Label-free stimulated Raman scattering microscopy visualizes changes in intracellular morphology during human epidermal keratinocyte differentiation

Epidermal keratinocyte (KC) differentiation, which involves the process from proliferation to cell death for shedding the outermost layer of skin, is crucial for the barrier function of skin. Therefore, in dermatology, it is important to elucidate the epidermal KC differentiation process to evaluate the symptom level of diseases and skin conditions. Previous dermatological studies used staining or labelling techniques for this purpose, but they have technological limitations for revealing the entire process of epidermal KC differentiation, especially when applied to humans. Here, we demonstrate label-free visualization of three-dimensional (3D) intracellular morphological changes of ex vivo human epidermis during epidermal KC differentiation using stimulated Raman scattering (SRS) microscopy. Specifically, we observed changes in nuclei during the initial enucleation process in which the nucleus is digested prior to flattening. Furthermore, we found holes left behind by improperly digested nuclei in the stratum corneum, suggesting abnormal differentiation. Our findings indicate the great potential of SRS microscopy for discrimination of the degree of epidermal KC differentiation.


Ex vivo visualization of intracellular morphologies of the epidermis. Vertical cross-section images
of abdominal skin of a 64-year-old subject observed by haematoxylin-eosin (HE) staining, transmission electron microscopy (TEM), SRS imaging at 2930 cm −1 , which mainly reflects proteins, and 2850 cm −1 images that mainly reflect lipids, are shown in Fig. 1. As shown in the illustration in Fig. 1, the same vertical section images for TEM and SRS. The optical vertical cross-section image observed by SRS was similar to the two conventional dermatological techniques with regard to the morphology of each epidermal layer. However, subtle differences such as thicknesses of the SC [ Fig. 1(a1, b1, c1)], size of cell [ Fig. 1(a5, a7, b5, b7, c5, c7)], and size of nuclei [ Fig. 1(a4, a6, b4, b6, c4, c6)] were observed. Specifically, the thickness of the SC and full epidermis by HE staining, TEM, and SRS were 9.9, 6.0, 9.3 and 50.1, 31.2, 48.1 µm, respectively. The average long diameters of cells and nuclei by HE staining, TEM, and SRS were 13.9, 10.9, 12.0 and 3.7, 3.0, 4.6 µm, respectively. The difference in values was suggested to be the effect of chemical pretreatment processes such as fixing/labelling/staining used in conventional dermatological techniques on the obtained images of tissue staining. Figure 2 shows typical molecular vibrational images of each epidermal layer in optical horizontal unsliced cross-sections of skin from a 70-year-old Caucasian female. Characteristic intracellular morphologies of each epidermal layer were observed without labelling, similar to the previously reported preliminary results of a 52-year-old Caucasian female at various depths 39 . In the SB and SS, nuclei were observed as a low intensity area both at 2850 cm −1 , reflecting mainly lipids (Fig. 2b), and at 2930 cm −1 , reflecting mainly proteins (Fig. 2a); whereas, nucleoli were observed as a high intensity area in 2930 cm −1 images (Fig. 2d). Lipid-rich granules in the SG analogized as lamellar granules in which lipids accumulated in the SG. A relatively high intensity at 2850 cm −1 was observed as a red colour in the SC in the merged image (Fig. 2c), thus reflecting the existence of intercellular lipids. In addition, visualization of the SC without distinct layers was observed in some skin (as for this subject), a feature not observed in the previously reported 52-year-old subject 39 .

3D intracellular morphologies of skin.
Supplementary videos S1 and S2 show typical horizontal cross-section images at 2930 cm −1 , reflecting mainly proteins, of abdominal skin of a 56-year-old subject and eyelid skin of a 59-year-old subject respectively. The 3D images consist of optical horizontal cross-sections at a 1-µm interval in depth. The differentiation process was clearly observed in 3D, as differences in intracellular morphology between the two tissues were clearly observed in each epidermal layer; for example, differences in the flattened features of the outlined KCs in the SB and SS, intranuclear features during the enucleation process in the upper part of SS and SG, and roughness in the outline of dead KCs in the SC.
Changes in the size of KCs during epidermal differentiation. Continuous cell size changes accompanying epidermal KC differentiation from SB to the SC were measured. Sizes of KCs in each layer were calculated from three values on one horizontal cross section image on the SB, SS, SG, and the lower/middle layers of the SS and two values on the upper layer of the SC where three could not be calculated, then the average of each layer was calculated. Changes in KC size during epidermal differentiation are shown in Fig. 3. The average cell size at the SB (14 abdominal skin of 31,31,33,39,43,49,49, 52, 56, 60, 62, 66, 70, and 74 years old subjects; and two eyelid skin of 59 and 60 years old subjects) was approximately 99.5 µm 2 , which increased up to 1120 µm 2 at the SG, and finally reached 1305 µm 2 at the SC. The size of cells increased more than 10 times at the SC compared with the SB during differentiation. intercellular morphological changes of nuclei in the SS. Figure 4 shows typical time-lapse optical cross-section images at 2850 cm −1 and 2930 cm −1 in the SS of skin of 39-and 49-year-old subjects from an abdominal site. Horizontal cross-section SRS images of skin mounted onto a cover slip with PBS were obtained as a measurement of 0 min, then in the absence of any chemical treatment the same area was measured using the same method after 75 or 120 min. The diagram in Fig. 4f shows suggested changes in the nuclei during the enucleation process and optical cross-section images at 2930 cm −1 . As already shown in Fig. 2a, nuclei in KCs in the middle to lower part of the SS were observed as a black low-intensity area, while nucleoli appeared as a white high-intensity area in living layers in the 2930 cm −1 image, reflecting mainly proteins. As shown in Fig. 4a-c, similar images were obtained in skin of a 39-year-old subject (corresponding to Areas 1 and 2) and in skin of a 49-year-old subject (corresponding to Area 3) at 0 min and after 75 or 120 min. Of note, filling with protein was observed in areas A and B of Fig. 4d,e in 2930 cm −1 images, but not in the 2850 cm −1 image in the skin of a 49-year-old subject (corresponding to Areas 1 and 2). existence of holes in the Sc. We observed holes in the SC in horizontal cross-section images, whereby the intensity at both 2850 cm −1 (reflecting mainly lipids) and 2930 cm −1 (reflecting mainly proteins) was quite low, almost black. We named these "SC holes".  Figure 6 shows the molecular vibrational characteristics of SC holes. Based on the observations of 2850 cm −1 (Fig. 6a) and 2930 cm −1 (Fig. 6b) images, the outline of SC holes consisted of lipid-rich components.

