Long-term expansion and differentiation of adult murine epidermal stem cells in 3D organoid cultures

Establishment of Murine Epidermal Organoid Cultures.

As originally established for mouse (25) and human intestinal stem cells (26), epithelial stem cells from multiple adult tissues can be cultured as 3D structures, called organoids, which mimic the cellular composition of their respective tissue-of-origin (27). Several culture systems for murine skin epidermis have been established. To circumvent some of their limitations, we designed an organoid culture system for adult murine epidermis. We systematically screened for media components that allow feeder- and serum-free expansion of epidermal keratinocytes. We isolated primary epidermal keratinocytes from mouse telogen back skin, as described by Watt and colleagues (28), and embedded unsorted single cells in basement membrane extract (BME) (Fig. 1A). We started from the basic medium utilized for the generation of mouse small intestinal organoids (25), consisting of advanced DMEM/F12 with antibiotics (penicillin/streptomycin), a buffering agent (Hepes), an l-glutamine source (GlutaMAX), B27 supplement (29), and a Rho kinase inhibitor (Y-27632) to inhibit anoikis (25). We enriched this basic medium with additional growth factors based on previously defined conditions for the expansion of epidermal stem cells. First, Wnt/β-catenin signaling is essential for epidermal stem cell self-renewal (30) and the Wnt agonist R-spondin 1 is already a component of the established intestinal organoid medium (25). Second, the extracellular signal-related kinase (ERK)/mitogen-activated protein kinase (MAPK) signaling cascade is activated by epidermal growth factor (EGF) or fibroblast growth factors (FGFs) and stimulates epidermal stem cell proliferation in vitro (31). Third, the signaling pathways of bone morphogenetic protein (BMP) and TGF-β stimulate HF stem cell quiescence (32–34) and are blocked by the respective inhibitors Noggin and A-83-01 (hereafter TGF-βi). Fourth, cyclic adenosine monophosphate (cAMP) increases epidermal keratinocyte expansion in vitro (35) and is activated by Forskolin (36).

To determine optimal culture conditions, we assessed different combinations of these growth factors. We identified that a defined epidermal expansion medium (EEM) containing Noggin (N), R-spondin 1 (R), Forskolin (F), and FGF1 provides the most optimal growth conditions with many large organoids being generated within 1 wk (Fig. 1 B and C and SI Appendix, Fig. S1A). No organoids formed in the absence of all of these factors (–N –R –F –FGF1). Withdrawal of either FGF1 or Forskolin significantly reduced organoid count and size (Fig. 1 B and C). Removal of Noggin or R-spondin 1 allowed formation of many small organoids, suggesting that both factors are required for organoid expansion (Fig. 1 B and C and SI Appendix, Fig. S1B). Culture medium containing either EGF or FGF1 generated proliferative organoids (Fig. 1 B and C and SI Appendix, Fig. S1 B, E, and F), suggesting that EGF and FGF1 do not synergize and can functionally replace each other in culture. FGF1 could be further replaced by FGF10, suggesting that both growth factors may act through FGF receptor 1b activation (37). Culture medium containing TGF-β inhibitor instead of FGF1 allowed the formation of organoids, but resulted in more differentiated organoids with less viability (SI Appendix, Fig. S1 B, E, and G). In the absence of FGF1, TGF-β inhibition did not improve organoid count and size, but addition of Forskolin partially restored organoid formation (Fig. 1 B and C and SI Appendix, Fig. S1B). Organoids could be maintained at comparable proliferation rates for several passages in culture conditions containing FGF, EGF, or TGF-βi (SI Appendix, Fig. S1 E and F). In conclusion, activation of ERK/MAPK through either FGFs or EGF in combination with BMP inhibition by Noggin, stabilization of Wnt/β-catenin signaling by R-spondin 1, and cAMP activation by Forskolin are essential for formation of murine epidermal organoids.

