Abstract Pancreatic adenocarcinoma is the fourth leading cause of malignancy-related deaths, with rapid development of drug resistance driven by pancreatic cancer stem cells. However, the mechanisms sustaining stemness and chemotherapy resistance in pancreatic ductal adenocarcinoma (PDAC) remain unclear. Here, we demonstrate that Bicaudal C homolog 1 (BICC1), an RNA binding protein regulating numerous cytoplasmic mRNAs, facilitates chemoresistance and stemness in PDAC. Mechanistically, BICC1 activated tryptophan catabolism in PDAC by up-regulating indoleamine 2,3-dioxygenase-1 (IDO1) expression, a tryptophan-catabolizing enzyme. Increased levels of tryptophan metabolites contribute to NAD^+ synthesis and oxidative phosphorylation, leading to a stem cell-like phenotype. Blocking BICC1/IDO1/tryptophan metabolism signaling greatly improves the gemcitabine (GEM) efficacy in several PDAC models with high BICC1 level. These findings indicate that BICC1 is a critical tryptophan metabolism regulator that drives the stemness and chemoresistance of PDAC and thus a potential target for combinatorial therapeutic strategy against chemoresistance. __________________________________________________________________ BICC1 promotes tryptophan metabolism in PDAC, resulting in stemness and resistance to chemotherapy. INTRODUCTION Pancreatic ductal adenocarcinoma (PDAC), arising from the epithelial cells of the pancreatic duct, is an exceptionally lethal malignancy with a dismal 5-year survival rate of less than 12% ([60]1). Owing to its aggressive clinical features and limited treatment options, chemotherapy, the standard regimen for the clinical treatment in most patients with pancreatic cancer, yields disappointing outcomes ([61]2). Drug resistance frequently emerges in patients with both resectable and advanced PDAC accompanied by cancer recurrence, metastasis, and poor survival ([62]3). Now, gemcitabine (GEM) stands as the gold standard first-line chemotherapy agent for pancreatic cancer; however, only 23.8% of patients experience a clinical benefit, with median survival ranging from 6.7 to 8.5 months (when combined with nab-paclitaxel) ([63]4, [64]5). These unsatisfactory results underscore the urgent need to understand the mechanisms underlying chemoresistance in PDAC and identify alternative therapeutic targets that may enhance treatment efficiency. It has been reported that chemo-resistant cancer stem cells (CSCs) play a pivotal role in cancer relapse following chemotherapy ([65]6). CSCs represent a subset of tumor cells that exhibit stem cell–like characteristics and have stem cell properties ([66]7). Typically, CSCs have an abundance of protective drug transporters and activate signaling pathways that counteract apoptosis, rendering them intrinsically resistant to chemotherapy ([67]8). Furthermore, pancreatic cancer stem cells (PCSCs) are characterized by their capacity for self-renewal and differentiation, as well as their insensitivity to therapeutic drugs ([68]9). Therefore, elucidating the mechanisms by which CSC properties are maintained in PDAC may reveal how chemotherapy tolerance is developed and how it can be overcome. Bicaudal C homolog 1 (BICC1) is an RNA binding protein that posttranscriptionally regulates numerous mRNAs within the cytoplasm. BICC1 has tandem repeats of heterogeneous nuclear ribonucleoprotein K homology (KH) and KH-like (KHL) domains at the N terminus, along with serine-glycine–rich sequences and a sterile alpha motif domain at the C terminus. Through its KH region, BICC1 can bind to the 3′ untranslated region (UTR) of mRNAs, thereby regulating these mRNAs via the “AU”-enriched region ([69]10, [70]11). Despite recent investigations into the involvement of BICC1 in renal growth, bone formation, and depression, there is limited research on its role in cancer ([71]12–[72]14). Our previous work has demonstrated that BICC1 exhibits high expression levels, leading to unfavorable clinical characteristics and outcomes in patients with PDAC ([73]15). This study also noticed that BICC1 expression levels were significantly correlated with chemotherapy responses in patients. However, the precise role of BICC1 in the regulation of pancreatic CSCs and its effect on chemotherapy responses remain inadequately elucidated. Here, we present evidence that BICC1 plays a crucial role in maintaining the stemness of PDAC and augmenting the resistance of PDAC cells to chemotherapy. Overexpression of BICC1 enhances cellular tryptophan catabolism and de novo nicotinamide adenine dinucleotide (NAD^+) synthesis by up-regulating the key enzyme, IDO1, thereby promoting the stemness and intrinsic chemoresistance of PDAC. Furthermore, treatment targeting the BICC1/IDO1/tryptophan metabolic axis, in combination with GEM, demonstrates a tumor-suppressive effect in various PDAC models. This effect is observed through pharmacological inhibition and nutrient restriction experiments. Our study thus identifies BICC1 as a potential biomarker for chemotherapy outcomes and proposes that the BICC1/IDO1/tryptophan metabolic axis is potentially an alternative therapeutic target for overcoming chemoresistance in PDAC. RESULTS BICC1 facilitates the resistance to GEM chemotherapy in PDAC We collected needle biopsy PDAC tissues from 68 patients with advanced PDAC who had been treated withgone GEM plus nab-paclitaxel chemotherapy and assessed BICC1 expression using immunohistochemical (IHC) staining. According to the intensity of BICC1 staining, the patients were divided into two groups. Our analysis revealed that patients in the BICC1-low group exhibited a significantly better response than those in the BICC1-high group ([74]Fig. 1, A and B). In addition, in another cohort of 65 patients with PDAC treated with GEM-based chemotherapy after radical resection, patients with high BICC1 levels had a significantly shorter relapse-free survival compared to patients with low BICC1 levels (fig. S1, A and B). Fig. 1. BICC1 facilitates PDAC cells resistance to GEM chemotherapy. [75]Fig. 1. [76]Open in a new tab (A) Representative images of IHC staining and computed tomography scans in patients with PDAC treated with AG chemotherapy (n = 68). AG, Abraxane plus GEM. (B) Correlation analysis between BICC1 and chemotherapy response. PR, partial remission; SD, stable disease. PD, progressed