Discussion
In the current study, we were able to achieve the first 3D constructions visualizing changes in intracellular morphology during epidermal KC differentiation in humans without any chemical processing that might affect the skin. These results demonstrated the great potential of intracellular morphology imaging with molecular vibration information using SRS microscopy for quick diagnosis of the differentiation state of epidermal layers. Our approach was different from most other studies using spontaneous/coherent anti-Stokes Raman microscopy, which relies on the use of characteristic molecular vibrations of target components 30,31,41 . Our vibrational contrast approach takes advantage of the benefits of high-speed evaluation to yield medical imaging applications.   By analysing changes in the morphology of KCs during epidermal differentiation, we evaluated intact cell size from SB to SC for the first time. Evaluating the size of cells at the skin surface is an alternative to measuring epidermal turnover using a stripped SC 42 . Our method enabled us to evaluate actual turnover conditions by observing the size of cells in each epidermal cell layer. Thus, it will be possible to elucidate mechanisms underlying turnover in humans in vivo using non-labelling molecular imaging with back-scattering detection.
In the upper part of the SS to SG, we observed the shrinking of nuclei and subsequent filling with protein in the upper part of the SS during the initial stage of the enucleation process. Time-dependent changes in spectral images allowed visualization of the contraction process of cell nuclei. In these observations, the inside of nuclei were exhibited as a high-intensity area in the 2930 cm −1 image from the outline of nuclei to the inside, followed by subsequent protein filling of nuclei, which occurred at 75 min (Fig. 4d), resulting because of nuclear shrinkage at the initial stage of the enucleation process or changes associated with tissue deterioration in refrigeration storage. In either case, this is the first example of visualizing such intracellular morphological changes ex vivo. In previous studies conducted by a labelling method using a 3D skin model and mice, change in morphology of nuclei in the last moment of the enucleation process was captured on a time scale within 60 minutes 43,44 . The time scale of morphological changes in the nucleus we captured do not seem to contradict these previous studies. In addition, our observation of tight junction-like features in the SG of humans (Supplementary videos S1 and S2) was similar to those reported using the transgenic labelling of mice 9 . Our findings demonstrating the visualization of human tight junctions using vibrational contrast illustrate the potential of SRS for applications in dermatological basic research.
In the SC, we obtained lipid-rich images based on high levels of intercellular lipids. Samples taken from younger subjects showed higher lipid levels that made the cell boundaries clearly visible. However, lipid levels were lower in elderly skin, yielding an ambiguous cell outline, suggesting that advancing age causes a deterioration of the epidermal KC differentiation process. We also observed holes, circular regions with low protein densities, in the SCs of some subjects. Notably, parakeratosis, a condition in which nuclei remain in the SC, was not observed in any skin by HE-staining ( Supplementary Fig. S1). A part of this phenomenon, referred to as the "ghost nucleus", was captured by a staining method in a previous report using tape-stripped SC cells at the outermost thin layer of skin 41 . However, it was impossible to clarify their identity in detail because of the limitations of staining/chemical treatment methods especially for the SC. In the process of healthy SC formation, the nuclei of KCs disappear in the SG, the inside of the KCs-including the lost part of the nucleus-is filled with keratin fibres in the SG to lower part of the SC, and the cell shape becomes flattened and finally a sheet form stacked as ten to several tens of layers is formed 1,45 . This layered structure of the SC possesses a strong barrier function. Barrier function is decreased during parakeratosis, which is in the category of non-diseased skin, because the SC is not completely filled with keratin fibres because the nuclei remain. Our observations suggest that the nuclear membrane would remain intact, such that keratin fibres did not fill the inside remnant of the nuclear membrane, thus leaving a distinct hole after digestion of the nucleus. Incomplete SC with holes remaining would lead to a decrease in physical barrier function. Such holes appearing would be the result of abnormalities in the enucleation process, especially the process of nuclear membrane loss and keratin fibre filling after enucleation. Importantly, the barrier function of skin with SC holes would be lower than that of normal skin, resulting in the absorption and retention of water in the SC, which would be affected. These are the first findings implying the order of digestion in the enucleation process. Thus, morphological changes during epidermal KC terminal differentiation from the upper part of the SS to the SG are considered important factors for cell death in the SC. Abnormalities in the steady process  www.nature.com/scientificreports www.nature.com/scientificreports/ of cell death in the SC, as discussed above, might lead to both a decline in the barrier function of the epidermis and reduced moisture retention.
As for the obtainable information, SRS microscopy still has some limitations compared with existing technologies. First, image resolution for evaluation of the intracellular morphology of SRS microscopy is not high compared with TEM. Second, functional groups detectable by SRS microscopy within a narrow range of wavenumber measurements are limited compared with spontaneous Raman scattering microscopy and coherent anti-Stokes Raman scattering microscopy. However, the biggest benefit of SRS microscopy is fast-speed imaging. Compared with other high-speed morphology imaging methods available for the epidermis, such as confocal microscopy 12 and optical coherence tomography 14 , SRS microscopy has the advantage of visualizing the morphology of intracellular components using vibrational contrast of video-rate images. Future investigation with back-scattering detection without slicing 18,21,32 would greatly expand the potential applications of SRS microscopy for evaluating skin diseases in clinical dermatology, as well as skin conditions in cosmetic dermatology fields. Furthermore, discrimination of the differentiation degree by intracellular morphology with vibrational contrast could be applied beyond dermatology for the evaluation of stem cells and cell differentiation in a variety of fields.
conclusions This is the first report demonstrating changes in 3D intracellular morphology during epidermal KC differentiation ex vivo using subcellular spatial distribution without fixing/staining/labelling. We analysed actual turnover conditions by observing the size of KCs in each epidermal layer, which was not possible using previous dermatological/optical techniques. In addition, we observed changes in intracellular and nuclear morphology during the enucleation stage for the first time, which allowed us to determine that the digestion of nuclei after enucleation is an important step linked to terminal differentiation in the SC. We also revealed the existence of holes in the SC that we posit are related to stiffness and water absorption/release levels. Our results can be applied to clarify which differentiation processes are important in forming healthy skin with a strong barrier function. We also showed that high-speed SRS imaging has great potential even in a narrow wavenumber range to evaluate the degree of differentiation by visualizing intracellular morphology.

