean ICL concentration than nLPS with high significance at 0%-80% SC depth (p < 0.01).
Next, the lamellar ICL organisation was examined by the ratio I 1080 /(I 1130 + I 1060 ), exploiting the fact that the bands at 1130 and 1060 cm −1 represent the all-trans conformation and the band at 1080 cm −1 the gauche-conformation of the ICL. 17As shown in Figure 1B, the ratio is significantly higher at 10%-40% SC depth for LPS (p < 0.01), indicating that ICL of LPS in this SC depth have a higher prevalence of gauche-conformers than nLPS, and thus suggesting a less ordered lamellar ICL organisation.
The lateral organisation of the ICL was determined using the I 2880 /I 2850 ratio, modified to exclude keratin interference to the Raman bands, as described in. 17Higher values are associated with a prevalence of the denser orthorhombic lateral organisation and lower values with the less ordered hexagonal lateral organisation of ICL. 17,68The results are presented in Figure 1C and show that LPS has a lower ratio over most of the SC thickness (highly significant from 0%-70% and significant at 80% SC depth) than nLPS, demonstrating a less dense lateral packing of ICL.
β-carotene and lycopene are highly concentrated carotenoids in the human SC with potent antioxidant and photo-protective properties, 69 supporting the lateral ICL organisation. 70For this reason, the carotenoid concentration was analysed, using the Raman band intensity at 1524 cm −1 . 71LPS and nLPS show no significant differences (Figure 1D) in carotenoid concentration over the entire SC.

| Natural moisturising factor (NMF)
For the calculation of the NMF concentration, 56 profiles of nLPS and 20 profiles of LPS were used in total, after combining the profiles for which a SC thickness could be determined with the ones for which the NMF concentration was able to be calculated with the Skin Tools 2.0 software.As demonstrated in Figure 2, LPS shows a much lower NMF concentration than nLPS over the entire SC thickness with a highly significant difference at 0%-60% and a significant difference at 70%-90% SC depth.

| Keratin structure
The secondary structure of keratin is largely determined by the number of α-helices, forming the stable coiled-coil structure, and β-sheets, allowing more exposed side chains. 72The β-sheet/α-helix ratio is described by the I 960 /I 938 ratio 22 and shown in Figure 3A.It indicates that LPS has a significantly higher ratio than nLPS in the superficial SC depth (p < 0.05 for 0% and p < 0.01 for 10%-30% SC depth), thus revealing a less stable secondary structure of keratin with higher number of β-sheets.
The tertiary keratin structure depends on the conformation and interaction of its side-chains. 21Figure 3B demonstrates the maximum position of the 2930 cm −1 band as obtained after decomposition of the broad band in the HWN region, confirming the non-homogeneous folding state of keratin in the SC. 22,63In LPS, the 2930 cm −1 band is shifted towards lower wavenumbers than in nLPS (highly significant difference at 0%-80% and significant at 100% SC depth), revealing a higher degree of folded keratin with reduced free side-chains in LPS.
Figure 3C demonstrates the amount of cysteine in keratin filaments which forms disulphide bonds, given by the I 690-712 /I 474-578 ratio, as the band at 474-578 cm −1 corresponds to the S-S bonds only, and the band at 690-712 cm −1 to the C-S groups of the total cysteine. 22The ratio is higher for LPS, indicating a lower amount of cysteine forming disulphide bonds than for nLPS, and signifying a less folded keratin state with respect to cysteine forming disulphide bonds.This is statistically highly significant at 10%-50% and 70% and significant at 60% and 80% SC depth.
Next, the stability of disulphide bonds in keratin was investigated using the ratio of the gauche-gauche-gauche to the total conformation of disulphide bonds (gauche-gauche-gauche + gauche-gauchetrans + trans-gauche-trans), calculated by I 474-508 /I 474-578. 22As can be seen in Figure 3D, the ratio is lower for LPS than for nLPS over the superficial and intermediate SC depth (with p < 0.01 at 30%-60% and p < 0.05 at 10%-20% SC depth), concluding that LPS has a lower percentage of the energetically most stable disulphide bonds than nLPS.
Finally, the ratio of buried to exposed aromatic rings of tyrosine in keratin side-chains was determined by the I 830 /I 850 ratio. 22,63is exploits the finding that buried tyrosine is associated with the 830 cm −1 , while exposed tyrosine with the 850 cm −1 band. 73,74gure 3E establishes that LPS has a higher amount of buried tyrosine in comparison to nLPS over the superficial and intermediate SC depth (p < 0.01 for 20%-70% and p < 0.05 for 10% SC depth), which indicates an increased folding of keratin in LPS.

