CircKIAA1617 promotes stemness via USP14/PGRMC1-mediated autophagy and lipid metabolism reprogramming in ER-positive breast cancer.

Background
Breast cancer (BC) is a growing global health threat, with approximately 2.26 million new cases and 680 thousand deaths reported worldwide each year, representing a significant threat to the lives and well-being of women [1]. In terms of the expression of estrogen receptor (ER), progesterone receptor (PR), and human epithelial growth factor receptor 2 (HER2), breast cancer can be classified into four molecular subtypes: luminal A, luminal B, HER2-enriched, and basal‐like. Among them, estrogen receptor-positive breast cancer is the most common subtype, accounting for more than 70% of all breast cancer cases, and generally progresses in an ER-positive manner [2]. By binding to estrogen receptors (ERs), estrogen (17‐β‐estradiol, E2) serves as a critical driver of ER-positive BC, triggering the aberrant expression of numerous target genes that govern tumor proliferation, metastasis, stemness, and metabolism [3–5]. Although the classical mechanisms by which estrogen affects the malignant behaviors of ER-positive BC have been extensively studied, additional regulatory mechanisms, particularly those involving novel non-coding RNAs, may yet be identified to provide new insights for therapeutic development.
Circular RNAs (circRNAs), which are endogenous single-stranded and covalently closed circular RNAs, were first identified in 1976 [6]. Unlike canonical RNAs, circRNAs are generated by backsplicing, in which the 3’ splice donor site of an exon is covalently joined to a 5’ splice acceptor of the same or upstream exon, resulting in the formation of a circular RNA structure [7]. Due to their unique closed-loop structure, circular RNAs exhibit increased stability, higher conservation across species, and increased tolerance to RNase R degradation compared with linear RNA [8]. CircRNAs exert their effects through different mechanisms, such as interactions with microRNAs or proteins and translation into novel peptides [9]. Previous studies have proved that circRNAs can act as a class of bona fide functional molecules, providing insights into the mechanism of cancer progression. For example, circPGR functions as a competing endogenous RNA (ceRNA) to sponge miR-301a-5p and suppress ER-positive breast cancer cell growth [10]. Du et al. verified that circ-Foxo3 bound p53 and the E3 ubiquitin ligase MDM2, thereby promoting MDM2-induced p53 ubiquitination and subsequent degradation to influence the apoptosis of breast cancer cells [11]. However, the effects of estrogen on circRNA expression and the underlying roles and mechanisms have not been fully evaluated, and the identification of estrogen-responsive circRNAs and an investigation of their molecular mechanisms could provide novel tailored targets for patients with ER-positive BC.
In the present study, we identified an estrogen-responsive circRNA termed circKIAA1617 (circBase ID: hsa_ circ_0017636 [12]) in ER-positive BC whose expression was significantly upregulated in ER-positive BC tissues and exhibited prognostic significance. In vitro and in vivo studies proved that circKIAA1617 promoted the proliferation and stemness of ER-positive BC cells by inducing autophagy. Mechanistically, circKIAA1617 could be activated by estrogen and further cyclized by eukaryotic initiation factor 4A3 (EIF4A3), a key regulator of mRNA splicing and translation [13]. Furthermore, circKIAA1617 could act as a scaffold to enhance the binding between progesterone receptor membrane component 1 (PGRMC1, which is known to participates in the regulation of autophagy) and ubiquitin-specific peptidase 14 (USP14, a deubiquitinating enzyme) [14], further inhibiting the K48-linked polyubiquitination of lysine 105 of the PGRMC1 protein and blocking its proteasome degradation, which accounted for the proliferation, stemness and autophagy of ER-positive BC. Moreover, we demonstrated that the autophagy induced by the circKIAA1617/USP14/PGRMC1 axis enhanced lipophagy to modulate lipid metabolic reprogramming in ER-positive BC. In conclusion, our study revealed the roles and mechanisms of estrogen-responsive circKIAA1617 in ER-positive BC, and it shows promise as a potential diagnostic biomarker for the clinical intervention of ER-positive BC.

