1. YAP1 was expressed in airway epithelial basal cells, but not localized in nuclei during homeostasis
In normal mice, immunofluorescent staining of the airway epithelial cells showed that YAP1 was expressed in the cytoplasm of airway epithelial basal cells, not in the nucleus during homeostasis (Fig. 1A and B). After being isolated, the cells underwent double staining with CD49f-APC and EpCAM-FITC markers for flow cytometry. FACS allowed the airway epithelial cells to be sorted into two subpopulations: CD49fhighEpCAMlow basal cells and CD49flowEpCAMlhigh luminal cells (Fig. 1C). Further staining with stemness markers showed that the basal cells expressed stemness phenotypes, including Krt5, p63 and Slug, while the luminal cells exhibited differentiation phenotypes, such as Krt18 and Ecad (Fig. 1D). Thus, YAP1 was expressed in the normal airway epithelial basal cells, primarily in cytoplasm during homeostasis, indicating the involvement of YAP1 in the homeostasis maintenance of airway epithelial basal cells.
2. Overexpression of constitutively active YAP1 increased tracheosphere formation by airway epithelial basal cells
After overexpressing constitutively active YAP1 in airway epithelial basal cells, we used 3D-matrigel to demonstrate the tracheosphere formation of different groups of cells. In the first passage, the YAP1-overexpression group formed a larger number of, and more densified colonies than, the control and vector control groups (Fig. 2A). In successive passages, YAP1 overexpression continued to increase the number and density of colonies (Fig. 2B and C). Moreover, immunofluorescent staining of the 3D-matrigel sections showed that the YAP1-overexpressed airway epithelial basal cells formed colonies in a more disordered and densified manner and had increased proliferative activity compared with the vector controls, as demonstrated by the increased expression of p63 and Krt14 (Fig. 2D). These findings suggest that overexpression of constitutively active YAP1 could enhance the tracheosphere formation by airway epithelial basal cells.
3. Deletion of endogenous YAP1 inhibited tracheosphere formation of airway epithelial basal cells
Using the Cre-Lox system, we established a YAP1 knockout mouse model. Induced with tamoxifen, Cre-recombinase was activated by fusing Cre to a mutant form of the estrogen receptor (ER) and translocated into the nucleus wherein it catalyzes loxP-specific recombination events upon binding of the active tamoxifen metabolite 4-hydroxytamoxifen (4-OHT; tamoxifen-OH). To track the Cre-recombination or the YAP1-knockout efficacy, we crossed CreERT2 mice with ROSAmT/mG reporter mice. Prior to Cre recombination, cell membrane-localized tdTomato (mT) fluorescence expression was widespread in cells, showing a bright red color. After Cre recombination, cell membrane-localized EGFP (mG) fluorescence expression replaced the red fluorescence with green fluorescence. The airway tissues were obtained from the YAP1-knocked out mice after sacrificing, then the airway epithelial basal cells were isolated and purified and seeded into 24-well plates. 3D-matrigel was also added into the plate to allow tracheosphere formation. Tamoxifen-OH was added into the culture plate to activate Cre-recombination in vitro. Tracheospheres with GFP fluorescence were observed under a microscope, suggesting that YAP1 was successfully knocked out of the mouse airway epithelial basal cells (Fig. 3A). The colonies were then processed to dissociate the cells for cell passage. Meanwhile, WT mice with mTmG reporter genes were also sacrificed to obtain the airway epithelial basal cells and the colonies formed by them. Adding 2 µM of tamoxifen-OH to the ROSA-CreERT2 cells to allow for YAP1 knockout, the number of tracheospheres was decreased remarkably compared with no tamoxifen-OH addition (Fig. 3B). Adding 4 µM of tamoxifen-OH to the WT cells, the number of tracheospheres showed no obvious change compared with no tamoxifen-OH addition (Fig. 3B). These results indicate that deletion of endogenous YAP1 inhibited tracheosphere formation of airway epithelial basal cells.