| Water mobility states
Water molecules can be characterised by the strength of hydrogen bonds they form with their surrounding molecules, determining the different water mobility states as tightly, strongly, weakly, very weakly bound and free water. 25,26To explore the water mobility states within the SC, the HWN Raman spectra were decomposed with 10 Gaussian functions after subtracting the fluorescence background. 26To obtain the SC depth profiles, the intensities of the water-related sub-bands were calculated after decomposition and normalised by the total water concentration.The depth profile of the tightly bound water (forming single-donor-doubleacceptor-bonds: DAA) in the SC is calculated at ≈3015 cm −1 (centre position) 26,75 and shown in Figure 4A.LPS has a higher amount of tightly bound water than nLPS over the entire SC (p < 0.05 at 10%-20%, p < 0.01 at 30%-100% SC depth).Figure 4B exhibits the depth profile of the strongly bound water (DDAA) in the SC calculated at ≈3225 cm −1 (centre position). 26,75Evidently, LPS has a lower amount of strongly bound water molecules than nLPS over the entire SC (p < 0.05 at 0%-40% and 100%, p < 0.01 at 50%-80% SC depth).The depth profile of weakly bound water (DA), calculated at ≈3451 cm −1 (centre position), 26,75 is displayed in Figure 4C and does not reveal any significant differences between LPS and nLPS apart from the two exemplary at 60% and 80% SC depth.The superposition of the very weakly bound (DDA) and free water, also characterised as unbound water, is calculated at ≈3633 cm −1 (centre position) 26,75 and illustrated in Figure 4D. Figure 4D shows similar profiles of unbound water with a significantly higher concentration in nLPS only at 50%-60% SC depth, compared to LPS.
The ratio of weakly to strongly bound water is a useful descriptor of the hydrogen-bonding state of water, calculated by the ratio of the corresponding decomposed band intensities as I 3451 /I 3225. 26,27LPS appears to have a higher ratio than nLPS (highly significant at 60%, 80% and significant at 70% SC depth), as demonstrated in Figure 4E.It is further evident that LPS shows a lower hydrogen bonding state of water with the surrounding molecules than nLPS in the entire SC as a visible trend apart from the significant differences at 60%-80% SC depth.
Ultimately, the relative water concentration (water/protein ratio), calculated as I 3350-3550 /I 2910-2965 67 and depicted in Figure 4F, is manifestly lower in LPS than nLPS over the entire SC depth, with significance at 10% and 100% and high significance at 20%-90% SC depth.

| DISCUSS ION
In this non-interventional in vivo study of the psoriatic SC, we demonstrated using CRM that LPS has an altered molecular structure suggesting a reduced SBF compared to nLPS.
The SC thickness could not be calculated for every recorded profile.The determination was successful for nearly 2.5-times more nLPS than LPS sites, reflecting in our opinion previously published results, 54,55 which describe a perturbed and poorly defined SC in psoriasis.We failed to obtain a significant difference between the SC thickness of nLPS and LPS in contrast to Egawa et al., 57 who report a significantly thicker SC in the lesional areas.It should be noted that most of the measured lesions were only slightly indurated (Table 1, Table S1), which might partly explain this lack of significant difference.Both nLPS and LPS were found to have higher mean SC thickness (also with much higher standard deviation) than values obtained with the same CRM method 63,75 from healthy subjects.