Materials and methods

Ethics statement and human tissue samples
Approval for all experimental protocols was secured through the Institutional Review Board of Shandong University Qilu Hospital, with written informed consent prospectively obtained from participant (Approval Number: KYLL-2022(ZM)-1058). Breast cancer tissue specimens were obtained from the Shandong University Qilu Hospital by prospective collection. Resected specimens underwent independent histopathological validation through triple-blinded consensus review by board-certified surgical pathologists, followed by immediate cryopreservation at -80 ℃. Clinicopathological data were retrieved from the institutional medical records. ER and PR status were determined by immunohistochemistry (IHC) on formalin-fixed, paraffin-embedded specimens, and HER2 status was assessed by IHC with fluorescence in situ hybridization (FISH) for cases with equivocal IHC results, following institutional laboratory protocols. Total RNA was extracted from tumor tissues and circKIAA1617 expression was quantified by qRT-PCR with normalization to endogenous controls. Patients were dichotomized into high- and low-expression groups using the median circKIAA1617 expression value as the cutoff. Survival analyses were performed using overall survival (OS) as the endpoint.

Cell culture and treatments
All cell lines were procured from the American Type Culture Collection (Manassas, VA, USA) and routinely maintained in standard media and conditions. MCF7 (RRID: CVCL_0031), T47D (RRID: CVCL_0553), and HEK293T (RRID: CVCL_0063) cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, MI, USA), 100 U/ml penicillin (Macgene, Beijing, China) and 100 µg/ml streptomycin (Macgene, Beijing, China). Cell cultures were incubated at 37 °C in a 5% CO2-humidified atmosphere. For experiments including the treatment of E2, cells were pre-treated with E2-free medium which contains charcoal-stripped FBS and phenol-red-free medium for 24 h.

Animal experiments
For xenograft proliferation studies, MCF7 overexpressing pLO5 and circKIAA1617 cells (1 × 106 cells) in 200µL of PBS containing Matrigel (1:3, v/v) was injected subcutaneously into the left flank of 4 to 6-week-old BALB/c nude female mice (n = 5 for each group). The day before MCF7 cell transplantation, mice were subcutaneously implanted with 17β-estradiol control release pellets (SE-121, Innovative Research of America). Tumor growth rate was monitored by measuring tumor diameters every 5 days. At the endpoint, the mice were sacrificed, and the xenografted tumors were measured and weighed. Hematoxylin and eosin (H&E) and immunohistochemistry (IHC) staining were performed on the sections. For in vivo limiting dilution assay, 1 × 105, 2.5 × 105, 5 × 105, 1 × 106 control or circKIAA1617-OV MCF7 cells were injected subcutaneously into 4 to 6-week-old BALB/c nude female mice. Animals were randomly assigned to experimental groups, and investigators were blinded to group allocation during data collection and analysis. All the animal experimental procedures conducted in this study received approval from the Ethical Committee of Shandong University (Approval Number: DWLL-202500074).

Statistical analysis
Statistical analysis was conducted by GraphPad Prism 10.1.2 and SPSS 25.0. All data were represented as mean ± standard deviation (SD) and are derived from a minimum of three independent experiments. Student’s t-test or one-way ANOVA was utilized to evaluate the relationship between parametric variables. Chi-square test was applied to analyze the relationships between nonparametric variables. Kaplan-Meier analysis was used to analyze the survival differences. P < 0.05 was regarded statistically significant.