disease; NPD, none progressed disease. (C) Indicated cells were treated with different GEM concentrations for 72 hours to detect their GEM 50% inhibitory concentration (IC[50]). (D) Indicated cells were treated with 400 nM GEM, and their viability was detected by real-time cell analysis (RTCA) system. (E) Indicated cells were treated with 400 nM GEM for 72 hours and collected to be labeled with annexin V and propidium iodide, followed by flow cytometry to detect their apoptosis rate. (F and G) Indicated cells (1 × 10^5) were orthotopically injected into BALB/C nude mice, treated by PBS or GEM chemotherapy (25 mg/kg) twice a week. The tumors were measured by IVIS weekly and then sectioned for TUNEL staining to determine the survival of the cancer cells (G). (H) Indicated organoids (7 to 10 days after passage) were transducted with scramble or shBICC1 lentivirus and then treated with 800 nM GEM for 72 hours, followed by TUNEL staining. Scale bars, 100 μm. All experiments were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. **P < 0.01, and ***P < 0.001. ns, not significant; NC, normal control; KO, knockout; APC, allophycocyanin; IVIS, in vivo imaging system. To precisely evaluate the effects of BICC1 on GEM resistance, we generated several PDAC cell lines with overexpression or silencing of BICC1 (fig. S1C). BICC1 overexpression markedly increased the median inhibitory concentration (IC[50]) of GEM in PDAC cells, whereas silencing of BICC1 decreased the IC[50] ([77]Fig. 1C and fig. S1D). In the absence of GEM treatment, BICC1 had minimal impact on the survival and proliferation of PDAC cells (fig. S1, E and F and H to J). However, upon GEM administration, PDAC cells overexpressing BICC1 exhibited higher viability ([78]Fig. 1D) and lower apoptosis rates ([79]Fig. 1E), indicating that forced expression of BICC1 significantly enhanced the tolerance of PDAC cells to GEM treatment in vitro. Conversely, silencing of BICC1 increased the sensitivity of PDAC cells to GEM ([80]Fig. 1, D and E, and fig. S1, G and K). We also established orthotopic tumors in mice using BICC1-overexpressing BxPC-3 cells and BICC1-silenced CFPAC-1 cells, followed by treatment with either GEM or phosphate-buffered saline (PBS). BICC1 overexpression had a negligible effect on tumors treated with PBS but significantly promoted tumor volume and growth rate in response to GEM treatment ([81]Fig. 1F and fig. S2, A to C), accompanied by increased cell mortality ([82]Fig. 1G and fig. S2D). In addition, we found that the xenografts of BICC1 knockdown PDAC cells exhibited a more fibrotic and immunosuppressing microenvironment (fig. S2, E and F). Furthermore, we rescued the expression of BICC1 in the BICC1 knockout (KO) cells, treated them with GEM, and observed that BICC1 rescued the GEM-induced apoptosis in BICC1 KO cells (fig. S2G). These data demonstrated that BICC1 levels directly regulated the PDAC cell sensitivity to GEM. Similar observations were made in PDAC patient-derived organoids. In the 12 cases of PDAC organoids, we found that those with less sensitivity to GEM were more likely to exhibited high BICC1 levels (fig. S3, A to C). Knocking down BICC1 in PDAC organoids resulted in a higher number of apoptotic cells induced by GEM compared to the controls ([83]Fig. 1H and fig. S3D). Collectively, these findings demonstrate a critical role of BICC1 in mediating chemoresistance to GEM in PDAC. In addition to GEM, we administered other chemotherapy agents or cell death inducer to the BICC1-silenced cells, followed by detection of apoptosis and IC[50]. The results revealed that BICC1 notably enhanced the resistance to 5-Fluorouracil and abraxane and moderatly increased resistance to the tumor necrosis factor–α + birinapant combination (fig. S4). Collectively, our findings indicated that BICC1 enhanced the PDAC cell resistance to various therapies, particularly GEM. BICC1 plays a pivotal role in promoting the stemness of PDAC cells Considering that the association between multi-drug resistance and the stem-like phenotype in cancer cells is well established, we sought to investigate the involvement of BICC1 in regulating PDAC cell stemness. We conducted a retrospective study in a cohort of 80 PDAC specimens. Our analysis revealed a notable correlation between high BICC1 expression and poor tumor differentiation with small-sized ducts or buddings, cubic cells with less mucin and more mitoses ([84]Fig. 2, A and B) ([85]16). To examine the relationship between tumorous BICC1 expression and markers of PCSCs, we performed a multiplex IHC assay on these specimens. The results of this assay, in conjunction with the IHC staining of serial sections, demonstrated that there was a positive correlation between BICC1 and CSC markers expression, while CSCs were more likely to exhibit a high BICC1 level ([86]Fig. 2, C to E, and fig. S5, A and B). Consistently, in a single-cell RNA sequencing analysis of human PDAC tissues (GSA:CRA001160) ([87]17), we observed a substantial correlations of BICC1 with CD133, ALDH1, and the stemness index in PDAC tumor cells (fig. S5, C to E) ([88]18), suggesting a positively correlation between the expression of BICC1 and PCSCs marker genes. Meanwhile, 28 cases of fresh PDAC tissues were collected and their BICC1 expression and PCSCs proportion were analyzed by IHC staining and flow cytometry. Notably, the proportions of PCSCs (ESA^+CD44^+CD24^+ cells) in patients with high tumorous BICC1 levels were significantly increased compared to those with low BICC1 levels ([89]Fig. 2, F and G). On the basis of the cell sorting assay, we confirmed that CD133^+ PDAC cells showed more stem-like properties and had higher BICC1 levels, which supported the correlation between BICC1 and stemness (fig. S6). Fig. 2. BICC1 promotes PDAC stemness. [90]Fig. 2. [91]Open in a new tab (A and B) Representative images of BICC1 IHC staining and hematoxylin and eosin (H&E) staining in human PDAC tissues (A) and the distribution of IHC results in poorly-differentiated or well-differentiated PDAC tissues (B). (C) Multiplex fluorescent IHC staining of BICC1 and CSC markers CD133, ALDH1, CD44, and CD24 (n = 60). CSC, cancer stem cell. (D and E) Machine learning analysis for identifying CSCs and their distribution of CSCs in BICC1^+ cells or BICC^− cells. Patients were divided into the