myCAFs enhanced CRC cell stemness and promote migration

To investigate the role of CAFs in CRC, we utilize single-cell sequencing data obtained from GSE166555 (comprising scRNA-seq profiles of tumor and adjacent normal tissue from 12 CRC patients without liver metastasis) and GSE178318 (encompassing scRNA-seq profiles of primary CRC, matched liver metastases, and blood samples from 6 CRC patients with liver metastases),clustering analysis found that the proportion of CAFs is higher in CRC patients with liver metastasis (CRLM) than those without LM. Specifically, myCAFs are significantly elevated in CRLM, while changes in iCAFs, apCAFs, and other CAFs are not as pronounced. (Fig. 1A, B). In the meta-GEO and TCGA cohort, a myCAF score was established to evaluate the levels of myCAFs in the bulk RNA-seq data. Patients with higher myCAF scores are predominantly diagnosed with CRC at stage III and IV, whereas those with lower myCAF scores are primarily in stages I and II. Kaplan–Meier analysis reveals that elevated myCAF scores are associated with significantly poorer prognoses in CRC patients (Fig. 1C, D, Supplementary Figure 1A).

Fig. 1

myCAFs enhanced CRC cell stemness and promote migration. A Uniform manifold approximation and projection (UMAP)plot showing the eight major cell types. Dots represent individual cells and colors represent different cell populations. (n = 18 patients). B Single cell data clustering analysis, data from GSE16655 and GSE178318. (n = 18 patients). C Proportion of patients with different Stages in high myCAF and low myCAF score groups, data obtained from the metaGEO database. (n = 913 patients). D Kaplan‒Meier survival curve of patients with high myCAF score and low myCAF score, data obtained from the metaGEO database. (n = 913 patients). E Workflows of the procedure for extracting myCAF and PDOs from CRC Patients and assessing invasion capability after co-culture with CRC Cells. F Western blot analysis of α-SMA, FAP and Collagen I protein expressions in NFs and myCAFs isolated from CRC patients. (myCAF and NF were isolated from four patients). G Immunofluorescence analysis of α-SMA, FAP and Collagen I protein expressions in NFs and myCAFs isolated from CRC patients. Scale bars = 100μm (left), Scale bars = 20μm (right). (n = 3, one of three biological replicates). H Representative α-SMA staining in CRC tissues from FUSCC cohort 2. Scale bars = 50μm. (n = 3, one of three biological replicates). I, J Transwell assay to evaluate the invasion after co-cultured DLD1 with myCAFs. Scale bars = 100μm. (Representative images are shown, data are from five biologically replicates). K, L Sphere forming assay to evaluate the stemness after co-cultured DLD1 with myCAFs. Scale bars = 100μm. (Representative images are shown, data are from five biologically replicates). M, N PDOs 3D invasion assay to evaluate the invasion of PDOs after co-cultured PDOs with myCAFs. Scale bars = 50μm. (Representative images are shown, data are from three replicates). O, P Representative images of subcutaneous tumors treated with FAPI gavage. (n = 5 mice per group). Q, R Liver metastasis assays were performed in vivo by splenic injection to evaluate the effect of FAPI on tumor metastasis. (n = 5 mice per group). Data are shown as mean ± s.e.m. For B-C, data were analyzed by chi-square test. For D, data were analyzed by log-rank test. For I-R, data were analyzed by two-tailed Student’s t-test. P values. *P < 0.05, **P < 0.01, ***P < 0.001

To investigate the influence of myCAFs on CRC progression in detail, we isolated myCAFs and normal fibroblasts (NFs) from CRC tissues and corresponding adjacent normal tissue. Additionally, organoid models were derived from the corresponding CRC tissues, following established methods (Fig. 1E). To confirm the identity of our NFs and myCAFs, we performed western blotting analysis to verify the expression of specific myCAFs biomarkers, including α-SMA, FAP, and Collagen I [26] (Fig. 1F). The data indicated that the myCAFs we extracted exhibited higher levels of these markers compared to NFs. We also assessed the cellular morphology of the two fibroblast types using light microscopy and immunofluorescence (IF) techniques. Under normal microscopy, both myCAFs and NFs displayed a similar spindle-shaped morphology (Fig. 1G, left). However, IF assays revealed that myCAFs showed higher fluorescence intensity for these markers (Fig. 1G, right). Furthermore, we utilized multicolor fluorescence staining of tissue samples from CRC patients with and without LM, to assess the proportion of myCAFs in the TME. The results also confirmed a substantial enrichment of myCAFs in CRLM patients, characterized by the red fluorescence of α-SMA (Fig. 1H).

