3D In vitro model of a functional epidermal permeability barrier from human embryonic stem cells and induced pluripotent stem cells

Abstract

Cornification and epidermal barrier defects are associated with a number of clinically diverse skin disorders. However, a suitable in vitro model for studying normal barrier function and barrier defects is still lacking. Here, we demonstrate the generation of human epidermal equivalents (HEEs) from human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs). HEEs are structurally similar to native epidermis, with a functional permeability barrier. We exposed a pure population of hESC/iPSC-derived keratinocytes, whose transcriptome corresponds to the gene signature of normal primary human keratinocytes (NHKs), to a sequential high-to-low humidity environment in an air/liquid interface culture. The resulting HEEs had all of the cellular strata of the human epidermis, with skin barrier properties similar to those of normal skin. Such HEEs generated from disease-specific iPSCs will be an invaluable tool not only for dissecting molecular mechanisms that lead to epidermal barrier defects but also for drug development and screening.

Figure 1

Epidermal Permeability Barrier within the SC of Normal Human Skin (A) Schematic drawing of normal human skin, depicting the epidermal permeability barrier within the SC. (B–F) TEM demonstrates the components of a competent permeability barrier in normal human skin. Lipids are packaged into lamellar bodies (LBs) in the SG cells. (B) LBs (arrows) are present in the uppermost SG cells. These LBs are not discrete droplets, but rather are part of a tubular network extending from the Golgi. (C) LBs, with all their contents, are secreted into extracellular space (asterisks) between the uppermost SG cells. De, desmosome. (D) Close-up of LB morphology. (E) Lipid from LBs is processed into lipid bilayers (asterisk) by SC enzymes that require acidity, thus forming lipid structures that are impermeable to passage of water and ions. (F) Ceramide-based lipids bind to the CE (asterisk), forming the cornified lipid envelope (CLE). Scale bars are noted in each panel.

Figure 2

Generation and Characterization of Normal Human iPSCs (A) Trypan blue staining of colonies in one well of a six-well dish. The average reprogramming efficiency from two independent experiments was ∼0.76%. (B–D) DNA methylation analyses. Hierarchical clustering (B), methylation portrait (C), and principal component analysis (D) suggest that both iKCL004 and iKCL011 have the methylation signature of PSCs. BJ, parental HFFs. n = 3 biological replicates (rounds of differentiation) for each cell line. (D) Undifferentiated iKCL004 and iKCL011 cultured on HFF feeders and under feeder-free conditions display the typical morphology of PSCs. (E) Expression of pluripotency markers is confirmed by alkaline phosphatase (AP) activity assay and immunostaining for NANOG, OCT4, TRA-160, and TRA-1-81 in iKCL004 and iKCL011 lines. (F) Differentiation of iKCL004 and iKCL011 into all three germ layers in vivo is confirmed by detection of specific markers. Teratomas were encapsulated and did not invade surrounding tissues. All sections were stained with hematoxylin and eosin (H&E). Specific stains are brown (immunohistochemistry) or light blue (Alcian). Positive staining for mitochondrial complex IV type II confirms the human origin of the teratoma tissue. Germ layer markers: Alcian blue/periodic acid Schiff (PAS)-stained cartilage and desmin for mesoderm, βIII tubulin and glial fibrillary acidic protein (GFAP) for ectoderm, and GATA4 and α-fetoprotein for endoderm. See also Figures S1–S3.

Figure 3

Efficient and Reproducible Differentiation of hESCs and iPSCs into Keratinocytes (A) Schematic of the differentiation protocol. T0 (day 0), T1 (day 7), T2 (day 14), and T3 (day 28) represent time points at which the cells were switched to a new culture condition and/or collected for analyses. (B) qPCR analysis measuring the expression of differentiation markers (epidermal: ITGB4 and KRT14; neural: OLIG2) suggests a difference in commitment toward epidermal lineage at time point T1. n = 9 for each cell line; each of three rounds of differentiations (biological replicates) had three technical replicates. (C) Diversity is lost at later stages, and cell colonies at the enrichment step are uniformly positive for keratinocyte markers K14 (red) and p63 (green). (D) qPCR analyses measuring expression of differentiation markers at T0, T1, T2, and T3 of the differentiation protocol. Expression of pluripotency markers OCT4 and NANOG is diminished by the end of differentiation, whereas KRT18 and p63 mRNA levels reflect stages during epidermogenesis. n = 9 for each cell line; each of three rounds of differentiations (biological replicates) had three technical replicates. (E) Levels of KRT14 mRNA expressed during three rounds of differentiation at four different time points for each of lines demonstrate the reproducibility of the protocol. Absolute quantification was performed using the ddCt method. GOI expression was normalized to the geometrical mean of two HK genes (GAPDH and ElF4A2). The SEM was calculated for mean expression in KCL034, iKCL004, and iKCL011. n = 9 for each cell line; each of three biological replicates/rounds of differentiations had three technical replicates. (F) The proliferation rate of hESC/iPSC-derived keratinocytes shows no marked difference compared with freshly isolated primary NHKs over a period of 5 weeks. Statistical significance was calculated for each passage using two-way ANOVA with Sidak’s multiple comparison test. Data are presented as a blank adjusted mean with SD (n = 3 biological replicates for each cell line). See also Figure S4.

