To the Editor,

Tumor heterogeneity plays a key role in cancer progression and therapy resistance [1]. However, knowledge of how tumor heterogeneity arises and contributes to disease progression is still limited [2]. Recent advances in organoid culture have been successfully established in a variety of solid tumors [3,4,5]. Tumor organoids retain the histological complexity and genetic heterogeneity of parental tumors, even after many passages [6], providing a wide range of applications for cancer research. Organoids have enormous potential for the identification of optimal treatment strategies in individual patients [6]. For example, human CRC organoids derived from primary tumors [5] and liver metastases [7] have been reported as precision medical models for assessing drug responses. However, paired organoids have not been studied as a model for CRC progression. In the present study, we used paired organoids derived from primary and liver metastatic tumors of CRC patients to model cancer metastasis. Through in vitro and in vivo studies and transcriptional analyses of the paired organoids, we revealed key genes associated with CRC liver metastasis, which could be translated into therapeutic targets or prognostic biomarkers for disease treatment. A total of 24 organoids have been established (Table S1). The library contained 2-paired organoid lines from patients P13 and P21. Particularly, P13 carried two primary tumors. The 13a and 13b organoids were established from the primary tumor, while 13L organoid was established from a synchronous liver metastatic tumor. Organoids of 21a and 21L were established from primary tumor and synchronous liver metastasis of the patient P21, which data demonstrated in Additional files (Supplementary Table S1 and Fig. S1, S2, S4, and S5). Histopathological structures and the intestinal epithelial marker CDX2 of parental tumor were well preserved in organoids (Fig. S1).

Invasion is a fundamental step in tumor progression toward metastasis. To study collective invasion, we cultured paired organoids in a 3D invasion matrix (Fig. 1a). Although we did not observe collective protrusive migration in organoids derived from primary lesions, metastatic organoids exhibited robust protrusive migration into 3D invasion matrix (Fig. 1b and Fig. S2A and B). Besides, the expression level of MMP-2 (matrix metalloproteinase 2) and Ki67 was significantly higher in metastatic organoids than that in the primary organoids (Fig. S2C-E). In subcutaneous xenotransplantation of paired organoids (Fig. 1c), the growth rate and volume of 13L organoids derived xenograft tumors was significantly higher than that of 13a and 13b organoids derived tumors (Fig. 1d and e). Furthermore, we successfully generated organoids from xenografts, histology, and Ki-67 expression analysis of xenografts, as well as organoids derived from these xenografts, demonstrated similarity to the original parental tumors (Fig. 1f and Fig. S2F). We next performed splenic injection of the paired organoids to assess the development of liver metastases (Fig. 1g). The 13L organoids formed macrometastatic tumors in the livers (Fig. 1h and i), whereas 13b organoids and 13a organoids failed to colonize and had a negative expression of Ki67 in the liver (Fig. S2G).

Fig. 1
figure 1

Patient-derived paired organoids provide invasion and transplantable models for human CRC progression. a Schematic of 3D invasion assay using paired organoids. b Representative micrographs of organoids in 3D invasion assay. Tumor organoids showed the smooth and protrusive leading fronts, respectively. The scale bar represents 100 μm. c Schematic of the subcutaneous organoid injection. d Organoids were injected subcutaneously into the flank region of nude mice for 60 days. Tumor volumes were monitored over time (top). n = 5 mice per group. Error bars indicate SEMs. *p = 0.0146 (one-way ANOVA). e Tumor volume of the mice at day 60 after inoculation. Each dot indicates individual mice. n = 5 mice per group. Error bars indicate SEMs. *p = 0.0433, **p = 0.002 (one-way ANOVA). f Representative bright-field images of organoids together with H&E staining of xenografts generated from organoids, and organoids derived from the xenografts. The scale bar represents 100 μm. g Schematic of the hepatic metastasis assay by splenic organoid injection. h Representative macroscopic photographs of the whole liver. n = 5 mice per group. Top, scale bar of the whole liver represents 1 cm. Bottom, high magnification of inset. Scale bar, 5 mm. i Representative histopathology and Ki67 staining of liver metastatic lesions generated from splenic injection of 13L organoids. Top, the scale bar represents 100 μm; bottom, the scale bar represents 50 μm