Subsequently, we employed both cell lines and organoid models to investigate the influence of myCAFs on CRC cells within the tumor microenvironment. In the cell line experiments, we utilized well-established protocols to establish a non-contact co-culture system involving two CRC cell lines and the isolated myCAFs [27] (Fig. 1E). Notably, CRC cells co-cultured with myCAFs demonstrated a significant enhancement in migratory capacity, increased sphere formation, and elevated expression of stemness markers, suggesting that myCAFs enhance CRC cell stemness and promote metastasis in vitro (Fig. 1I-L, Supplementary Figure 1B-K). Additionally, we incorporated myCAFs into the patient-derived organoid (PDO) 3D invasion matrix using previously validated methodologies. The PDOs co-cultured with myCAFs exhibited markedly more pronounced protrusive migration into the matrix compared to those without myCAFs co-culture (Fig. 1M, N). These observations provide compelling evidence that myCAFs substantially augment the metastatic potential of CRC cells in vitro.

Finally, we employed subcutaneous tumor models and LM models in nude mice to validate the effects of CAFs in vivo. After co-injecting myCAFs with CRC cells, the mice were randomly divided into two groups: the Fibroblast Activation Protein Inhibitor (FAPI) group and the saline group. FAP is highly expressed in stroma component of TME, especially in the myCAFs. FAPI has demonstrated potential in both inhibiting fibroblast activation and diagnosing metastases in various cancer types [28]. The mice in the FAPI group were administered FAPI inhibitor orally, while the mice in the saline group were administered saline. Our results showed that the size of subcutaneous tumors and liver metastases in the FAPI group was smaller than those in the physiological saline group (Fig. 1O–R).

These findings strongly emphasize the critical role of myCAFs in the metastasis of CRC, elucidating potential avenues for therapeutic interventions targeting the dynamic interplay of myCAFs and CRC cells on tumor progression.

myCAF-derived exosome PWAR6 enhanced cell stemness and promote CRC cell migration

As previously introduced, exosomes are the primary vehicles for communication between myCAFs and other components of the TME. Using our previously established myCAFscore, we performed differential gene expression and pathway enrichment analyses on samples from the meta-GEO cohort with high and low myCAFscore. Pathways related to exosome biogenesis and secretion are highly upregulated in samples with higher myCAFscore (Supplementary Figure 2A).

To elucidate whether exosomes also contribute to our pervious finding, we co-cultured myCAFs with CRC cells in the presence of the exosome inhibitor GW4869 (Supplementary Figure 2B-C). Strikingly, subsequent Transwell and sphere formation assays showed that adding GW4869 to the co-culture system reversed the migration rate and stemness properties of CRC cells (Fig. 2A–C, Supplementary Figure 2D-H). Next, we isolate exosomes from myCAFs and co-culture them with DLD1/HCT116 cells. Similarly, we found that exosomes can promote the metastasis and stemness of these cells (Supplementary Figure 2I-M). In our 3D invasion experiments with PDOs, the addition of GW4869 to the co-culture system also significantly reversed the protrusive migration (Fig. 2D), suggesting that myCAFs facilitate metastasis and enhance stemness in CRC cells via exosome.