To investigate the optimal cell concentration for expansion of organoids, we tested different cell-seeding densities. Organoid formation was optimal at 2,500 cells per 10 µL BME drop in a well of a 48-well plate and deteriorated below 1,000 cells per 10 µL of BME (SI Appendix, Fig. S1 C and D). Single-cell suspensions generated organoids (Fig. 1D). Compact organoid structures with a keratinizing inner core formed within 1 wk, independent of the passage number, even though the amount of keratinization was higher in later passages (Fig. 1E). Organoid cultures could be passaged every 7 d in an approximate ratio of 1 to 4. Murine epidermal organoids were cultured for up to 6 mo while retaining this morphology (Fig. 1F).

Lgr5, Lgr6, and SCA-1 Identify Multiple Sources of Organoids.

We questioned whether different (integrin α6⁺) epidermal stem and progenitor cell compartments can initiate organoid formation. We and others previously described the existence of HF stem cells marked by leucine-rich repeat-containing G protein-coupled receptor 5 (Lgr5) and stem cells in several epidermal compartments marked by Lgr6 (8, 30, 38–41). While Lgr5 is highly expressed in the lower HF (Fig. 2A), Lgr6 expression is confined to the upper HF (SI Appendix, Fig. S2A). Cellular expression of stem cell antigen-1 (SCA-1, encoded by the Ly6a gene) marks the infundibulum and the IFE, including the respective stem cell populations (Fig. 2B). We used our Lgr5^{eGFPiresCreERT2} and Lgr6^{eGFPiresCreERT2} knockin mouse models (41, 42) as well as SCA-1 immunolabeling to purify Lgr5-eGFP⁺, Lgr6-eGFP⁺, and SCA-1⁺ keratinocytes (Fig. 2 C and D and SI Appendix, Figs. S2 B and C and S3). In line with previous reports (38, 40, 43), we found that these keratinocyte subpopulations were enriched for CD49f/ITGA6⁺ cells (∼95%), which is consistent with their basal layer identity (SI Appendix, Fig. S3). Similar to unsorted epidermal keratinocytes from wild-type mouse back skin (SI Appendix, Fig. S1B), individual Lgr5-eGFP⁺ and Lgr6-eGFP⁺, and SCA-1⁺ stem cells formed organoids within 1 wk (Fig. 2 E and F and SI Appendix, Fig. S2D), which could be passaged. However, both the number and the average size of the organoids differed. The SCA-1⁺ and Lgr5-eGFP⁻ fractions were more efficient in both initiating organoids and forming larger organoids (Fig. 2 E and F and SI Appendix, Fig. S2D). In short, while our culture conditions support the maintenance and expansion of multiple epidermal compartments in vitro, IFE and infundibular stem cells are more efficient in performing that task.

Whole-Transcriptome Analysis Identifies Epidermal Organoids as IFE-Like.

Multiple combinations of growth factors allowed the initial outgrowth of organoids (Fig. 1 B and C). To elucidate the differences between these conditions and to reveal the molecular phenotype of the organoids, we performed messenger RNA (mRNA)-sequencing to comprehensively compare the transcriptomes of organoid cultures. We generated libraries from organoids of primary (1 wk) and early (4–7 wk) passages kept under different media compositions. RNA isolated from HFs microdissected from murine back skin during the anagen phase or SCA-1⁺ IFE cells (further referred to as IFE) sorted from telogen back skin were used as primary tissue control (44). Our dataset consisted of 23 samples covering ∼20,000 genes in total (Fig. 3 and SI Appendix, Fig. S4).

Hierarchical clustering revealed differences between epidermal organoids and primary tissue controls (Fig. 3A). In support, differential gene-expression analysis confirmed the differences between the organoids and IFE (3,084 genes with <0.01 false discovery rate [FDR]) (SI Appendix, Fig. S4A), and between the organoids and anagen HFs (4,340 genes with <0.01 FDR) (SI Appendix, Fig. S4A). Genes encoding markers for the basal layer, IFE, HF stem cells, and HF differentiation were expressed in the organoids following the trend found in IFE (Fig. 3B). The expression of IFE and infundibulum marker Ly6a (encoding SCA-1) and absence of expression of HF stem cell and HF differentiation genes Lgr5, Lef1, and keratin genes, such as Krt73 and Krt71, in combination implicated an IFE/infundibulum identity of the murine epidermal organoids. The combined expression of basal layer markers Krt5 and Itga6 and differentiation marker Krt10 confirmed that our in vitro system mimics the in vivo epidermal stratified epithelium.