ICL concentration in LPS was found to be much lower than in nLPS (Figure 1A), in agreement with the results of other groups, who have shown a lower concentration of FFA. 29,38Further known is an altered ceramide composition of the psoriatic skin, 39,57 with the extent of ceramide-decrease correlating with the PASI score 76 and clinical severity. 77Prosaposin, which plays a role in the generation of ceramides, has been reported to be lower concentrated in nLPS and even lower in LPS than HS. 78The different ceramide composition of the psoriatic SC might be related to the pathological keratinocyte differentiation 40 and/or increased presence of interferonγ, which has been suspected to reduce the number of long-chain fatty acids in ceramides. 79Indeed, the decrease of long-chain ceramides in psoriasis has also been found by other groups, 54,80 and is crucially associated with a shift of ICL organisation towards hexagonal lateral packing of lipids. 81Takahashi et al. 29 did not find any significant differences between FFA of nLPS and HS, which would be in accordance with our results showing a similar ICL concentration in nLPS to that in HS. 17,75 Apart from the ICL concentration, even more relevant for the SBF is their lateral and lamellar organisation in the SC.The curves describing the lamellar ICL organisation (Figure 1B) reach their minimum values at 10%-40% SC depth, as also known for HS, 17 corresponding to the highest order of lamellar ICL packing (more trans-conformers) in this SC depth.This is also the region, where LPS differs significantly from nLPS, thus indicating the less ordered lamellar organisation (more gauche-conformers) of LPS, which is in total agreement with results of previous studies. 54,55Regarding the nLPS, it is unclear whether its lipid organisation differs from HS.Our results for nLPS compared to HS 75 seem to suggest no difference as also reported by Leroy et al., 54 in contrast to other studies. 53,55e lateral ICL organisation was observed to be much denser in nLPS than in LPS (Figure 1C), which demonstrates further clear evidence of SBF impairment in LPS.It has a shifted ICL lateral organisation towards decreased orthorhombic packing, consequently showing a higher SC permeability. 82This loss of crystalline organisation in favour of the more fluid-like in LPS is in agreement with the results of Osada et al., 56 as well as the findings regarding reduced long-chain ceramides, 81,83 and also reflects the well-established higher TEWL values for psoriatic skin. 29,84,85Similar to the lamellar ICL organisation, our results for the lateral organisation of nLPS do not seem to differ from HS. 75 It should be noted that some factors might potentially disturb the correct interpretation of lipid organisation in psoriatic skin, such as the abnormal patterns of intercellular bilayers and lamellae, 35,86 or the presence of lipid droplets, which have been observed in psoriatic SC. 54 Carotenoids without hydroxyl groups (carotenes, lycopene) have been shown to facilitate the formation of the orthorhombic organisation of ICL, 70 which maintains the intact SBF.In this study, we observed an analogous distribution of carotenoids in nLPS and LPS (Figure 1D).Compellingly, the means for both LPS and nLPS seem to be lower (especially at the intermediate SC depth) than for HS (unpublished data from the subjects in). 75Previous studies 87,88 have also reported lower carotenoid concentrations in psoriatic skin compared to HS, with Lopes et al. 89 even proposing potential topical therapies for psoriatic skin with cyanobacterial carotenoids.
Our results suggest a very high difference in NMF concentration: LPS has a much lower mean NMF concentration in the SC compared to nLPS (Figure 2) and HS. 27This result is in agreement with other studies, 29,57 where the lower NMF concentration is explained as a consequence of reduced profilaggrin-filled keratohyaline granules in psoriatic skin, 90 resulting in less filaggrin expression. 91The reduced filaggrin levels are related to increased tumour-necrosis-factorα in LPS and nLPS 92 and a lower expression of filaggrin-2 in LPS has recently been linked to specific inflammatory cytokine molecules. 93Especially at the superficial SC depth, where NMF concentration is physiologically the highest and NMF the primary water-binding element, 22 the reduced NMF in LPS indicates a serious alteration of the water-binding properties of the SC, and as a result reduced SC hydration (Figure 4F) and, non-directly, reduced SBF.The NMF concentration in nLPS seems to be comparable to that of HS, 27,75 also in agreement with Takahashi et al. 29 The secondary structure of keratin in LPS is found to be less dense than in nLPS, as it is shifted towards more β-sheets than α-helices (Figure 3A).This outcome is in agreement with Takahashi et al., 29 who name as possible explanations of this effect the stretching of psoriatic keratinocytes and/or higher expression of proliferationrelated keratin types K6/K16/K17, as has been described for psoriatic skin. 9,94It is not clear to which extent these alterations apply to the nLPS; the β-sheet/α-helix ratio for nLPS appears in our study to be as in HS (unpublished data from the subjects in). 75e tertiary folding status of keratin directly influences the water binding properties of the SC. 22In LPS we found tertiary keratin structure parameters indicating more folded conformations than in nLPS, such as the higher ratio of buried tyrosine rings (Figure 3C) and the lower-shifted band position at 2930 cm −1 (Figure 3B) in disagreement with previous results. 56In contrast, there seems to be a lower amount of cysteine-forming disulphide bonds in keratin of LPS (Figure 3D) and the disulphide bonds themselves seem to be less stable (Figure 3E), that is, shifted away from the energetically stable gauche-gauche-gauche conformation in accordance with Osada et al. 56 -both rather indicating a more unfolded keratin tertiary structure, as also described for the in vitro psoriatic skin model. 55wever, as the band position at 2930 cm −1 describes the folding properties of the entire tertiary keratin structure, it can be argued that keratin is probably in a more folded state in LPS than nLPS and consequently has a lower ability to bind water.It has been known for decades that the protein maturation in psoriasis is deranged, 95 rendering the direct comparison of LPS with HS difficult.Regarding nLPS, neither our results for nLPS seem to differ from HS (unpublished data from the subjects in 75 ) nor those of others. 53Further research is needed to clarify the exact alterations of tertiary keratin structure in psoriatic skin.