Fig. 2

myCAF-derived exosome PWAR6 enhanced glutamine uptake and promote CRC cell migration. A–C Evaluation of invasion and stemness of DLD1 Cells after GW4869 treatment using Transwell and spheroid formation assays. Scale bars = 100μm(upper), Scale bars = 50μm(lower). (Representative images are shown, data are from five biologically replicates). D PDO 3D invasion assay to evaluated the invasion of PDOs after GW4869 treatment. Scale bars = 50μm. (Representative images are shown, data are from three biologically replicates). E Workflows of the extraction of fibroblasts from LM tissues (myCAFs) and adjacent normal tissues (NFs) in CRLM patients, followed by the extraction of exosomes for lncRNA sequencing and metabolomics analysis. F Expression of the exosome markers TSG101, HSP70 and Alix confirmed by western blot. (n = 3, one of three biological replicates). G Transmission electron microscopy (TEM) images of exosomes secreted from myCAFs and NFs. Scale bars = 100nm. (n = 3, one of three biological replicates). H Top 100 upregulated and downregulated lncRNAs in exosomes secreted by myCAFs from CRLM tissues and paired NFs. I The expression levels of PWAR6 among normal colonic epithelial cell line (NCM460), CRC cell lines, CRC tissues (T), normal intestinal epithelial tissues (N), NFs, NF-derived exosomes, myCAFs, and myCAF-derived exosomes. (n = 3 technical replicates, one of three biological replicates). J ISH experiments in CRC cells and PDO indicate that PWAR6 is localized in both the nucleus and the cytoplasm. Scale bars = 100μm. (n = 3, one of three biological replicates). K Knockdown of PWAR6 in myCAFs. (n = 5, one of three biological replicates). L Knockdown of PWAR6 in myCAFs reduces its level in exosomes secreted by myCAFs. (n = 5, one of three biological replicates). M Incubation with exosomes from PWAR6-knockdown myCAFs significantly reduces PWAR6 levels in DLD1 cells. (n = 5, one of three biological replicates). N, O PDO 3D invasion assay to evaluated the invasion of PDOs after knocking down PWAR6. Scale bars = 50μm. (Representative images are shown, data are from three biologically replicates). P, Q PWAR6 knockdown significantly reduces the size of subcutaneous tumors. (n = 5 mice per group). R, S PWAR6 knockdown significantly reduces the size of liver metastases. (n = 3 mice per group). Data are shown as mean ± s.e.m. For A-D, I and K–O, data were analyzed by one-way ANOVA. For P-S, data were analyzed by two-tailed Student’s t-test. P values. *P < 0.05, **P < 0.01, ***P < 0.001

To elucidate the mechanisms of myCAF-derived exosomes (CAF-Exo) on CRC cells, we collected five pairs of CRLM tissues and adjacent normal liver tissues, successfully extracting NFs and myCAFs from these samples. After that, exosomes from the conditioned medium of myCAFs and NFs were isolated through ultracentrifugation [29] (Fig. 2E). To ensure that exosomes were successfully extracted, we performed Western blotting to assess the presence of several key markers in the isolated exosomes, including TSG101, HSP70, and Annexin [30] (Fig. 2F). Transmission electron microscopy (TEM) was also used to directly observe the membrane structure of exosomes secreted by myCAFs and NFs, which appeared as closed, round vesicles (Fig. 2G).

We then performed non-coding RNA sequencing on exosomes derived from myCAFs and normal fibroblasts (NF-Exo). Among all the detected lncRNAs, PWAR6 exhibited the most significant upregulation in myCAF-Exo compared to NF-Exo (Fig. 2H). To confirm these findings, we assessed PWAR6 expression levels across various samples, including myCAFs, NFs, CAF-Exo, NF-Exo, CRC cell lines, a healthy intestinal epithelial cell line, CRC tumor tissues, and adjacent normal tissues. Consistently, PWAR6 expression was significantly higher in CAFs, myCAF-Exo, and CRC tumor tissues, with no notable difference observed between CRC cell lines and the healthy intestinal epithelial cell line (Fig. 2I). In the GSE41568 dataset, PWAR6 expression was also significantly upregulated in CRC metastases compared to primary tumors (Supplementary Figure 3A). These findings underscore the substantial upregulation of PWAR6 in myCAFs and myCAF-Exo, suggesting its potential role in CRC tumor progression. Additionally, the FISH assay on PDO and CRC cells revealed that PWAR6 is primarily localized in the nucleus (Fig. 2J).

To investigate the biological function of exosomal-PWAR6 in CRC, we utilized short hairpin RNA (shRNA) to knock down the expression of PWAR6 in CAFs (Fig. 2K) and observed a corresponding decrease in the levels of PWAR6 in the secreted exosomes from myCAFs (Fig. 2L). Moreover, we co-cultured the myCAFs after PWAR6 knockdown and unmodified myCAFs with both CRC cell lines and PDOs (Fig. 2M and Supplementary Figure 3B-C). After co-culturing with PWAR6-knockdown myCAFs, the protrusive migration of PDOs into 3D matrix was significantly reduced, compared to the control group (Fig. 2N, O). Likewise, CRC cells co-cultured with PWAR6-knockdown myCAFs exhibited reduced migration and sphere formation abilities (Supplementary Figure 3D-I), along with significantly decreased expression of stemness and epithelial-mesenchymal transition markers (Supplementary Figure 3J-L). These in vitro experiments demonstrated that myCAF-derived exosome PWAR6 could enhance stemness and metastatic potential in colorectal cancer.