BICC1-low and BICC1-high groups based on the IHC staining score. (F and G) Single-cell suspensions were prepared from 28 cases of fresh PDAC tissues and stained with specific antibodies against EP-CAM and PDAC CSC subsets (CD44^+CD24^+ cells). Representative dot plots of CD44^+CD24^+ cells (gated on EP-CAM^+ epithelial cells). Patients were divided into BICC1-low and BICC1-high groups based on the IHC staining score. (H) Western blots for BICC1, CD133, ALDH1, CD44, and CD24 were analyzed in indicated cell lines. β-Tubulin was used as loading control. Representative results are shown. (I to K) Proportion of CD44^+CD24^+ cells and CD133^+ cells in the indicated cells was analyzed using flow cytometry. Representative dot plots (I) and percentage of CD44^+CD24^+ cells (J) were shown. Percentage of CD133^+ cells (K) was shown. (L) Sphere formation assays were performed in the indicated cell lines. Representative images and sphere number analysis were shown. (M) In vivo limited dilution assays were performed to determine the CSC self-renewal function of the indicated cells. CSC probabilities were calculated by extreme limiting dilution assay (ELDA). Scale bars, 100 μm. All experiments above were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001. Furthermore, BICC1-overexpressing PDAC cell lines and the BICC1-silent cells were used to determine whether BICC1 regulated PDAC stemness. Using quantative polymerase chain reaction (qPCR) and Western blotting, we demonstrated that BICC1 up-regulated stemness-related genes ([92]Fig. 2H and fig. S7, A and B). Consistently, the proportion of CD44^+CD24^+ PCSCs, CD133^+ PCSCs, and ALDH^+ PCSCs was significantly increased in BICC1-overexpressing cells compared to that in the control group. By contrast, silencing of BICC1 resulted in a substantial reduction in the proportion of PCSCs ([93]Fig. 2, I to K and fig. S7, C to F). In addition, an in vitro sphere formation assay revealed that BICC1 enhanced cellular sphere formation capacity of PDAC cells ([94]Fig. 2L and fig. S8, A and B). In an in vivo limited dilution assay, we found ectopic expression of BICC1 markedly increased tumor incidence and the frequency of CSCs, whereas silencing of BICC1 reduced both the tumor incidence and CSC frequency ([95]Fig. 2M and fig. S8, C and D). Moreover, overexpressed BICC1 could rescue the decrease in PDAC stemness in BICC1 KO cells (fig. S8E). These findings strongly suggest that BICC1 has an intrinsic effect on maintaining PDAC stemness traits, which may contribute to its role in reducing the sensitivity of PDAC to GEM chemotherapy. Tryptophan metabolism, a crucial process in PDAC stemness, is regulated by BICC1 To gain further insights into the mechanism by which BICC1 promotes tumor stemness and chemoresistance, we performed Gene Ontology enrichment analysis using mRNA sequencing data from BxPC-3 cells with BICC1 knockdown and control cells (Gene Expression Omnibus database S-BSST748). The analysis revealed substantial alterations in several metabolic processes involved in stemness maintenance, including the metabolism of fat-soluble vitamin, tryptophan, and thyroid hormone ([96]Fig. 3A). Building upon these findings, we conducted an untargeted metabolomic analysis using the same cell lysates to identify the key metabolic processes regulated by BICC1. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis highlighted the most prominent changes in tryptophan metabolism ([97]Fig. 3B and fig. S9A). Subsequently, we performed liquid chromatography–mass spectrometry (LC-MS) analysis of tryptophan metabolites in PDAC cells, which revealed a strong decrease in tryptophan catabolites within the kynurenine pathway (KP) and an increase in tryptophan levels upon BICC1 knockdown ([98]Fig. 3C). These results suggested that BICC1 might promote tryptophan catabolism within the KP. Moreover, qPCR demonstrated a statistically significant up-regulation of KP enzymes induced by BICC1, and The Cancer Genome Atlas (TCGA) data revealed a positive correlation between BICC1 and the expression of these KP enzymes in PDAC, providing further support for our hypothesis (fig. S9, B and C). Fig. 3. Tryptophan metabolism is regulated by BICC1 and plays a crucial role in PDAC stemness. [99]Fig. 3. [100]Open in a new tab (A) Gene Ontology pathway analysis was performed to explore the differential signal pathways between the BxPC-3 BICC1 knockdown cells and the control cells. (B) Untargeted metabolomics analysis was performed in the BxPC-3 BICC1 knockdown and the control cells. The most significantly different KEGG pathways were shown. (C) LC-MS of tryptophan metabolites from the BICC down-regulated and control BxPC-3 cells (n = 6). (D) Sphere formation assays were performed in indicated cell lines. Representative images (left) and sphere number analysis (right) were shown. (E) Indicated cells were supplemented with tryptophan metabolites (25 μm kynurenine) for 48 hours. The proportion of CD44^+CD24^+ cells was analyzed using flow cytometry. Scale bars, 100 μm. All experiments were repeated three times independently. Paired Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001. QA, quinolinic acid; NAD, nicotinamide adenine dinucleotide; PA, picolinic acid; PE, phycoerythrin. Considering previous reports on the role of tryptophan and its KP metabolites in promoting tumorigenesis in colon cancer ([101]19) and maintaining pluripotency in embryonic stem cells ([102]20), we sought to investigate whether they exhibit similar functions in PDAC malignancy and the maintenance of PCSCs stemness. Consequently, we treated PDAC cells with kynurenine, an immediate upstream product of tryptophan, and examined the effects of KP metabolites on PDAC stemness. Our findings demonstrated that kynurenine significantly increased the proportion of CD24^+CD44^+ PCSCs, enhanced cellular sphere formation capacity in PDAC, and completely restored stemness inhibited by BICC1 knockdown ([103]Fig. 3, D and E). In addition to kynurenine, several other KP metabolites exhibited similar effect on the up-regulation of stemness-related genes and enhancement of cellular sphere formation capacity (fig. S9, D to H). Together, our data indicated that BICC1 regulated tryptophan metabolism in PDAC, which might ultimately affect stemness. BICC1 up-regulates IDO1 posttranscriptionally by