Compared with HFs, organoids had up-regulated epidermal suprabasal marker Krt6b (+16.4×) and genes involved in epidermal barrier formation, such as Sprr2d (+15.6×) (45) and Ly6a (+7.4×). Ly6a expression in organoids was similar to that of the IFE (+1.8×). Epidermal basal layer marker Krt14 (+1.4×) was significantly higher in organoids than IFE, and Krt5 expression was increased (+4.1×) in organoids compared with anagen HFs, indicating an expansion of basal keratinocytes in culture. Compared with anagen-HFs, organoids had down-regulated the expression of SG marker Scd1 (-2.0×), as well as HF markers Cux1 (−3.2×), Lef1 (−9.4×), Lgr4 (−2.0×), Lgr5 (−7.2×), and Tcf7l2 (expressing TCF4; −1.7×). In addition, genes encoding HF keratins, such as Krt35 (−12.6×), Krt71 (−17.1×), and Krt85 (−14.9×) were robustly down-regulated in cultures compared with HFs. Furthermore, organoids showed down-regulated expression of melanocyte marker Mitf (−5.5×) (46).

Next, we asked whether the transcriptome of organoids changed with time. Hierarchical clustering suggested high similarity of primary and early-passage organoids in comparison with the tissue controls (Fig. 3A). Differential gene-expression analysis revealed some minor differences between primary and early-passage cultures (668 genes with <0.01 FDR). We found decreased expression of matrix metalloproteinases, such as Mmp3 (+4.2×) and Mmp9 (+3.3×), as well as nonepithelial markers Dkk3 (+5.2×), Ednra (+3.1×), and Thy1 (+3.1×) in early-passage organoids compared with primary cultures. Several cell-type–specific genes, including Krt1, Krt10, Krt6a, Krt6b, and Krt15 did not change significantly, suggesting that the cell-type composition is not altered upon short-term culture.

We subsequently asked whether media composition significantly affected gene expression in culture. Overall, major signaling pathway components were similarly expressed in organoids and tissue controls, albeit with slightly lower expression of BMP signaling components in organoids (SI Appendix, Fig. S4B). Consistent with their comparable organoid-forming capacity (Fig. 1B), cultures with either FGF1 or FGF10 displayed almost identical gene-expression patterns (1 gene with <0.01 FDR), suggesting that they are indeed interchangeable (SI Appendix, Fig. S4C). Addition of either FGF1 or FGF10 to the medium triggered a slight change in gene expression (135 genes with <0.01 FDR). Addition of Forskolin (42 genes with <0.01 FDR) triggered only minimal changes in the gene-expression pattern, even though it greatly enhanced organoid growth. This was consistent with similar organoid morphology observed under mentioned culture conditions (SI Appendix, Fig. S1 G). Addition of TGF-βi induced the highest number of, yet still minor, changes (378 genes with <0.01 FDR) (SI Appendix, Fig. S4B). The most prominent classes of genes down-regulated by TGF-β inhibition were extracellular matrix proteins, such as Col4a1 (−4.3×), Col4a2 (−3.9×), Sparcl1 (−5.0×), as well as Tgfb3 (−4.0×) and Spon2 (−3.0×). While most cell-type–specific markers remained unchanged, we observed a decrease in Krt14 (−1×) expression at the expense of Krt15 (+2.4×). None of the culture conditions resulted in significant changes in expression of Ly6a. Thus, organoids grown under different media formulations remain very similar, with a slight decrease in basal cell (BC) markers upon TGF-β inhibition.

In conclusion, molecular profiling demonstrated that murine epidermal organoids transcriptionally resemble IFE despite the influence of small changes in media composition and remain transcriptionally stable in short term culture.

Long-Term Expansion of Epidermal Organoids.