Subsequently, MC38 cells were co-cultured with either PWAR6-knockdown or unmodified myCAFs, then injected subcutaneously and into the spleens of C57BL/6 mice. We observed a significant decrease in the number and weight of liver metastases and subcutaneous tumors in the MC38/myCAFs-shPWAR6 group compared to the control group (Fig. 2P–S). Immunohistochemical staining (IHC) was also performed on the liver metastases of mice. The analysis revealed a remarkable decrease in the expression of Vimentin and N-cadherin, while E-cadherin expression was notably increased in the MC38/myCAFs-shPWAR6 group (Supplementary Figure 3M-N).

In summary, our findings reveal that myCAF-derived exosome PWAR6 promotes CRC stemness and metastasis in both in vitro and in vivo experiments.

Expression of exo-PWAR6 and its clinical significance in CRC patients

As previously introduced, FAPI shows potential as a novel radiotracer for tracking myCAFs and tumor lesions [31]. Recently, 68Ga-FAPI Positron Emission Tomography (68Ga-FAPI-PET) has been developed and clinically applied for imaging metastases and diagnosing tumors [32]. In our study, we conducted a comprehensive analysis of PWAR6 levels in histopathological samples from 15 patients with different 68Ga-FAPI-PET SUVmax values from the Fudan University Shanghai Cancer Center (FUSCC Cohort1). We found that the expression levels of PWAR6 were significantly higher in patients with elevated SUVmax values and lower in those with reduced SUVmax values (Fig. 3A, B). Besides, we also compared the organoid derived from these patient tissues. A higher number of organoids were observed from patients with elevated 68Ga-FAPI-PET SUVmax values (Fig. 3C left). Moreover, these organoids exhibited a higher positivity rate of PWAR6 (Fig. 3C right). Besides, we also demonstrated that samples with higher myCAF scores in GSE39582 exhibited correspondingly elevated expression levels of PWAR6 (Fig. 3I).

Fig. 3

Expression of exo-PWAR6 and its clinical significance in patients with CRC. A, B Representative image and PWAR6 expression of patients with varied 68Ga-FAPI-PET SUV max values (n = 15 patients). C Representative image of PDOs and PWAR6 ISH using tissues from patients with varied 68Ga-FAPI-PET SUV max values. Scale bars = 50μm. (n = 15 patients). D, E Representative ISH image and PWAR6 expression in healthy donors (enteritis) and CRC patients with and without LM. Scale bars = 50μm (n = 25 patients). F Representative CT or MRI Images of CRC Patients with and without LM (n = 25 patients). G Heatmap of Chi-Square test based on the association between PWAR6 and different clinicopathological factors. (n = 150 patients). H Univariate and multivariate Cox regression analyses of OS in the FUSCC cohort 3. (n = 150 patients). I PWAR6 expression level of patients with high myCAF score and low myCAF score, data obtained from the GSE39582. (n = 133 patients). K Kaplan‒Meier survival curve of patients with PWAR6-high and PWAR6-low, data obtained from the FUSCC cohort3. (n = 150 patients). Data are shown as mean ± s.e.m. For A-D and I, data were analyzed by one-way ANOVA. For G, data were analyzed by Chi-square test. For H, data were analyzed by Cox regression analysis. For J, data were analyzed by log-rank test. P values. *P < 0.05, **P < 0.01, ***P < 0.001

Next, we collected tissue samples from 5 paired normal tissues, 10 patients with non-LM CRC, and 10 CRLM patients (FUSCC cohort2). After extracting exosomes from the tissues, we used qPCR to measure PWAR6 expression in these samples. Meanwhile, we employed the chromogenic in situ hybridization (CISH) method to compare PWAR6 expression in their intestinal tissues. The results of qPCR and CISH both indicated patients with CRLM has the highest PWAR6 expression, followed by those with non-metastatic CRC and lowest in patients with enteritis (Fig. 3D, E). Representative abdominal magnetic resonance imaging (MRI) or computed tomography (CT) images of these patients are shown in Fig. 3F.

Subsequently, we investigated the pathological implications and prognostic value of PWAR6 in a separate FUSCC cohort comprising 150 paraffin-embedded CRC tissues of stages I-IV that were carefully preserved in a −80 °C refrigerator (FUSCC cohort3). Kaplan–Meier survival analysis based on FUSCC cohort3 demonstrated a clear distinction in the clinical outcomes between patients with low and high expression levels of PWAR6. Patients with low PWAR6 expression exhibited a significantly better overall survival (OS) compared to those with high expression (Fig. 3J). To further investigate the associations between PWAR6 expression and key clinicopathological factors, we conducted a Chi-square test. Our analysis revealed that distant metastasis (M stage), survival, neural invasion, vascular invasion, age and tumor size were significantly associated with PWAR6 expression (Fig. 3G). We next performed univariate and multivariate Cox regression analyses of OS in the FUSCC cohort 3. In univariate Cox regression analysis, M stage and PWAR6 expression levels were significantly associated with prognosis (P < 0.05). Multivariate Cox regression analysis further identified high PWAR6 expression as an independent predictor of poor OS (Fig. 3H).