stabilizing its mRNA As BICC1 regulated tryptophan metabolism in PDAC cells from metabolics study, we explored the potential BICC1-regulated mRNAs involved in tryptophan metabolism using our result from whole transcriptomic RNA-seq. Three genes involved in tryptophan metabolic process were regulated by BICC1 based on analysis of mRNA sequencing data, two of which were KP enzymes and the other one was not documented to be related to KP metabolism of tryptophan (fig. S10A). Next, we investigated whether BICC1 directly regulated these targets. Previous studies have demonstrated that BICC1 contained two KH domains, which bind to AU-rich RNA motifs, thereby enhancing mRNA stability and up-regulating the protein levels ([104]21). Because both IDO1 and KYNU had several AU-rich motifs in the 3′UTR region of their mRNAs, we performed an RNA immunoprecipitation (RIP) assay to investigate whether BICC1 binds to the mRNA of these candidate KP-related genes. The result suggested that BICC1 directly bound to IDO1 mRNA but not the KYNU mRNA ([105]Fig. 4A). Fig. 4. BICC1 up-regulates IDO1 expression by enhancing its mRNA stability. [106]Fig. 4. [107]Open in a new tab (A) RIP assessing if specific binding exists from BICC1 to IDO1 or KYNU mRNA in CFPAC-1 cells. Representative results are shown. LCN2 was used as an already-known positive control. (B) Western blots for BICC1 and IDO1 were analyzed in the indicated cell lines. β-Tubulin was used as loading control. Representative results are shown. (C) Schematic image of the mutation in IDO1 mRNA binding sites. (D and E) RNA pull-down assay using IDO1 wild-type (WT) or mutated mRNA and followed by Western blot analysis of BICC1 in CFPAC-1 cells. (F) Dual luciferase activity of reporter plasmids with the wild-type or mutated 3′UTR of IDO1 fused to the luciferase gene following BICC1 cotransduction in human embryonic kidney–293T cells. (G) Decay curve of IDO1 mRNA in the SW1990 vector/BICC1 and BxPC-3 scramble/shBICC1 cells. All experiments were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, ***P < 0.001. IgG, immunoglobulin G; EV, empty vector. Subsequent Western blotting experiments confirmed that BICC1 indeed up-regulated IDO1 expression in PDAC cells ([108]Fig. 4B). To further understand the underlying mechanism of BICC1-mediated up-regulation of IDO1, we analyzed the 3′UTR region of IDO1 mRNA and identified two AU-rich sequences (AUUUA) as potential BICC1 binding sites. To determine the essential binding site for BICC1, we mutated the sequences from TAAAT to GGGGG and performed RNA pull-down and dual-luciferase assays. The mutation of binding site 2 nearly abolished the interaction between BICC1 and IDO1 mRNA, as well as the induction of luciferase activity ([109]Fig. 4, C to F). In addition, the chromatin immunoprecipitation (ChIP) and dual-luciferase assays indicated that BICC1 had no effect on transactivation of the IDO1 promoter, suggesting that BICC1 does not influence IDO1 transcriptional activity (fig. S10, B and C). Using RNA decay assays, we verified the effect of BICC1 on the IDO1 mRNA stability. The IDO1 mRNA half-lives were measured by qPCR following treatment with actinomycin D. The results showed that BICC1 extended the half-life of IDO1 mRNA and increased its stability while having no effect on β-actin mRNA stability ([110]Fig. 4G and fig. S10, D and E). Furthermore, we observed a positive correlation between BICC1 and IDO1 expression in our cohort of human PDAC tissue specimens (fig. S10, F and G). Collectively, our data indicate that BICC1 up-regulates IDO1 by stabilizing its mRNA in PDAC, thereby explaining how BICC1 enhances tryptophan catabolism. The effect of BICC1 on PDAC stemness and resistance to GEM is dependent on IDO1 Given the crucial role of IDO1 in limiting the rate of tryptophan catabolism ([111]22), we next inquired whether the up-regulation of IDO1 was necessary for BICC1 to promote stemness and chemoresistance to GEM in PDAC cells. We conducted IDO1 knockdown experiments using BICC1-overexpressing BxPC-3 and SW1990 cells. Consistent with our previous findings, BICC1 overexpression led to increased levels of cellular KP metabolites, up-regulation of stemness-related genes, and enhanced sphere formation capacity. However, when IDO1 was knocked down in BICC1-overexpressing cells, the effects of BICC1 were completely abolished ([112]Fig. 5, A to D). To further validate these results in an in vivo setting, we performed a limited dilution assay, which demonstrated that forced expression of BICC1 significantly increased the frequency of CSCs. This increase was abrogated after IDO1 knockdown ([113]Fig. 5, E and F). As anticipated, IDO1 knockdown also remarkably inhibited the resistance to GEM induced by BICC1 in PDAC cells ([114]Fig. 5, G to I). Fig. 5. BICC1 enhances PDAC stemness and resistance to GEM via IDO1. [115]Fig. 5. [116]Open in a new tab (A) LC-MS analysis of tryptophan metabolites from the indicated cells (n = 6). (B) Quantification of KP enzyme levels normalized to β-actin detected by qPCR. (C) Western blots for BICC1, CD133, ALDH1, CD44, CD24, and IDO1 were analyzed in the indicated cell lines. β-Tubulin was used as loading control. Representative results are shown. (D) Sphere formation assays were performed in the indicated cell lines. Representative images (left) and sphere number analysis (right) were shown. (E and F) In vivo limited dilution assays were performed to determine the CSC self-renewal function of the indicated cells. Representative tumor incidence (E) and CSC probabilities (F) are shown. (G to I) Indicated cells (1 × 10^5) were subcutaneously injected into BALB/c nude mice, treated by PBS or GEM chemotherapy (25 mg/kg) twice a week. The tumors were measured weekly (H) and then sectioned for TUNEL staining to determine the survival condition of the cancer cells (G). Scale bars, 100 μm. All experiments were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001. In another assay, we used the selective IDO1 inhibitor epacadostat to treat BICC1-overexpressing PDAC cells and observed similar results to those obtained with IDO1-silent cells, in that BICC1-induced up-regulation of stemness-related genes, increase in CD24^+CD44^+ PCSCs proportion and enhancement of sphere formation capacity were all abrogated (fig. S11, A to E). Besides, epacadostat greatly