The SC hydration was found to be lower in LPS than nLPS (Figure 4F), which is in agreement with previous results 29,57,85 and indirectly confirmed by the lower NMF concentration (Figure 2) and possibly the increased folding of tertiary keratin structure in LPS (Figure 3B).Our results show that the difference increases in the intermediate and bottom SC depth, which is responsible for binding water by keratin. 22In our study, SC hydration for nLPS was found apparently analogous to HS (from unpublished data).
So far, to the best of the authors' knowledge, no studies exist comparing the different water mobility states in the SC of psoriatic skin.Our study revealed a lower amount of strongly bound water in LPS than nLPS (Figure 4B), which amounts to the greatest difference observed among the different water mobility states.It might explain the lower hydration level of LPS, as at normal physiological conditions strongly bound water exceeds 45% of the total water in the SC. 26 The lowest concentration of strongly bound water in LPS is observed near the SC surface, which correlates well with the very low concentration of NMF at this SC depth (Figure 2).In contrast, tightly bound water is observed at a slightly higher amount in LPS (Figure 4A), but it represents physiologically less than 5% of the total water in the SC.This is also the case for unbound water, 26 which is detected marginally lower in LPS than in nLPS (Figure 4D).Finally, weakly bound water, physiologically amounting to the most common water state in the SC, 26 also demonstrates a weak divergence between LPS and nLPS with a tendency for more weakly bound water in LPS.The results for nLPS appear similar to results for HS 75 regarding the different water mobility states.
The exact interpretation of the water profiles for LPS and nLPS is not clear; while secondary and tertiary keratin structure-provided hydration do not correlate, the tertiary structure seems to have a higher influence on binding water, as the lower SC hydration of LPS agrees with the lower-shifted band position at 2930 cm −1 (Figure 3B), which shows an increase of folded keratin and thus lower ability to bind water molecules.The observed transformation of water from strongly to weakly bound states (Figure 4E) can also be interpreted as an additional hallmark of swelling of keratinocytes 96 in LPS, with others being the increase of SC thickness and the higher prevalence of keratin βsheets (Figure 3A).A comparison with HS 75 suggests that LPS has a higher ratio of weakly to strongly bound water than HS in the bottom half of the SC, while nLPS does not seem to differ from HS.
The performed study has some limitations.All skin sites of different patients corresponding to nLPS and LPS were considered similar and pooled together, without a mixed model analysis for dependent data.Furthermore, numerous t-tests were applied without correction for multiple testing.The performed analysis was motivated by the intention to facilitate comparability with previous studies using the same analysis, and the lack of the above mentioned corrections was compensated by a cautious interpretation of the results, for example regarding only clustered significant differences as relevant.A further possible limitation is the application of methods established with healthy skin (e.g.determination of SC thickness) to psoriatic skin, for which different conditions might apply.The high number of excluded profiles due to unobtainable SC thickness limits the direct applicability of the results to mild lesions (corresponding to mildly perturbed SC structure) and reduces the effectively analysed sample size with a resulting uneven distribution between male and female patients.Arguably, the study design also poses some limits to the study's findings, as only Caucasians with skin type I-III and between ages 19-59 were examined, reducing the impact of the results to this population.
While the primary focus of the study was on the comparison between nLPS and LPS, we also used results from previous studies with the same CRM and analytical methods to compare with HS.
Unfortunately, it was not possible to include a direct comparison, as the HS data are published in different studies.We have chosen reference 76 as the most frequent for adequate comparability, because the same CRM and analytical methods were used and the time of measurements was nearly overlapping with the psoriasis study.Lastly, even though short-term previous topical treatments were excluded for all patients, many of the patients had received topical and systemic therapies in the past, which could potentially influence the results and should be studied in detail in the future.