Culture conditions that support short-term growth of organoids were comparable in maintaining an IFE-like phenotype. Thus, we further cultured organoids in EEM (N/R/F/FGF1) and obtained long-term stable growth (Fig. 1 E and F). To characterize organoids collected after long-term culturing, we first isolated RNA from organoids of early (∼passage [P]5) mid (∼P12), and late passages (∼P20), as well as keratinocytes cultured on a feeder layer (28), and performed qPCR (Fig. 4A). Although robustly expressed, we did not find significant differences in expression of markers of proliferation, the basal layer, and differentiation between different passages of organoid lines (Fig. 4A). While we saw a transient increase in Krt14 expression in early passages and a decrease in Mki67 expression in the long run, neither were statistically significant (Fig. 4A). The HF stem cell marker Cd34 was expressed in organoids at low, yet comparable levels to control keratinocytes (Fig. 4A). In addition, the marker of IFE and the HF infundibulum Ly6a was expressed at very high levels in culture, peaking at the latest passage analyzed (Fig. 4A), in line with our transcriptome data showing that epidermal organoids were enriched for Ly6a mRNA transcripts (Fig. 3B). We further found elevated levels of differentiation markers Krt10 and Flg1 (encoding Filaggrin1), which were highly expressed in the early passages (Fig. 4A).

We next performed immunohistochemistry on paraffin sections prepared from early- (∼P5) and late-passage (∼P30) organoids cultured in EEM, as well as control back skin tissue to confirm expression of key markers at the protein level. We found strong immunoreactivity against the proliferation marker KI67 in the basal layer of the organoids irrespective of passage number (Fig. 4B). In addition, suprabasal layers of the organoids were positive for the early differentiation markers KRT1 and KRT6A, as well as for the marker of terminal differentiation and keratinization Loricrin (Fig. 4B). Consistent with our transcriptome analysis of early-passage organoids, Loricrin was expressed at the protein level in passaged organoids indicating de novo differentiation in both early- and late-passage organoids. Immunofluorescence staining of organoid whole mounts showed that the late-passage organoids expressed the basal layer marker KRT5 (Fig. 4C).

It has previously been reported that cultured murine keratinocytes senesce over time, immortalize through oncogenic mutations, or acquire increased chromosome numbers (20–22). We therefore performed karyotyping on late-passage organoids (∼P30). Analysis of 2 independent cultures showed the presence of 40 chromosomes per metaphase spread, suggesting long-term genomic stability in vitro (SI Appendix, Fig. S5A). To quantify the number of actively cycling cells (in S-G₂-M phases) in our cultures over time, we flow-sorted single cells incubated with the DNA-label DAPI for subsequent cell-cycle analysis. Approximately 30% of keratinocytes in both early- and late-passage organoids were actively cycling, as opposed to 10% in freshly isolated keratinocytes (SI Appendix, Fig. S5B), demonstrating that our organoids are enriched for proliferative epidermal cells. Next, we evaluated the percentage of differentiated (cornified) organoids over different passages in our cultures by assessing positivity for eosin. This analysis revealed that 41.3 ± 2.0% (mean ± SEM) organoids established from primary keratinocytes were eosin⁺, while early-passage (P6) and late-passage (P25) organoid cultures contained, 26.4 ± 2.3% and 20.7 ± 2.0% eosin⁺ organoids, respectively (SI Appendix, Fig. S5C).

We then performed RNA sequencing on 6 late-passage organoid cultures and compared the whole-transcriptome expression levels to the above-described dataset. Hierarchical clustering suggested high similarity of primary, early-, and late-passage organoids and IFE, while HFs clustered apart (SI Appendix, Fig. S6A). Principle component analysis revealed that overall, the transcriptome of organoids remained similar over time with a slight change in transcriptome upon long-term passaging (SI Appendix, Fig. S6B). Continuous expression of Ly6a (and other key genes), demonstrated that late-passage organoids molecularly remain IFE-like over time (SI Appendix, Fig. S6C), despite the presence of differentially expressed genes upon passaging (SI Appendix, Fig. S6E). Gene ontology term analysis of the top 500 differentially expressed genes in each direction between primary and early-passage organoids and late-passage organoids did not implicate changes in epidermal specific programs (SI Appendix, Fig. S6 F and G).