attenuated GEM resistance induced by BICC1 and sensitized the cells to GEM treatment (fig. S11, F and G). Collectively, our data suggested that IDO1 was essential for BICC1-mediated promotion of PDAC stemness and resistance to GEM. To further validate the role of IDO1, we overexpressed IDO1 in those BICC1-silenced cells and examined their metabolic features, stem phenotypes, and chemotherapy sensitivities. As expected, IDO1 rescued the effectS of BICC1 in CFPAC-1 cells, including increased tryptophan catabolism, elevated levels of stemness-related genes, and reduced sensitivity to GEM (fig. S12). De novo NAD^+ synthesis and OXPHOS contribute to BCC1-induced stemness in PDAC cells To gain further insight into how tryptophan catabolism sustains stemness in PDAC, we investigated the metabolic roles of IDO1. IDO1 catalyzes the initial rate-limiting step of the KP by donating precursors for NAD^+ synthesis via the de novo pathway ([117]23) ([118]Fig. 6A). Both bioluminescent assay and LC-MS analysis demonstrated that the NAD^+/reduced form of NAD^+ (NADH) level in PDAC increased owing to BICC1 overexpression, which were completely abrogated by IDO1 knockdown ([119]Fig. 6, B and C). Considering that NAD^+ synthesis from KP metabolites was controlled by the rate-limiting enzyme quinolinate phosphoribosyltransferase (QPRT) ([120]24), we treated PDAC cells with the QPRT competitive inhibitor phthalic (PTH) acid and the KP metabolite kynurenine. As shown, PTH completely blocked the de novo NAD^+ synthesis pathway ([121]Fig. 6D). Fig. 6. BICC1 induced KP metabolism contributes to de novo NAD^+ synthesis and OXPHOS in PDAC cells. [122]Fig. 6. [123]Open in a new tab (A) Schematic diagram of the KP metabolism and NAD^+ synthesis pathways. (B to D) Intracellular NAD^+/NADH level detected by bioluminescent assay in the indicated cells (n = 6). (E to H) OCR of the indicated cells detected by the seahorse XF analyzer. (I) LC-MS analysis for M+6 NAD^+ in the EPA- or PTH-treated cells using mass-labeled ^13C-Trp. M+6 NAD^+ was synthesized from ^13C-Trp. (J) Western blots for stemness-related genes were analyzed in KYN- and PTH-treated cells. β-Tubulin was used as loading control. Representative results are shown. (K and L) Sphere formation assays were performed in treated cells. Representative images (left) and sphere number analysis (right) were shown. Scale bars, 100 μm. All experiments were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001. OCR, oxygen consumption rate; EPA, epacadostat; PTH, phthalic acid; KYN, kynurenine. It was well established that high mitochondrial activity represents an important determinant of stemness in PDAC ([124]25, [125]26). Consistent with the essential role of NAD^+/NADH in mitochondrial function and aerobic respiration, we assessed the oxygen consumption rate in PDAC cells using a Seahorse assay. Compared to control cells, BICC1-overexpressing cells exhibited significantly higher levels of oxidative phosphorylation and mitochondrial respiration. Strikingly, this effect was completely abolished by the knockdown of IDO1 ([126]Fig. 6, E and F). Moreover, PTH abrogated the kynurenine-induced oxidative phosphorylation in PDAC cells ([127]Fig. 6, G and H, and fig. S13A). Labeled precursors ^13C-Trp and D^4-NA were used to measure the NAD^+ proportions synthesized via different pathways. We observed that BICC1 specifically increased the NAD^+ level (^13C labeled, M+6) synthesized from the de novo synthesis pathway, which was abolished by the IDO1 and QPRT inhibitors ([128]Fig. 6I and fig. S13, B and C). Because the PCSCs rely more on oxidative phosphorylation (OXPHOS) than glycolysis ([129]26), we hypothesized that the elevated NAD^+ level and activated mitochondrial respiration played an intrinsic role in BICC1-induced stemness and resistance to GEM. Furthermore, Western blotting and sphere formation assays demonstrated that PTH acid inhibited the kynurenine-induced stemness and chemoresistance ([130]Fig. 6, J to L, and fig. S13D). To further investigate the role of OXPHOS in BICC1-induced stemness, we introduced the metformin as an OXPHOS inhibitor. Metformin strongly inhibited OXPHOS and completely abolished the kynurenine-induced increase of stemness (fig. S14). Consistent with previous findings, these results revealed that the activation of de novo NAD^+ synthesis pathway and OXPHOS is indispensable for BICC1 to facilitate stemness traits in PDAC. Flutamide inhibits stemness and sensitizes PDAC cells to GEM by down-regulating BICC1 expression Using a large-scale toxicogenomics database, we found that several drugs (flutamide, danazole, and omeprazole) down-regulated the mRNA levels of BICC1 in hepatocytes ([131]27) (E-MTAB-798), indicating their potential as BICC1 inhibitors. To determine whether these drugs had a similar effect on PDAC, we performed Western blotting and identified flutamide, a selective androgen receptor (AR) antagonist, as a potent inhibitor of BICC1 expression in PDAC ([132]Fig. 7A). Further analysis using Western blotting and flow cytometry to detect PCSCs revealed that flutamide suppressed PDAC stemness (fig. S15, A and B). In addition, flutamide increased the sensitivity of PDAC cells to GEM-induced apoptosis (fig. S15, C and D). Fig. 7. Targeting BICC1-IDO1-Trp metabolism in PDAC increases the efficacy of GEM. [133]Fig. 7. [134]Open in a new tab (A) Western blots for BICC1 in the CFPAC-1 cells treated with candidate drugs. β-Tubulin was used as loading control. Representative results are shown. FLU, flutamide. OME, omeprazole. DAN, danazole. (B) Survival of mice carrying orthotopic KPC xenograft after treatment with GEM, flutamide, epacadostat, low tryptophan diet, or combined (n = 6). (C) Western blots for BICC1 were analyzed in eight cases of PDX tumors . β-Tubulin was used as loading control. Representative results are shown. (D) Rate of PDX tumor growth inhibition was calculated at 100%; (treated tumor size/control tumor size). (E) PDX tumor sectioned for TUNEL stained to determine the survival condition of the cancer cells. Representative images (left) and TUNEL^+ cells proportion (right) were shown. (F) BICC1 high and low expression organoids (7 to 10 days after passage) were treated with 800 nM GEM with or without flutamide (50 μM), epacadostat (50 nM), and 20% tryptophan-containing medium