We then performed gene expression analysis in the paired organoids and tumor tissue from patient 13. There were 33 genes (P < 0.05; fold change > 2.5) that were significantly upregulated in metastatic organoids (Fig. 2a and Fig. S3A and B), including the transcription factor SOX2. Previous studies have shown that SOX2 plays critical roles in embryonic pluripotent stem cells [8] and that SOX2 is abnormally expressed in many types of cancer [9,10,11,12]. The differential expression of SOX2 in paired organoids was consistent with the RNA-seq data (Fig. 2b and c), and SOX2 was also highly expressed in the metastatic tissues (Fig. 2d), while relatively low expression in normal colon tissues (Fig. S3C-F). SOX2 is also highly expressed in metastatic organoids and tissues of the other paired organoids (Fig. S4A and B).

Fig. 2
figure 2

SOX2 plays an important role in colorectal metastasis. a Hierarchical clustering heatmaps of the 33 significantly upregulated genes in 13L organoids. b qRT-PCR analysis of SOX2 in paired organoid lines. Values were normalized to mean levels in 13a organoids. Error bars indicate SEMs. **p = 0.0050, ****p < 0.0001 (one-way ANOVA). c Western blot analysis of the SOX2 protein expression in paired organoid lines. α-tubulin was used as a loading control. d Representative IHC sections for SOX2 in human colorectal tumor tissues and organoids. The scale bar represents 100 μm. e Representative micrographs of Dox-transduced the 13 L-shRNA-1/2 organoids after 5 days and stained with phalloidin–F-actin. The scale bar represents 100 μm. f Representative micrographs of colonies arising from the 13L-shRNA-1/2 organoids (top), with magnified insets showing colonies (bottom). The scale bar represents 200 μm. Colony forming efficiency in f was calculated and compared (right). Error bars indicate SEMs. **p = 0.0078, (two-way ANOVA). g Proliferation of the 13L-shRNA-1/2 organoids were examined by CTG cell viability assays following 3- or 5-days growth in the presence or absence of Dox. Error bars indicate SEMs. **p = 0.0019, ****p < 0.0001 (two-way ANOVA). h Representative macroscopic photographs (top) and histology sections (H&E, bottom) of the livers of NOD mice transplanted with the 13L-shRNA-1/2 organoids. Arrowheads, metastatic foci. n = 5 mice per group. Scale bars, 1 cm (top) and 4 mm (bottom). i Representative SOX2 immunostaining of liver metastases of the 13L-shRNA-1/2 organoids. Scale bars, 200 μm (top) and 100 μm (bottom)

To investigate the role of SOX2 in CRC progression, doxycycline (Dox) inducible expression of shRNA targeting SOX2 was established in metastatic organoids (Fig. S4C and D). SOX2 organoids exhibited the reduced ability of invasion, colony-forming efficiency, and cell viability in metastatic organoid lines (Fig. 2e-g and Fig. S4E-G). Furthermore, the metastatic organoids efficiently formed large metastatic tumors in control groups (Dox untreated), whereas the SOX2 organoids showed no or few engraftments (Fig. 2h). The downregulation of SOX2 and Ki67 was further confirmed by immunohistochemistry (Fig. 2i and Fig. S4H). We then overexpressed SOX2 in primary organoids and found that organoids with overexpressed SOX2 exhibited increased ability of invasion and proliferation when compared with control organoids (Fig. S5). Taken together, these findings demonstrate that SOX2 expression is sufficient and necessary for CRC organoids to exhibit the metastatic potential.

In summary, the present study highlights the potential of patient-derived paired primary and metastatic cancer organoids as an experimental model for investigating CRC progression. We identified a significantly dysregulated gene between paired organoids, SOX2, which could be a prognostic biomarker, and perhaps a potent therapeutic target in the treatment of CRC.