Finally, we performed colony-forming efficiency (CFE) assays as following the gold standard protocol published by Jensen et al. (28) to confirm that our organoid culture retained murine epidermal stem cells long term. Keratinocytes from late-passage organoid cultures subjected to CFE assays generated holoclones containing typical cobblestone-like epidermal stem cells (Fig. 5 A–C). However, the average number of colonies almost doubled from primary organoid cultures to late-passage organoids (Fig. 5D), suggesting that our culture system enriches for colony-forming cells long term. Collectively, these data demonstrate that the proliferative capacity of murine epidermal stem cells is retained long term in the in vitro organoid system, while their ability to differentiate into IFE-like structures is similarly maintained.

Murine Epidermal Organoids Are Amenable to Genetic Manipulation.

Murine epidermal stem cells are widely used to study the biology of mammalian epidermis (1, 4, 5, 10). We therefore aimed to test whether murine epidermal organoids are amenable to genetic manipulation. We first attempted to introduce an eGFP expression vector by standard Lipofectamin-2000 transfection. However, only very few transfected cells survived. We next introduced lentiviral expression vectors by transduction (SI Appendix, Fig. S7A). Single cells derived from organoids transduced with a construct encoding Neon-tagged histone H2B (pLenti-H2B-neon-Puro) showed a robust transduction efficiency of about 10% 4 d after passaging (SI Appendix, Fig. S7B). After puromycin-based selection, we used the H2B-mNeonGreen–tagged organoids for imaging and assessed cell divisions in vitro following a previously published protocol (SI Appendix, Fig. S7C) (47). We confirmed normal cell division in late-passage organoids.

CRISPR−Cas9 genome editing is an elegant and efficient way to introduce genetic alterations into mammalian cells (48). We therefore tested whether genes-of-interest can be targeted murine epidermal organoids using CRISPR−Cas9 technology. We first introduced a lentiviral iCas9-FLAG expression vector by transduction (Fig. 6A). Subsequently, organoid lines stably expressing inducible Cas9 served as a platform to introduce gene knockouts (KO) for desmoplakin (Dsp), a gene that is mutated in multiple diseases, including skin fragility-wooly hair syndrome (49). Using 2 different single-guide (sg)RNA sequences, we successfully generated homozygous KO organoid clones lacking expression of Dsp mRNA (Fig. 6 B and C and SI Appendix, Fig. S8A). We then analyzed epidermal marker genes by qPCR (SI Appendix, Fig. S8 A and B). While expression of the basal layer marker gene Krt14 and proliferation marker gene Mki67 remained similar between the Dsp KO organoids and control organoids, expression of differentiation marker genes Krt10, Filaggrin, and Involucrin were significantly down-regulated in the Dsp KO organoids.

Transmission electron microscopy (TEM) revealed clear morphological changes upon genetic ablation of Dsp. Semithin sections stained with Toluidine blue showed clear multilayered organization of wild-type organoids (Fig. 6 D, Left), while mutant organoids (Fig. 6 D, Right) appeared less organized and less cornified. The multilayered phenotype in wild-type organoids was further confirmed by high-resolution TEM (Fig. 6 E, Left), with the BC located on the outside of the organoids. While the spinous layer (SP), granular layer (GR), keratin granules (KG), and stratum corneum (SC) were found toward the center of the organoids. The mutant organoids lacked this clear organization (Fig. 6 E, Right). Magnifications of the organoids (SI Appendix, Fig. S8 C–E, G, and H) showed less attachment between the cells in the mutant (SI Appendix, Fig. S8 D, Bottom, and H), altered adherens junction, desmosome patterning (SI Appendix, Fig. S8 E, Bottom), and atypical organization of the GR (SI Appendix, Fig. S8 E, Bottom).