4. YAP1 was related to the stemness of CD49fhighEpCAMlow cells in human LSCC tissues
Using cancer tissue samples from 18 cases of LSCC and LUAD, we prepared cell suspension, and performed double or multiple staining with TIC markers, including CD49f-APC, EpCAM-FITC, Muc1, CD44, CD24, THy and CD271. Flow cytometry analysis of the cells revealed that the proportion of CD49fhighEpCAMlow cells was significantly higher in LSCC than in LUAD (3.43% vs 0.23%) (Fig. 4A). Furthermore, stemness identification confirmed that CD49fhighEpCAMlow was the molecular phenotype of LSCC TICs. In addition, the CD49fhighEpCAMlow subpopulation of LSCC highly expressed YAP1 and TIC transcriptional factors, such as Oct-4, Sox-2, Nanog and Klf4 (Fig. 4B). 3D-matrigel culture showed that the CD49fhighEpCAMlow subpopulation formed obviously more colonies than the CD49flowEpCAMhigh subpopulation (Fig. 4C), demonstrating a strong colony-forming ability. To assay the TIC frequency, TICs (CD49fhighEpCAMlow) and non-tumor cells (NTCs; CD49flowEpCAMhigh) were inoculated into nude mice. For each cell type, the mice were divided into three groups, each group inoculated with 50,000, 5,000, and 500 cells, respectively (n = 8). In each group, the number of tumors formed in the mice was significantly larger for the TICs than for the NTCs. Thus, the TIC frequency was 100 times higher in the CD49fhighEpCAMlow subpopulation than in the CD49flowEpCAMhigh subpopulation, validating the stemness features of CD49fhighEpCAMlow cells (Fig. 4D). These results suggest that YAP1 was closely related to the stemness of TICs in human LSCC.
Combining all these above results, we speculated that YAP1 maintains and promotes colony formation by airway epithelial basal cells in the normal airways of mice. YAP1 overexpression led to disordered proliferation of airway epithelial cells, suggesting that YAP1 may promote lung cancer tumorigenesis, especially LSCC. Therefore, we planned to establish a YAP1KITrp53KO mouse model to further explore this role of YAP1.
5. Generation of homozygous YAP1KI mice and establishment of YAP1KITrp53KO mouse model
The conditional YAP1 knock-in was designed by Cyagen Biosciences (Guangzhou, China) (Supplementary Fig. S1A). The targeting vector was first constructed (Supplementary Fig. S1B). The YAP1 targeting construct was linearized by restriction digestion with NotI, followed by phenol/chloroform extraction and ethanol precipitation. The linearized vector was transfected into C57BL/6 ES cells according to Cyagen’s standard electroporation procedures. The transfected ES cells were subject to G418 selection (200 µg/mL) 24 hours post electroporation. A total of 94 G418-resistant clones were picked and amplified in 96-well plates. Two copies of 96-well plates were made, one copy frozen down and stored at − 80°C and the other copy used for DNA isolation and subsequent PCR screening for homologous recombination. PCR screening identified 25 potential targeted clones (Supplementary Fig. S1C), from among which six were expanded and further characterized by Southern blot. Five of the six expanded clones (1B6, 1C104, 1D4, 1D8 and 1H3) were confirmed to be correctly targeted (Supplementary Fig. S1D).
Targeted ES cell clone 1D8 was injected into C57BL/6 albino embryos, which were then re-implanted into CD-1 pseudo-pregnant females. Founder animals were identified by their coat color, and their germline transmission was confirmed by breeding with C57BL/6 females and subsequent genotyping of the offspring. The Neo cassette was self-deleted in germ cells so the offspring were Neo cassette-free. Four male and three female heterozygous targeted mice were generated from clone 1D8 as final deliverables. Thus, the Yap knocked-in (TurboKnockout®) mice were obtained. Next, the F1 mice of heterozygous YAP1KI genotype were crossed. The F2 mice were genotyped and crossed, and later the F3 were bred and crossed to generate sufficient homozygous mice. Genotyping of mouse tail DNA via PCR showed that 2 and 7 homozygous YAP1KI mice were obtained.
Mice with Trp53 conditionally knocked out (Trp53flox/flox) were purchased from Cyagen Biosciences. The homozygous YAP1KI mice were mated with Trp53KO mice. Their offspring underwent genotyping, and YAP1KITrp53KO mice were selected (Supplementary Fig. S1E). The whole process of YAP1KI Trp53KO mouse model establishment is shown in Supplementary Figure S1F.