In conclusion, our study provides the first in vivo, non-invasive, depth-dependent analysis of the SC structure, using CRM to compare molecular parameters between LPS and nLPS.The depthdependent analysis yields insights into the exact structure of the SC and allows a better comparison of LPS with nLPS, which may be important in the non-invasive evaluation of disease severity and treatment control/optimisation.The impaired SBF of LPS for which a SC thickness can be obtained (i.e. is not greatly perturbed) is clearly related to the structural alterations evident from our results regarding ICL concentration and organisation, as well as NMF and water concentration.The secondary/tertiary keratin structure and their interactions with the different water mobility states also provide indications thereof, which should be further elucidated.A thorough understanding of the psoriatic SC and SBF offers useful information about the disease's pathophysiology and forms the basis for targeted topical therapies.The CRM method used in this study is well established for the determination of SC structure and SBF-related parameters in patients and may, together with the study results, provide a valuable tool for future interventional studies, comparing the psoriatic SC before and after specific treatments.

1
Depth profiles of the ICL-related parameters in the SC of LPS and nLPS: (A) The ICL concentration given by the ratio of the intensities at 2880 and 2850 cm −1 normalised to the keratin-related band intensity at 2930 cm −1 after spectra decomposition.(B) The lamellar ICL organisation calculated with the ratio of the band intensities at 1130 and 1060 to 1080 cm −1 .(C) The lateral ICL organisation analysed by the ratio of intensity at 2880 to 2850 cm −1 , modified to exclude keratin interference.(D) The carotenoid concentration represented by the band intensity at 1524 cm −1 .[*p < 0.05, **p < 0.01; ICL, intercellular lipids; SC, stratum corneum; (n)LPS, (non-) lesional psoriatic skin; the error bars show the standard error of the means].

F I G U R E 2 F I G U R E 4
Depth profiles of the NMF concentration in the SC of LPS and nLPS.[*p < 0.05, **p < 0.01; NMF, natural moisturising factor; SC, stratum corneum; (n)LPS, (non-) lesional psoriatic skin; the error bars show the standard error of the means].F I G U R E 3 Depth profiles of the secondary (A) and tertiary (B-E) structure of keratin in the SC of LPS and nLPS: (A) The ratio of β-sheet/ α-helix in secondary keratin structure plotted as the I 960 /I 938 ratio.(B) The 2930 cm −1 band position, which measures the folding of tertiary keratin structure.(C) The amount of cysteine in keratin filaments, which forms disulphide bonds, given by the I 690-712 /I 474-578 ratio.(D) The stability of disulphide bonds in keratin, analysed by the I 474-508 /I 474-578 ratio.(E) The ratio of buried/exposed aromatic rings of tyrosine in keratin side chains determined by the I 830 /I 850 ratio.[*p < 0.05, **p < 0.01; SC, stratum corneum; (n)LPS, (non-) lesional psoriatic skin; the error bars show the standard error of the means].Depth profiles of the water mobility states and concentration in the SC of LPS and nLPS calculated after decomposition of HWN Raman spectra with 10 sub-bands: (A) The relative amount of tightly bound (DAA) water in the SC (sub-band corresponding to centre position ≈3015 cm −1 ) normalised to the total water amount.(B) The relative amount of strongly bound (DDAA) water in the SC (sub-band corresponding to centre position ≈3225 cm −1 ) normalised to the total water amount.(C) The relative amount of weakly bound (DA) water in the SC (sub-band corresponding to centre position ≈3451 cm −1 ) normalised to the total water amount.(D) The relative amount of very weakly bound (DDA) and free water in the SC (sub-band corresponding to centre position ≈3633 cm −1 ) normalised to the total water amount.(E) The ratio of weakly to strongly bound water in the SC calculated as I 3451 /I 3225 .(F) The relative water concentration in the SC calculated as I 3350-3550 /I 2910-2965 .[*p < 0.05, **p < 0.01; SC, stratum corneum; (n)LPS, (non-) lesional psoriatic skin; HWN, high wavenumber; DA, single donor, single acceptor; DAA, single donor, double acceptor; DDA, double donor, single acceptor; DDAA, double donor, double acceptor; the error bars show the standard error of the means].