for 72 hours. Representative TUNEL staining images of the organoids treated with GEM and the percentage of apoptosis cells were shown. (G) KPC mice were treated with GEM alone or combined therapy with flutamide (5 mg/kg, p.o.), epacadostat (100 mg/kg, p.o.), and a 20% tryptophan diet. Representative macroscopic images of pancreatic tumors in KPC mice were shown. The tumor burden was measured by positron emission tomography–computed tomography after 14 days of treatment. Scale bars, 1 cm (G) and 100 μm. All experiments were repeated three times independently. Student’s t test was used for statistical analysis. Data are shown as the mean value ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001. To further explore the underlying mechanism for the effect of flutamide, we treated PDAC cells with the flutamide and the AR agonist DHT. DHT up-regulated BICC1 expression, whereas flutamide down-regulated BICC1 expression, suggesting that BICC1 expression might be controlled by AR (fig. S15E). Consistently, the mRNA levels of AR were strongly correlated with those of BICC1 in the TCGA database (fig. S15F). Because AR is a ligand-dependent nuclear transcription factor ([135]28), we found two androgen response elements (AREs) within the BICC1 promoter. A ChIP assay confirmed that AR bound directly to the first ARE in the BICC1 promoter (fig. S15G). Therefore, our results demonstrated that flutamide reduced BICC1 expression by AR signaling, leading to suppression of PDAC stemness and chemoresistance. Targeting the BICC1/IDO1/tryptophan metabolism axis enhances the efficacy of GEM in PDAC To investigate whether targeting the BICC1-IDO1-Trp metabolic axis could restore the sensitivity of PDAC cells to GEM chemotherapy, we conducted experiments using BALB/c nude mice carrying orthotopic xenografts derived from KPC cells. The combined treatment of GEM with flutamide, epacadostat, or a low tryptophan diet markedly prolonged the overall survival and inhibited tumor growth compared to GEM treatment alone, while the individual treatments with flutamide, epacadostat, or a low tryptophan diet had negligible effects on overall survival and tumor growth ([136]Fig. 7B and fig. S16A). Furthermore, we used a patient-derived xenograft (PDX) model to accurately assess treatment efficacy. We evaluated the expression of BICC1 in eight cases of PDX model and selected three cases with high BICC1 expression and three cases with low BICC1 expression ([137]Fig. 7C). We divided each of these six PDX models into five groups and administered five different treatments. Tumor volume measurements and TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) analysis revealed that the combination of GEM with flutamide, epacadostat, or a low tryptophan diet had a distinct inhibitory effect on GEM-insensitive PDAC tumors with high BICC1 expression, characterized by reduced tumor growth rates and increased apoptosis compared to the group treated with GEM alone. By contrast, the inhibitory effect of the combined treatment was almost abrogated in GEM-sensitive group with low BICC1 expression ([138]Fig. 7, D and E, and fig. S16B). In addition, we carried out similar treatments in patient-derived organoid models, followed by IC[50] determination and TUNEL detection. The results showed that flutamide, epacadostat, and a low tryptophan diet led to increased sensitivity to GEM in organoids from patients, in which highly BICC1 expressed. Those combined treatments exhibited equal efficacy as GEM alone in organoids derived from patients with low BICC1 expression ([139]Fig. 7F and fig. S17A). To validate the efficacy of the combined strategy, we developed LSL-Kras^G12D/+, LSL-Trp53^R172H/+, and Pdx1-Cre (KPC) mouse models carrying spontaneous PDAC tumors. Similar effects were observed in these KPC tumors, expressing high BICC1 levels as detected by multiplex immunohistochemistry. Flutamide, epacadostat, or a low tryptophan diet significantly enhanced the effect of GEM; decreased CSC proportion on these tumors; and exhibited weaker fibrosis with fewer macrophages ([140]Fig. 7G and fig. S17, B and C). Furthermore, because of the myelosuppression and toxicity of GEM, we also checked the safety of the combined therapy. Data from laboratory tests and hematoxylin and eosin staining indicated that the combined treatments did not enhance the GEM side effects in major organs and blood cells (fig. S18). DISCUSSION Chemotherapy consists of the first-line treatment for all the stages of PDAC, but it yields suboptimal outcomes in the majority of patients ([141]3). Although cytotoxic chemotherapies can effectively eliminate bulk tumor cells, they often fail to eradicate aggressive stem cells, which have properties similar to those of normal stem cells, including drug transporters, enhanced DNA damage repair capacity, and recruitment of a protective niche ([142]7, [143]29). Hence, there is an urgent need to develop strategies that can overcome the aggressive stemness properties and GEM resistance in PDAC. In this study, we have confirmed that BICC1 regulated tryptophan metabolism in an IDO1-dependent manner, resulting in the enhancement of stemness and chemoresistance in PDAC. In our previous study, we observed widespread overexpression of BICC1, which promoted tumor growth by facilitating angiogenesis in PDAC ([144]15). Further investigation of the role of BICC1 in PDAC using clinical data revealed that patients with high BICC1 levels exhibited a poorer response to GEM-based chemotherapy and were more likely to show poor differentiation. These findings suggested that BICC1 might contribute to chemoresistance and stemness in PDAC. Supporting this notion, overexpression of BICC1 in PDAC indeed maintained stemness and strengthened cell survival in the presence of GEM. These findings have expanded our understanding of BICC1 function in tumor progression. On the basis of the RNA sequencing data, we identified multiple enriched involved in metabolic reprogramming. Because recent research suggested that the self-renewing CSCs exhibited an altered metabolic signature ([145]30), we focused on the metabolic processes regulated by BICC1. Both RNA-seq data and metabolomics analysis demonstrated that tryptophan metabolism was the most likely candidate for mediating BICC1-regulated metabolic processes. Published evidence