Our study highlights that elevated PWAR6 expression, as assessed by 68Ga-FAPI-PET and IHC analysis, correlates with worse DFS and OS in CRC patients. Elevated PWAR6 levels are significantly correlated with a higher proportion of CAFs and adverse clinicopathological features among CRC patients.

CAF-derived exo-PWAR6 enhances the glutamine uptake of CRC cells via upregulation of amino acid transporters SLC38A2

Metabolomic sequencing data on previous myCAF-Exo and myNF-Exo elucidated notable differences in the glutamine metabolism pathways between myCAFs in liver metastases and NFs in normal liver tissues (Fig. 4A, B).

Fig. 4

myCAF-derived exo-PWAR6 enhances the glutamine uptake of CRC cells via upregulation of amino acid transporters SLC38A2. A, B Metabolomics Sequencing of exosomes secreted by myCAFs from CRLM tissues and paired NFs (n = 5 patients). C, D Evaluation of glutamine uptake, glutamate and α-Ketoglutarate in CRC cells co-cultured with myCAFs after knocking down PWAR6. (n = 5, one of three biological replicates). E qRT-PCR to analyze the transcriptional profiling of known glutamine transporters treated with shPWAR6/LvPWAR6 exosomes derived from myCAFs. (n = 3, one of three biological replicates). F, G Western blot to evaluate the protein level of SLC38A2 in DLD1/HCT116 cells treated with myCAF exosomes. (n = 3, one of three biological replicates). H SLC38A2 expression is elevated in metastatic compared to primary tumors, data from GSE41568 (n = 133 patients). I Immunofluorescence of NRF2 and SLC38A2. (n = 3, one of three biological replicates). J, L Evaluation of invasion and stemness of DLD1 Cells after shSLC38A2 treatment using Transwell and spheroid formation assays. Scale bars = 100μm(upper), Scale bars = 50μm(lower). (Representative images are shown, data are from three biologically replicates). M, N Expression of SLC38A2 in FUSCC Cohort 2 and representative IHC staining images. Scale bars = 50μm. (Representative images are shown, n = 25 patients). O, P Expression of SLC38A2 in FUSCC Cohort 3 and representative IHC staining images. Scale bars = 50μm. (Representative images are shown, n = 150 patients). Q IHC Staining of SLC38A2 in PDOs from patients with varied 68Ga-FAPI-PET SUVmax. Scale bars = 50μm. (Representative images are shown, n = 15 patients). R, S SLC38A2 knockdown significantly reduces the size of subcutaneous tumors. (n = 5 mice per group). T, U SLC38A2 knockdown significantly reduces the size of liver metastases. (n = 3 mice per group). Data are shown as mean ± s.e.m. For C-E, J-L, data were analyzed by one-way ANOVA. For H, M-U, data were analyzed by two-tailed Student’s t-test. P values. *P < 0.05, **P < 0.01, ***P < 0.001

Glutamine metabolism, involving a series of interconnected reactions, holds crucial implications. Notably, exogenous glutamine primarily enters cancer cells through the family of glutamine transporters. Once internalized, glutamine undergoes conversion to glutamate, which is subsequently catabolized to α-ketoglutarate (α-KG) and enters the tricarboxylic acid (TCA) cycle.

Encouragingly, our findings revealed a significant increase in glutamine uptake, as well as elevated levels of glutamate and α-ketoglutarate in CRC cells co-cultured with myCAFs, but the enhancement could be reversed by knocking down PWAR6 in CAFs (Fig. 4C, D).

This suggests an active modulation of glutamine metabolism in the presence of myCAF-derived PWAR6, likely contributing to the altered metabolic phenotype and tumor progression observed in CRC.

Subsequently, we postulated that PWAR6 might influence glutamine influx through the regulation of amino acid transporters. Glutamine transporters belonging to the solute carrier (SLC) family play a crucial role in mediating glutamine transport. To investigate this, we performed qRT-PCR to analyze the transcriptional profiling of known glutamine transporters in DLD1/HCT116 cells. We found that only SLC38A2 exhibited downregulated/upregulated expression at mRNA when DLD1/HCT116 cells were treated with shPWAR6/LvPWAR6 exosomes derived from myCAFs (Fig. 4E).