Figure 4

Profiling of hESC/iPSC-Derived Keratinocytes (A) qPCR analyses measuring expression of KRT14 and p63 in hESC/iPSC-derived keratinocytes at T3 (n = 9; three rounds of differentiation for each of three lines) and NHKs (n = 2) found no significant difference between the two groups. Multiple t tests. Statistical significance was determined using the Holm-Sidak method. For KRT14, p = 0.120881; for p63, p = 0.155472. (B) Populations of hESC/iPSC-derived keratinocytes at T3 have a similar percentage of K14⁺/integrin β4⁺ cells as NHKs. The percentage is an average from three rounds of differentiation for each of the three lines. At T0, no K14⁺/integrin β4⁺ cells were detected. Fluorescence-activated cell sorting (FACS) images represent one of the three rounds. (C and D) Scatterplots showing the relative expression of all genes on the chip. (C) Global gene-expression profiles of NHKs from two donors (NHK1 and NHK2) shown as a scatterplot against each other. (D) Global gene-expression profiles of hESC/iPSC-derived keratinocytes at T3 (average of n = 3 biological replicates/rounds of differentiation for each line) shown in scatterplots against NHKs (average of n = 2). Normalized linear expression levels for each probe are plotted as XY coordinates for each sample. Yellow lines indicate boundaries of a 2-fold relative expression difference. (E) Venn diagram of the transcripts with a 2-fold relative expression difference from NHKs in KCL034-, iKCL004-, and iKCL011-derived keratinocytes at T3 shows that 122 out of >47,000 were common to all three lines. (F) Heatmaps showing relative expression patterns across multiple samples for specific gene lists. HLA genes were selected by text searching for “HLA” within the “Gene Symbol” annotation, and p63-related genes were identified from the p63 transcriptional network in the NCI Pathway Interaction database. The colors of the heatmaps indicate high (yellow) or low (blue) expression relative to the average for that gene across the samples. The values are the average of three independent runs for KCL034, iKCL004, and iKCL011 at T0 and T3. Transcripts for NHKs from two different donors are shown independently on the HLA heatmap, and as an average on the p63 transcriptional network heatmap. See also Figures S5 and S6, and Tables S1–S5, S6, and S7.

Figure 5

Pluristratified HEEs Generated In Vitro from hESCs and iPSCs (A) The TEER during HEE formation over a period of 14 days reflects permeability barrier formation (n = 4 for NHK, n = 12 for each of lines; each point represents an average of measurements from three different spots). Error bars represent SD. Only 14-day-old HEE cultures that displayed a TEER of >1,200 ohm cm² were used for subsequent analyses unless indicated otherwise. (B) H&E staining demonstrates normal morphology with the presence of all epidermal layers regardless of the source of keratinocyte population. SB, stratum basale; SS, stratum spinosum.

Figure 6

The Distribution of Epidermal Markers in HEEs Mirrors that in Human Skin (A) Immunostaining for three markers of keratinocyte terminal differentiation: filaggrin, loricrin, and involucrin. Each of these markers is expressed at the appropriate site, denoting normal epidermal differentiation. (B–D) Immunostaining for K14 (basal layer) and K10 (suprabasal layers) (B), desmocollin 1 (suprabasal layers) (C), and p63 (basal layer in general) (D) demonstrates normal epithelial stratification in all HEE cultures. (E and F) Basal layer keratinocytes express laminin (E) and form BMZ-like structure (F). BM, basement membrane; Cy, cytoplasm; SB, stratum basale; TM, transwell membrane. Yellow arrowheads point to laminin-positive staining of BM.

Figure 7

HEE with a Functional Permeability Barrier (A) Lipid bilayer formation assessed with TEM. Arrows, lamellar bodies; asterisks, lipid bilayers. Upper row: normal lipid secretion between the SG and SC is seen in all cultures. Middle row: LBs are seen in the SG of all cultures (arrows). LB morphology was normal in all cultures, although it appeared slightly smudged in KCL034. Bottom row: lipid was successfully processed into lipid bilayers (asterisks) in all cultures. Like LB morphology, lipid bilayer morphology was normal in all cultures, although it seems slightly disrupted in KCL034. (B) ER Ca²⁺ sequestration at day 7 of HEE culture in cells transfected with the ER-targeted Ca²⁺ sensor D1ER. Data are presented as the mean of the intensity ratio (I) between the yellow-channel (higher Ca²⁺) and blue-channel (lower Ca²⁺) images (I(yfp)/I(cfp)) ± SEM. Higher ratios denote higher Ca²⁺ stores in the ER. n = 10–14 cells from two biological replicates in each group. Significance was calculated using a one-way ANOVA. Distributions with p < 0.05 were assumed to be statistically different based on unpaired t tests between the populations. (C) The epidermal Ca²⁺ gradient captured on TEM as electron-dense Ca²⁺ deposits at day 14. Arrows indicate the direction from the basal layer toward the SC. Ca²⁺ precipitates (black deposits) denote the presence of Ca²⁺ in the tissue. Precipitates are seen in the viable SG but are absent from the SC, denoting a functional barrier to the passage of water and ions. (D) Permeability barrier integrity assessed by lanthanum perfusion. Lanthanum is visualized as electron-dense deposits in the extracellular spaces of the viable SG (arrowheads), demonstrating that lanthanum and, by extension, water and other small ions can pass between keratinocytes in this stratum. In contrast, lanthanum spreads along the base of the SC, but cannot penetrate further into the SC because a functioning lipid barrier is blocking its movement upward. All cultures demonstrated a functional permeability barrier.