confirmed that tryptophan and its KP metabolites contributed to tumor growth through their immunosuppressive effects in various types of cancer ([146]31, [147]32). Moreover, studies investigating their roles in tumor cells also reported that tryptophan metabolites activated PI3K-Akt or β-catenin pathway, promoting tumor proliferation and tumorigenesis in colon cancer ([148]19, [149]33). Besides, tryptophan metabolism has been recognized to play a crucial role in maintaining pluripotency in human embryonic stem cells by regulating glycolysis ([150]20). Therefore, we explored whether tryptophan metabolites exerted a similar effect on maintaining stemness in PDAC. Treatment of PDAC cells with kynurenine and several other metabolites resulted in a stronger self-renewal capacity, thereby directly supporting our conclusion. It is well established that IDO1 plays an indispensable role as the rate-limiting enzyme in the first step of tryptophan catabolism. Despite numerous works that revealed various transcriptional mechanisms underlying IDO1 overexpression in many tumors, recent studies demonstrated that posttranslational mechanisms also influence the expression and activity of IDO1 under certain circumstances ([151]34, [152]35). Given that BICC1 is an RNA binding protein with KH domain, which specifically binds to the AU-rich region of mRNA, our detection of BICC1 binding to the “AUUUA” sequence in the 3′UTR region of IDO1 mRNAs reveals a mechanism, by which BICC1 regulates tryptophan metabolism. Although IDO1 and tryptophan metabolites maintain pluripotency in human embryonic stem cells, their roles in promoting PDAC stemness remained unclear. A study by Chowdhry et al. ([153]36) found that PDAC relied more preferentially on the Preiss-Handler pathway and the de novo synthesis pathway to produce NAD^+ rather than salvage pathway, implying that tryptophan catabolism might increase NAD^+ production in PDAC. Our work demonstrated that BICC1 and tryptophan metabolites induced higher NAD^+/NADH levels in in vitro experiments. As NAD^+/NADH is required in mitochondrial respiration, we found that BICC1 and kynurenine facilitated OXPHOS in PDAC. Since it had been confirmed that PCSCs depended on OXPHOS for phenotype maintenance and survival ([154]25, [155]26), our work provides a comprehensive explanation to the BICC1-tryptophan catabolism-OXPHOS regulatory effect on PDAC stemness and drug resistance. On the basis of our findings on the BICC1 function, we considered BICC1/IDO1 inhibition as a supplementary therapeutic strategy. In consideration of the convenience and safety in clinical use, we prefer available drugs that are already in clinical applications or have been tested in previous clinical trials. Flutamide, an androgen antagonist widely used in prostate cancer, but exhibited controversial efficacy in PDAC before ([156]37, [157]38), inhibited BICC1 expression in our study. Another candidate drug was epacadostat, an IDO1 inhibitor that failed in phase III clinical trials. Regarding the importance of identifying specific patient subgroups to evaluate the clinical response, we explored whether the drug showed selective higher efficacy when combined with GEM in PDAC models with high BICC1 levels rather than those with low BICC1 levels. Notably, dietary intervention is an effective treatment with less side effect ([158]39). Hence, we practiced a low tryptophan content diet as an altered choice. These three treatments remarkably sensitized the response to GEM in both PDX and organoids model with high BICC1 expression, which had little effect in the lower BICC1 expressed model. Blockade of AR signaling also directly down-regulated BICC1, resulting in the inhibition of IDO1 and tryptophan catabolism in PDAC. To our knowledge, the interaction between these two important processes in cancer had not been previously reported. Our work serves as a basis for further research on the effect of AR signaling on tryptophan metabolism and induction of tumor stemness or immunosuppressive microenvironment, which might be highly valuable in clinical transition in pancreatic cancer and beyond. Although our findings provide a deeper understanding of the function of BICC1 in PDAC and offer experimental evidence for potential combined treatment strategies with GEM ([159]Fig. 8), several limitations to the present study should be noted. First, we only focused on the molecular function in tumor cells only, with no consideration of the cross-talk in the tumor microenvironment. Because tryptophan metabolism plays a crucial role in immune modulation, further investigations could explore the role of BICC1 in the tumor microenvironment. Second, flutamide itself was still a drug with complex pharmacological effects. A specific BICC1 selective inhibitor should be more suitable for evaluating the therapeutic potential of BICC1-targeted therapy. Meanwhile, our preliminary translational experiments only provide a potential targeting strategy to overcome GEM resistance, and there is still a long distance from its clinical application. Further clinical trials are required to evaluate the significance and usefulness of BICC1 and its downstream therapeutic targets. Fig. 8. BICC1 drives pancreatic cancer stemness and chemoresistance by facilitating tryptophan metabolism. [160]Fig. 8. [161]Open in a new tab Schematic diagram. In PDAC cells with high BICC1 levels, BICC1 up-regulated IDO1 expression by stabilize its mRNA, which contributed to the tryptophan catabolism. Increased tryptophan metabolites promoted de novo NAD^+ synthesis and OXPHOS, leading to the stemness and chemoresistance. AR antagonist flutamide could inhibit BICC1 transcription by blocking AR signaling. Flutamide and epacadostat and reduction of tryptophan sensitize the PDAC cells to GEM by targeting the BICC1-IDO1-Trp metabolic axis. MATERIALS AND METHODS Human samples collection A total of 68 needle biopsy specimens were obtained from patients with advanced PDAC at the Tianjin Medical University Cancer Institute and Hospital in China between 2017 and 2021. These patients had all undergone AG chemotherapy. In addition, 80 PDAC patients who received radical surgery R0 resection between 2011 and 2016 were retrospectively included in this study. The follow-up rate was 90%, and of the 65 cases, systematic GEM-based chemotherapy