Western blot revealed that silencing PWAR6 attenuated SLC38A2, whereas overexpressing PWAR6 enhanced their expressions in DLD1/HCT116 cells treated with myCAF exosomes (Fig. 4F, G). In the GSE41568 dataset, SLC38A2 expression was also significantly upregulated in CRC metastases compared to primary tumors (Fig. 4H). Additionally, immunofluorescence assay showed that NRF2 co-localizes with SLC38A2 in the nucleus of DLD1 cells (Fig. 4I). More importantly, increase in glutamine uptake, cell migration and cell stemness by PWAR6 overexpression could be partially reversed by SLC38A2 knockdown (Fig. 4J–L and Supplementary Fig. 4A-G).

Next, we further explore the clinically significant of SLC38A2 using our previous FUSCC cohorts. In FUSCC cohort 1, our analysis revealed that patients with elevated 68Ga-FAPI-PET SUVmax values exhibited a corresponding increase in SLC38A2 expression (Fig. 4Q). Similarly, in FUSCC cohort 2, tissues from CRLM patient demonstrated significantly higher levels of SLC38A2 expression compared to those without LM (Fig. 4M, N). Furthermore, in FUSCC cohort 3, we observed that CRC tissues with elevated PWAR6 expression were associated with a concomitant upregulation of SLC38A2 (Fig. 4O, P). The TCGA database shows that there is a positive correlation between SLC38A2 and PWAR6 (Supplementary Fig. 3H).

Additionally, we also found that knocking down SLC38A2 significantly reduced the size of subcutaneous tumors and liver metastases in C57BL/6 mice model constructed as previous described (Fig. 4R–U).

Collectively, these findings indicate that myCAF-derived exosome PWAR6 promotes increased glutamine uptake in CRC cells by upregulating the amino acid transporter SLC38A2. Moreover, in vivo and in vitro experiments also demonstrate that myCAF-derived exosome PWAR6-induced upregulation of SLC38A2 can promote the metastasis of CRC cells.

PWAR6 blocks NRF2 ubiquitination and degradation

To elucidate the mechanisms by which myCAF derived PWAR6 modulates glutamine metabolism and enhances CRC metastasis, we firstly cocultured the DLD1 cells with myCAFs (Fig. 5A). Next, biotinylated sense and antisense RNAs corresponding to PWAR6 were synthesized, and subsequent RNA pulldown assays were carried out to elucidate the protein interactome associated with PWAR6 in CRC cells (Fig. 5B). The captured protein complexes were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) followed by mass spectrometry analysis. After excluding other genes, NRF2 emerged as a plausible candidate protein forming a complex with PWAR6(Fig. 5C). Nuclear fraction assay revealed that NRF2 mainly located in the nuclear (Fig. 5D), FISH assays further substantiated the co-localization of PWAR6 and NRF2 within the cellular context (Fig. 5E). To verify the PWAR6-NRF2 interaction, we performed a pull-down followed by Western blot analysis using an anti-NRF2 antibody, revealing NRF2-PWAR6 complex enrichment in the sense group (Fig. 5F upper part). Additionally, RNA immunoprecipitation (RIP) with an anti-NRF2 antibody and subsequent PCR verified the presence of PWAR6 (Fig. 5F, lower part). Next, we examined the secondary structure of PWAR6 using Vienna RNA (http://rna.tbi.univie.ac.at/) (Fig. 5G). To pinpoint the specific regions of PWAR6 binding to NRF2, we generated a series of truncated PWAR6 variants informed by the secondary structure and employed RNA pulldown combined with Western blot analysis. The results demonstrated that nucleotides 4000–4617 bp of PWAR6 interact with NRF2 (Fig. 5H).