Brightfield images as well as H&E stainings and stainings for KI67 and KRT10 (Fig. 6G) confirmed a similar phenotype in the Dsp KO organoids to that described for Dsp KO mice in vivo (50). Similarly, Dsp KO organoids showed altered polarization of KRT5 within basal layer cells consistent the Dsp KO mice phenotype (SI Appendix, Fig. S8F) (51). As a proof-of-concept, these experiments demonstrated that murine organoids may serve as a platform for gene tagging and gene editing in a self-renewing multilayered epithelium in vitro.

Mice.

Wild-type C57BL/6 mice at 8–18 wk of age were used, unless stated otherwise. Generation and genotyping of the Lgr5^{eGFPiresCreERT2} and Lgr6^{eGFPiresCreERT2} knockin mouse models are described elsewhere (41, 42). All mice were collected under experimental protocols approved by the Hubrecht Institute Animal Welfare Body within a project license granted by the Central Committee Animal Experimentation of the Dutch government.

Murine Epidermal Keratinocyte Isolation and Culture.

Epidermal keratinocytes were isolated from telogen back skin collected from 8- to 18-wk-old mice, as previously described (28, 44). Viable keratinocytes were plated at a density of 2,500 cells per 10 µL BME (Cultrex) drop. Cells were cultured in EEM consisting of advanced DMEM/F12 (Gibco) supplemented with penicillin/streptomycin (100 U/L; Gibco), Hepes, (10 mM; Gibco), GlutaMAX (1×; Gibco), B27 supplement (50× stock; Gibco), N-Acetylcysteine-1 (1 mM; Sigma-Aldrich), Noggin-conditioned medium (5%, in-house), R-spondin 1-conditioned medium (5%, in-house, derived from 293T-HA-RspoI-Fc cells provided by Calvin Kuo, Stanford University, Stanford, CA), acidic FGF1 (100 µg/mL; Peprotech), Heparin (0.2%; Stemcell Technologies), Forskolin (10 ng/mL; Tocris), Rho kinase inhibitor (Y-27632; 10 µM; Sigma-Aldrich), and Primocin (500× stock; Invivogen). Additional components tested include FGF10 (100 µg/mL; Peprotech), TGF-β inhibitor (A-83-01; 2 µM; Tocris), and EGF (50 ng/mL; Invitrogen). Upon seeding or passaging, medium was refreshed every 3–4 d. Organoid cultures were passaged every 7 d in an approximate ratio of 1 to 4 at a density of 2,500 viable cells per 10-µL BME drop. Organoid cultures were stored long term in −80 °C or in liquid nitrogen. Two-dimensional feeder layer-dependent culture of murine epidermal keratinocytes was described elsewhere (28). Further details are given in SI Appendix, SI Materials and Methods.

Organoid-Forming Efficiency Assays.

Organoid-forming efficiency was determined by quantification of organoid numbers and size. In brief, 8-bit binary images of whole organoid drops were analyzed with the ImageJ (NIH) images using the “analyze particles” option set to the following parameters: size, 25-infinite and circularity, 0.5–1.0 (n = 4). CellTiter-Glo luminescent cell viability assays (Promega) were used to assess organoid cell viability (n = 4).

Organoid-Forming Capacity of Sorted Keratinocytes.

Back skin from wild-type C57BL/6 mice and Lgr5^{eGFPiresCreERT2} and Lgr6^{eGFPiresCreERT2} knockin mice in telogen were dissociated into single cells. Single cells were sorted on a BD FACSJazz. Wild-type keratinocytes were stained with anti–SCA-1 conjugated APC primary antibody (clone D7, eBioscience, 17-5981-81). Postsorting, cells were plated at a density of 5,000 cells per 10 µL BME. One-week after seeding, organoid-forming capacity was assessed by quantification of organoid numbers and size, as described earlier. Eight replicative measurements were pooled per condition and mean and SEM were calculated using GraphPad Prism (biological n = 2, technical n = 4).

Determination of the Percentage of CD49f/ITGA6⁺ and CD34⁺ Cells.