was administered. Furthermore, 28 fresh PDAC tissues were collected immediately after surgical resection from January 2020 to March 2022. All usage of these specimens and patient information was approved by the Ethics Committee of the Tianjin Medical University Cancer Institute and Hospital. All patients involved were provided written consent for the use of their specimens and disease information for future investigations according to the ethics committee and in accordance with recognized ethical guidelines of Declaration of Helsinki. Cell culture Human PDAC cell lines BxPC-3, SW1990, and CFPAC-1 were acquired from the Type Culture Collection Committee of the Chinese Academy of Sciences. The murine cell line from KPC (LSL-KrasG12D/+; LSL-Trp53R172H/+; Pdx-1-Cre) mouse was generously provided by T. Liang from the Department of Surgery at the Second Affiliated Hospital, Zhejiang University, China. These cell lines were cultured in RPMI 1640 or Iscove’s Modified Dulbecco Medium basic medium supplemented with 10% fetal bovine serum at 37°C in a humidified atmosphere of 95% air and 5% CO[2]. BICC1 overexpression was induced using doxycycline (1 μg/ml). LC-MS metabolics study T Metabolites were extracted from 1 × 107 cells lysates using methanol and analyzed using the UHPLC-Q Exactive system (Thermo Fisher Scientific). Untargeted metabolomics analysis was performed using Progenesis QI software, and pathway enrichment analysis was conducted on the basis of the KEGG pathway database. Isotope labeling was carried out using labeled compounds, including l-tryptophan (^13C[11]; Cambridge Isotope Laboratories) and NA (D4, ZZBIO) or unlabeled compounds such as l-tryptophan (Sigma-Aldrich) and NA (Sigma-Aldrich). These labeled or unlabeled compounds were added to customized RPMI media lacking l-tryptophan. PDX model experiments To establish PDXs, primary tumor specimens were collected from patients with PDAC who underwent tumor resection at Tianjin Medical University Cancer Institute and Hospital in China. Eight-week-old NSG mice were used for PDX transplantation under pathogen-free conditions. The PDX-carrying mice from the same case were randomly divided into five groups. Combined treatment was initiated when the xenografts reached a longest diameter of 3 mm. The PDX-carrying mice were administered with flutamide (5 mg/kg, qd, p.o.), epacadostat (100 mg/kg, bid, p.o.), a low-tryptophan diet (20% content of tryptophan compared to normal chow), and GEM (25 mg/kg, qw, i.p.). Tumor size was measured weekly using calipers. After 2 weeks, the mice were euthanized, and the tumors were collected and sectioned. Tumor volume was calculated using the formula: volume = length × width × width/2. All procedures were approved by the Ethics Committee of Tianjin Medical University Cancer Institute and Hospital and in accordance to National Institutes of Health (NIH) guidelines. RIP assay RIP assays were conducted using an RIP kit (Millipore), following the manufacturer’s instructions. In brief, the cells were lysed, and then mRNA fragments were immunoprecipitated using the BICC1 antibody. The immunoprecipitated samples were amplified by PCR using specific primers as indicated and further analyzed by DNA agarose gel electrophoresis. LCN2 was used as a known positive control, and ACTIN was used as a negative control. The primer sequences are listed in table S2. RNA pull-down assay RNA pull-down assays were conducted using the Magnetic RNA-Protein Pull-Down Kit (Thermo Fisher Scientific). Following the manufacturer’s instructions, we constructed plasmids with T7 promoter and wild-type or mutant IDO1 mRNA sequences as template and synthesized the mRNAs using T7 RNA polymerase in vitro. Then, we labeled these mRNA products with biotin and mixed them with the cell extract to pull-down proteins. Subsequently, Western blotting assays were performed on both the elution product and the wash supernatant fluid. NAD^+/NADH detection by bioluminescent assay The intracellular NAD^+/NADH levels were determined using the NAD/NADH-Glo Assay (Promega). Briefly, 4000 indicated cells were cultured in 96-well plates and incubated with NAD/NADH-Glo Detection Reagent for 1 hour. The luminescence was then recorded using a luminometer. The NAD^+/NADH concentration was calculated on the basis of the luminescence and a standard curve constructed using NAD^+ standards (MedChemExpress). PDAC organoid experiments The construction of PDAC organoids followed previously described methods ([162]40). For knockdown BICC1 in the organoids, the dome was digested to a single-cell suspension. Subsequently, the suspension was mixed with specific lentiviruses and polybrene for 3 hours, followed by plating on Matrigel and culturing in human complete feeding medium. To assess the response of organoids to the indicated treatments, organoid cell viability was determined using the Cell Titer-Glo assay (Promega) according to the provided instructions. For TUNEL staining, the organoids dome was dehydrated using 30% sucrose solution and fixed in 4% paraformaldehyde. Those domes were rapidly frozen in optimal cutting temperature compound using liquid nitrogen. TUNEL staining was performed after slicing using in situ cell death detection kit (Roche). KPC mouse model experiments LSL-Kras^G12D/+, LSL-Trp53^R172H/+and Pdx1-Cre mouse models were generated in-house and genotyped as previously described ([163]41). Two-month-old KPC mice with tumors of sizes similar to chosen in situ were randomized into five groups and treated with flutamide (5 mg/kg, qd, p.o.), epacadostat (100 mg/kg, bid, p.o.), a low tryptophan diet (20% content of tryptophan compared to normal chow), and GEM (25 mg/kg, qw, i.p.). Tumor burden was measured using ultrasonography and PET-CT. Tumor tissues were harvested at the end of the studies and subjected to the indicated analysis. Statistical analysis Statistical analysis was performed using SPSS Statistics 21 or GraphPad Prism software 8. Kaplan-Meier curves were used for survival analysis. Spearman’s correlation coefficients were used to investigate the relationships between the variables. Each experiment was performed in triplicate, and the values are presented as the means ± SD. Student’s t test was used for group comparisons. P values < 0.05 were considered statistically significant. Acknowledgments