Fig. 5

PWAR6 blocks NRF2 ubiquitination and degradation. A Co-culture of CRC cells and myCAFs before pull-down assay. B, C Silver stained SDS-PAGE gel of proteins immunoprecipitated by the sense and antisense of PWAR6. The differentially exhibited lanes were used for the mass spectrum. D Nucleus-cytoplasm separation experiment showed that NRF2 is mainly located in the nucleus. (n = 3, one of three biological replicates). E Representative images of immunofluorescent staining of NRF2 and PWAR6 in DLD1/RKO cells. Scale bars = 50μm. (n = 3, one of three biological replicates). F Western blotting assays (upper part) and RIP assay (lower part) of the specific interaction of PWAR6 with NRF2. (n = 3, one of three biological replicates). G The secondary structure of PWAR6 is shown as predicated by the centroid method (http://rna.tbi.univie.ac.at/). H Truncated PWAR6 fragments interacting with NRF2 in DLD1 cell lysates. (n = 3, one of three biological replicates). I Western blot to evaluate the protein level of NRF2 in DLD1/HCT116 cells treated with myCAF exosomes. (n = 3, one of three biological replicates). J–M Protein stability assay by using cycloheximide (CHX, 50 μg/mL) to treat cells at the different time was performed to evaluate the effect of PWAR6 overexpression (J, K) or knockdown (L, M). (n = 3, one of three biological replicates). N Western blot analysis of NRF2 in DLD1/HCT116 cells treated with myCAF overexpression or knockdown exosomes with the proteasome inhibitor MG132 (10 μM). (n = 3, one of three biological replicates). O, P Co-IP was used to assess NRF2 ubiquitination with and without PWAR6 overexpression (O) or knockdown (P), following MG132 and HA-Ub treatment. (n = 3, one of three biological replicates). Q–T Transwell assay to evaluate the DLD cell invasion (upper part and R), Scale bars = 100μm, spheroid formation assays to evaluate the DLD cell stemness (middle part and S), PDOs 3D invasion assay to evaluate the invasion of PDOs after knockdown of NRF2 (lower part and T). Scale bars = 50μm. (Representative images are shown, data are from biologically replicates). Data are shown as mean ± s.e.m. For F, K, M and Q-T, data were analyzed by one-way ANOVA. P values. *P < 0.05, **P < 0.01, ***P < 0.001

Given the observed interaction between PWAR6 and NRF2 in CRC cells, we sought to explore the regulatory role of PWAR6 on the expression of NRF2. Intriguingly, PWAR6 exhibited no impact on the mRNA abundance of NRF2(supplementary Fig. 5A-B) However, the protein level of NRF2 displayed a significant decrease upon PWAR6 knockdown and a notable increase upon PWAR6 overexpression (Fig. 5I). Next, we treat CRC cells with the protein synthesis inhibitor cycloheximide (CHX, 50ug/mL) for different periods to exam the protein stability of NRF2. Interesting, PWAR6 knockdown decreased the stability of NRF2 and accelerated its degradation (Fig. 5J, K), while PWAR6 overexpression prolonged the half-life of NRF2 degradation (Fig. 5L, M). MG132, a protease inhibitor, can accurately inhibit the ubiquitin-mediated proteasome pathway. Notably, inhibition of the proteasome through MG132 treatment resulted in the accumulation of endogenous NRF2 in CRC cells upon PWAR6 silencing, suggesting that PWAR6 regulated the NRF2 expression at the post-translational level in a ubiquitination-proteasome dependent manner (Fig. 5N).

We next used Western blotting to assess ubiquitination levels, finding that PWAR6 overexpression significantly reduced NRF2 ubiquitination (Fig. 5O). Conversely, PWAR6 knockdown led to a marked increase in NRF2 ubiquitination, with no change in the Input group (Fig. 5P). These results suggest that PWAR6 may inhibit proteasome-dependent degradation of NRF2 in CRC cells. We then analyzed cell migration, stemness, and glutamine uptake in infected cells. NRF2 silencing or overexpression was combined with PWAR6 modulation to explore NRF2's rescue effect on PWAR6-driven functions. Western blotting confirmed NRF2 upregulation in the Lv-PWAR6 group, reversible by PWAR6 depletion (supplementary Fig. 5C-D). Transwell assays showed that PWAR6 enhanced CRC cell migration, which was blocked by NRF2 depletion (Fig. 5Q upper part, Fig. 5R and supplementary Fig. 5E,F). Sphere-forming assays revealed increased stemness in the Lv-PWAR6 group, reversed by NRF2 silencing (Fig. 5Q middle part, Fig. 5S and supplementary Fig. 5E lower part, supplementary Fig. 5G). Stemness markers such as CD133 and LGR5 followed a similar pattern (supplementary Fig. 5J-K). PDO 3D invasion assays showed that PDOs co-cultured with Lv-PWAR6 CAFs had smooth, protrusive fronts, with robust migration into the 3D matrix, an effect negated by NRF2 depletion (Fig. 5Q lower part and Fig. 5T). Additionally, glutamine uptake was reduced in the sh-PWAR6 group, but NRF2 overexpression rescued this inhibition (supplementary Fig. 5H-I). Overall, these findings demonstrate that PWAR6 promotes cell migration, stemness, and glutamine uptake in CRC cells by upregulating NRF2.