Back skin from wild-type C57BL/6 mice and Lgr5^{eGFPiresCreERT2} and Lgr6^{eGFPiresCreERT2} knockin mice in telogen were dissociated into single cells. Wild-type keratinocytes were stained with anti–SCA-1 conjugated APC primary antibody (clone D7, eBioscience, 17-5981-81). Keratinocytes isolated from Lgr5^{eGFPiresCreERT2} mice were stained with anti-CD34 (Clone RAM34, BD Biosciences, 560230). All cells were stained with anti-CD49f (Clone GoH3, BD Biosciences, 555736). Single cells were sorted on a BD FACSJazz. Ten thousand cells per condition were analyzed using FlowJo software. Two biological replicative measurements were pooled per condition and mean and SEM were calculated using GraphPad Prism.

Immunohistochemistry and Immunofluorescence.

Isolated back skin or organoids were fixed in 4% formaldehyde and embedded in paraffin wax using standard protocols. Isolated back skin was embedded in optimal cutting temperature medium (Leica) for frozen sections and organoids for whole-mount staining were stored in PBS containing sodium azide at 4 °C until further use. Stainings of paraffin sections, frozen sections, tail whole mounts, and organoid whole mounts are described in SI Appendix, SI Materials and Methods.

qPCR and mRNA Sequencing.

Total RNA was extracted from microdissected anagen HFs from primary tissue using TRIzol (Ambion). RNA representative of the IFE was extracted from SCA-1⁺ basal layer cells sorted from telogen back skin using TRIzol (Ambion). To isolated IFE-specific cells from telogen back skin, we applied the sorting strategy, as shown in SI Appendix, Fig. S3A, followed by selection of CD49f⁺ cells and subsequent gating of SCA-1⁺ cells seen in SI Appendix, Fig. S3D. Total RNA was extracted from organoids using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol and processed for qPCR or mRNA-sequencing. qPCR was performed using iQTM SYBR Green Supermix (Bio-Rad) following the manufacturer’s instructions. Messenger RNA-sequencing was performed using the CEL-Seq2 method (55, 56). Primary passage organoids were harvested at 7 d after initial organoid establishment (P0d7), early-passage organoids at ∼P7d7, and late-passage organoids were harvested at ∼P26d7 and isolated with the RNeasy mini kit or viable cells were sorted into TRIzol and RNA was extracted using chloroform-based protocol following standard procedure. A full description of qPCR, calculations and mRNA-sequencing can be found in SI Appendix, SI Materials and Methods.

Determination of Cornification.

The percentage of cornification within a culture was determined by scoring the percentage of eosin⁺ organoids over the total number of organoid in H&E-stained sections of paraffin-embedded organoids (n = 4–7).

CFE Assay.

CFE assays were essentially performed following a previously published protocol (28); 5,000 cells were seeded per well. A full description of the CFE assays can be found in SI Appendix, SI Materials and Methods.

Organoid Transduction.

Organoids were grown until day 4 postpassaging after which cells were trypsinized to single cells or small organoid clumps. Cells transduced with H2B-mNeonGreen (pLV-H2B-mNeonGreen-ires-Puro). For the CRISPR–Cas9 approach, stably expressing FLAG-iCas9-Puro lines were generated by lentiviral transduction. pCW-Cas9 was a gift from Eric Lander, Broad Institute, Massachusetts Institute of Technology (MIT), Cambridge, MA and David Sabatini, Whitehead Institute for Biomedical Research, MIT, Cambridge, MA (Addgene plasmid #50661) (57). Then, stably expressing iCas9 lines served as basis for lentiviral transduction with 2 different sgRNAs targeting the Dsp gene (SI Appendix, Table S2). Dsp sgRNAs were cloned into lentiGuide-blast (cloned from LentiGuide-Puro). lentiGuide-Puro was a gift from Feng Zhang, Broad Institute, MIT, Cambridge, MA (Addgene plasmid #52963) (58). A full description of transduction of murine skin organoids can be found in SI Appendix, SI Materials and Methods.

Electron Microscopy.

For TEM, organoids were either grown according to standard procedure until the time of fixation or grown on specific planchettes for 1 d before cryofixation. A full description of electron microscopy procedures can be found in SI Appendix, SI Materials and Methods.

Statistics.

Significance was determined using unpaired Student’s t tests performed in GraphPad Prism or Microsoft Excel.