Methods
SRS microscopy. The details have been described previously 46 . Briefly, a Ti:sapphire (Ti:S) laser at 790 nm with 0.14-nm spectrum width and wavelength-tunable Yb fibre (YbF) laser at 1015-1045 nm were used as light sources. An apparatus overview is shown in Fig. 8a. The power of Ti:S and YbF lasers at the input of laser scanners was 110-120 mW each, and that in the sample plane was estimated to be −60 mW each. Horizontal cross-section SRS images (80 × 80 μm; 500 × 500 pixels; transmission mode; horizontal spatial resolution of −0.5 μm) were acquired in the wavenumber range of 2800 to 3100 cm −1 , corresponding to the C-H stretching region 20,21,39 . The frame rate of the microscope was 30 frames/s. Images were accumulated to improve the signal-to-noise ratio (continuous wavenumber scanning mode; 10 accumulations, discrete scanning mode; 50 or 100 accumulations). The wavelength dependence of the YbF laser intensity was measured on each measurement day and used for calibration.
Ex vivo molecular imaging of human skin. Human skin from the abdomen and eyelid (n = 19, 31-74 years old, Caucasian females) were purchased from Biopredic International (Rennes, France) via KAC Co., Ltd. (Kyoto, Japan). The tissues had been collected during plastic surgery after informed consent had been obtained according to French laws including ethics regulations. Our study was approved by the ethics committee of Shiseido, in accordance with the guidelines of the National Institute of Health. Skin were dermatomed to approximately 400-μm thickness on the production date in France and then transported to Japan. Tissues were refrigerated during transit and until measurement without any chemical treatment. Skin were immersed in PBS, mounted onto a cover slip, sandwiched with another coverslip, and sealed with enamel. The tissues were molecularly imaged in the horizontal optical cross-section at a depth interval of 1 µm using SRS microscopy without physical slicing. Typical SRS spectra of epidermal layers are shown in Fig. 8b. Spectral images at 2850 cm −1 (CH 2 symmetric stretching), which mainly reflect lipids, and 2930 cm −1 (CH 3 symmetric stretching), which mainly reflect proteins, were used to visualize the intracellular morphology of epidermal layers 25,29,34,35,47 .