6. Tumor formation was detected in YAP1KITrp53KO mice by lung volume measurement
After genotyping confirmation, the homozygous and heterozygous YAP1KITrp53KO mice were selected and bred for 1 year. Micro-CT for animals was used to perform scanning of the murine lungs in vivo. Mice that received micro-CT scanning were divided into three groups (5 mice in each group): double homozygous YAP1KITrp53KO mice, heterozygous YAP1KITrp53KO mice (heterozygous for both YAP1 and Trp53), and WT mice, with the latter two as controls. By micro-CT imaging, the lung volume for each mouse was observed and delineated (Table 1). Coronal view of the murine lungs showed that the lung volume was largest in the homozygous group and smallest in the heterozygous group (Fig. 5A). Further statistical analysis showed that the lung volume was significantly larger in the homozygous group than in the WT group and heterozygous group (p < 0.05), and was significantly larger in WT than in heterozygous mice (p < 0.05) (Fig. 5B). These results indicated that lung tumors formed in double homozygous YAP1KITrp53KO mice, and tumor formation was probably facilitated by YAP1 knock-in and Trp53 knock-out.
Table 1
The lung volume of the wild-type (WT), heterozygous and homozygous C57 mouse (V/cm3)
WT | Heterozygote | Homozygote |
373.21 | 266.06 | 448.82 |
372.11 | 213.03 | 528.91 |
331.29 | 134.18 | 354.86 |
310.77 | 177.69 | 470.44 |
322.63 | 116.68 | 392.59 |
7. Pathohistological detection revealed that LSCC was formed in YAP1KITrp53KO mice
After the discovery of tumor formation in YAP1KITrp53KO mice by micro-CT imaging, we further validated tumor formation and identified the histological tumor subtype by pathohistological staining. The mice were sacrificed, and the lungs from YAP1KITrp53KO and WT mice were dissected. Compared with the lungs from WT mice, the volume of the lung was larger in YAP1KITrp53KO mice, and the tumor xenograft could be identified clearly (Fig. 6A). The lung tissues were then processed and sliced into sections. H&E staining demonstrated that a tumor nest was formed (Fig. 6B). IHC staining was performed for the detection of pan-CK, CK5, p63, Ki67, p53, CK7, YAP1, TTF-1, NCAM1, NSE, and CgA proteins. The p53 protein was not expressed in either the nucleus or the cytoplasm (Fig. 6G), validating the successful knockout of the Trp53 gene. In contrast, the YAP1 protein was expressed primarily in the nucleus (Fig. 6I), indicating the successful knock-in of the YAP1 gene and the proliferative activity of the cells. The proteins p63 and Ki67, which are generally expressed in LSCC, were positive in the xenograft tumor tissue (Fig. 6E and F). Moreover, the LSCC-related proteins pan-CK and CK5 were also expressed in the tumor tissue (Fig. 6C and D). Meanwhile, the proteins CK7, TTF-1, NCAM1, NSE and CgA, which are mainly expressed in LUAD, were negative in the tumor tissue (Fig. 6H, J, K, L, and M). All these results confirmed the formation of lung cancer in YAP1KITrp53KO mice and indicated that the pathohistological subtype of the cancer was LSCC.
8. Genome sequencing and bioinformatic analysis of the YAP1KITrp53KO mouse
To further investigate the cancer-related genes, we performed genome sequencing and comprehensive bioinformatic analysis on the YAP1KITrp53KO mice and compared the data with those from C57 WT mice. Protein-protein interaction (PPI) network analysis generated a network of 300 differentially expressed genes (DEGs) between LSCC and normal tissues (Supplementary Fig. S2A). Further analysis revealed the top 10 primary modules of PPI sub networks, including Snap25, Syt2, Sv2b, Sv2a, Vamp2, Dlg4, Syn1, Slc17a7, Dnm1, and Stxbp1 (Supplementary Fig. S2B). The corresponding top 10 genes with a higher degree of connectivity are shown in Supplementary Table S3. The GO enrichment analysis of biological processes (BP), cellular components (CC), and molecular functions (MF) of the DEGs showed that the DEGs were mainly related to neurological functions (Supplementary Fig. S2C). GO analysis results of DEGs associated with tumor-related pathways are shown in Supplementary Table S4. YAP1 is constitutively expressed, which may partially explain the absence of YAP1 in the top 10 DEGs. The comparison between transgenic and WT mice may also produce inaccurate results. KEGG pathway enrichment analysis showed that DEGs were most enriched in pathways of neurodegeneration-multiple diseases, with significant differences (Supplementary Fig. S2D). KEGG pathway analysis revealed 24 genes related to tumorigenesis, and PPI network analysis screened out 10 genes most closely related to tumorigenesis (Supplementary Fig. S2E), among which three genes overlapped, Sv2b, Sv2a, and Dlg4 (Supplementary Fig. S2E). DEGs associated with tumor-related pathways revealed by KEGG analysis are shown in Supplementary Table S5.