To explore the mechanisms behind PWAR6-driven CRC progression, we examined its interaction with NRF2, a key protein. Using immunoprecipitation (IP) with anti-NRF2 and IgG antibodies, followed by Mass Spectrometry, we identified Keap1 as a significant NRF2-interacting protein (supplementary Fig. 6A), suggesting its role in the ubiquitin–proteasome pathway. Keap1 is known to regulate NRF2 stability by promoting its ubiquitination and degradation via Cul3-based ubiquitin ligase under non-stressed conditions, thus keeping NRF2 levels low and preventing overactivation of downstream genes.

Immunofluorescence further confirmed the co-localization of NRF2 and Keap1 within the cell (supplementary Fig. 6B). We then transfected DLD1 cells with Flag-tagged Keap1 and Myc-tagged NRF2. Co-immunoprecipitation (co-IP) analysis validated the binding between exogenous Keap1 and NRF2, underscoring their strong association (supplementary Fig. 6C-D).

Previous results indicated that PWAR6 inhibits NRF2 ubiquitination, while Keap1 promotes it, suggesting a potential interaction between them. Silencing PWAR6 significantly upregulated Keap1, whereas overexpression of PWAR6 caused a clear downregulation of Keap1 (Fig. 6G, H). These observations suggest a reciprocal regulatory relationship between PWAR6 and Keap1. To further explore Keap1's role in NRF2 downregulation via the ubiquitin–proteasome pathway, we manipulated Keap1 expression and assessed NRF2 ubiquitination levels. Overexpressing Keap1 notably increased NRF2 ubiquitination, while Keap1 knockdown greatly reduced it (supplementary Fig. 6E-F).

Fig. 6

PWAR6 accelerates migration, stemness and glutamine uptake by elevating the expression of NRF2. A The average intensity curves for NRF2 signals at TSSs in a region comprising ± 2 Kb in DLD1 and HCT116 cell lines. B The IGV shows the CUT&Tag signals of NRF2 at the SLC38A2 promoter. C Heatmap of CUT&Tag-seq peaks associated with the NRF2 in DLD1 and HCT116 cell lines, signals are displayed from −2.0 kb to + 2.0 kb surrounding the TSS. Histone H3 as positive control and IgG as negative control. D The binding sequence of NRF2 within the promoter region of the SLC38A2 gene. E Luciferase assays confirmed that NRF2 regulates SLC38A2 transcription, and PWAR6 overexpression significantly boosts NRF2 activity. (n = 3, one of three biological replicates). F Luciferase reporter assays of the transduced DLD1 cells transfected with reporter plasmids containing the SLC38A2 promoter, respectively. Wild type: −2000–0 construct; mutant: −2000–0 constructed with a point mutation at the NRF2 binding site. (n = 3, one of three biological replicates). G Westen blot of SLC38A2 after NRF2 knockdown. (n = 3, one of three biological replicates). H, I Expression of NRF2 in FUSCC Cohort 2 and representative IHC staining images. Scale bars = 50μm. (Representative images are shown, n = 25 patients). J, K Expression of NRF2 in FUSCC Cohort 3 and representative IHC staining images. Scale bars = 50μm. (Representative images are shown, n = 150 patients). L IHC Staining of NRF2 in PDOs from patients with varied 68Ga-FAPI-PET SUVmax. Scale bars = 50μm. (Representative images are shown, n = 15 patients). M, N NRF2 knockdown significantly reduces the size of subcutaneous tumors. (n = 5 mice per group). O, P NRF2 knockdown significantly reduces the size of liver metastases. (n = 3 mice per group). Data are shown as mean ± s.e.m. For E, data were analyzed by one-way ANOVA. For F, H-P, data were analyzed by two-tailed Student’s t-test. P values. *P < 0.05, **P < 0.01, ***P < 0.001

To investigate their relationship further, we introduced Flag-tagged Keap1 or Keap1 siRNA into DLD1 cells expressing Lv-PWAR6. Our findings revealed that PWAR6 could partially protect NRF2 from Keap1-mediated ubiquitination and degradation. Quantitative co-immunoprecipitation (co-IP) confirmed that PWAR6 modulates Keap1's impact on NRF2 post-translational ubiquitin modification. Notably, in CRC cells treated with MG132, PWAR6 overexpression reduced the interaction between Keap1 and NRF2, reinforcing PWAR6's regulatory role in this process (supplementary Fig. 6I-J). Besides, knockdown of Keap1 could enhance glutamine uptake of CRC cells (supplementary Fig. 6K-L).

Collect