Results

CircKIAA1617 expression is upregulated in ER-positive BC
CircRNA-seq was initially performed to screen for potential estrogen-responsive circRNAs associated with the estrogen-induced progression of ER-positive BC. As shown in Fig. 1. A, circRNAs that were dysregulated in E2-treated MCF7 cells were first analyzed, and the results revealed 97 upregulated circRNAs and 202 downregulated circRNAs in E2-stimulated ER-positive BC cells. In parallel, circRNAs that have been reported to be differentially expressed between normal breast tissues and ER-positive BC tissues were further screened to identify potential oncogenic circRNAs [15]. As shown in Fig. 1. B, we identified 518 upregulated circRNAs and 149 downregulated circRNAs in ER-positive BC tissues. We subsequently intersected the aforementioned results, and 21 circRNAs that were dysregulated in both E2-stimulated ER-positive BC cells and tissues were further identified (Fig. 1. C, Figure S1. A). Among the 21 circRNAs, the top 5 dysregulated circRNAs upon E2 treatment detected using RNA-seq were further selected, and detailed information on the circRNAs is shown in Figure S1. B. As shown in Figure S1. C, D, qRT-PCR assays revealed that circKIAA1617 was the only circRNA whose expression was upregulated in both E2-treated ER-positive BC cell lines and ER-positive BC tissues at the most significant level. Although the magnitude of the fold change differed between the circRNA-seq screening and the qRT-PCR validation, the upregulation trend of circKIAA1617 remained consistent. This quantitative discrepancy is commonly observed and likely attributable to the inherent differences in sensitivity and dynamic range between circRNA-seq screening and the qRT-PCR validation. Crucially, the qRT-PCR results, which serve as the gold standard for quantification, confirmed the significant overexpression of circKIAA1617 identified by the initial screening. Thus, this circRNA was selected for further study. Moreover, E2 was used to treat ER-positive BC cells in a concentration- and time-dependent manner, further indicating that circKIAA1617 is an estrogen-responsive circRNA (Fig. 1. D, E). Furthermore, an ER-specific siRNA and the ER antagonist fulvestrant were also used to antagonize the effects of ER, and the results showed that the regulatory effects of E2 on circKIAA1617 were suppressed upon ER inhibition (Figure S1. E-H). In addition, E2 was also used to treat TNBC (Triple-negative breast cancer) cell lines, and the dose and time course of E2 did not significantly affect circKIAA1617 expression in MDA-MB-231 cells, whereas the overexpression of ESR1 in MDA-MB-231 cells further led to the upregulation of circKIAA1617 expression upon E2 treatment, indicating that circKIAA1617 expression is E2 dependent (Figure S1. I-K). The detection of circKIAA1617 in breast cancer cell lines revealed that circKIAA1617 was significantly overexpressed in ER-positive BC cells (Fig. 1. F). Additionally, the expression of circKIAA1617 in breast cancer tissues was also examined using ISH and qRT-PCR, and the results further confirmed that circKIAA1617 was significantly overexpressed in ER-positive BC tissues (Fig. 1. G, H). A total of 188 patients with ER-positive BC, and the expression of circKIAA1617 was examined using qRT-PCR to explore the clinical significance of circKIAA1617; the results showed that circKIAA1617 expression was correlated with the tumor size, N status and Ki67 status in ER-positive BC (Table 1). The prognostic analysis revealed that circKIAA1617 expression was associated with shorter overall survival (OS) of patients with ER-positive BC (Fig. 1. I), and univariate and multivariate analyses further indicated that circKIAA1617 expression was an independent prognostic factor for the overall survival (OS) of patients with ER-positive BC (Table 2). Our data suggest that circKIAA1617 is an estrogen-induced circRNA that might play vital roles in ER-positive BC.

Annotations from the USCS Genome Browser revealed that circKIAA1617 originated from the 5th to 9th exons of KIAA1617 (Fig. 1. J). Specific divergent and convergent primers were designed to amplify circKIAA1617 and the linear KIAA1617 mRNA, and the specific joint sequence of head-to-tail splicing was detected in ER-positive BC cells by Sanger sequencing (Fig. 1. J). Moreover, PCR assays were performed to assess the expression of circKIAA1617 and actin using cDNA and genomic DNA (gDNA) templates from the MCF7 and T47D cell lines. Notably, while the convergent primers successfully amplified both circKIAA1617 and actin, the divergent primers specifically amplified circKIAA1617 from cDNA but not from gDNA (Fig. 1. K). RNase R was subsequently used to treat total RNA from ER-positive BC cells, and the results showed that compared with the KIAA1617 mRNA, circKIAA1617 was more resistant to digestion by RNase R, indicating the circular characteristics of circKIAA1617 (Fig. 1. L). In addition, actinomycin D assays were performed, confirming that circKIAA1617 was more stable than the linear mRNA of its parental host gene (Fig. 1. M). Total RNA from ER-positive BC cells was reverse-transcribed using either random primers or oligo (dT) primers to further validate the circular characteristics of circKIAA1617, and qRT-PCR revealed that compared with oligo (dT) primers, circKIAA1617 was clearly enriched in samples that were reverse transcribed using random primers, indicating the lack of a poly(A) tail in the circKIAA1617 sequence (Fig. 1. N). In addition, fluorescence in situ hybridization (FISH) assays and nuclear-cytoplasmic RNA extraction revealed that circKIAA1617 was located in both the cytoplasm and nucleus of MCF7 and T47D cells (Fig. 1. O, P). In conclusion, our results indicated that circKIAA1617 was an endogenously expressed estrogen-responsive circular RNA in ER-positive BC cells.

CircKIAA1617 promoted the proliferation and stemness of ER-positive BC cells in vitro and in vivo
Previous studies have proven that estrogen plays an important role in facilitating the malignant behaviors of ER-positive BC, including proliferation and stemness [16, 17]; thus, we evaluated whether estrogen-induced circKIAA1617 plays regulatory roles in ER-positive BC cells. CircKIAA1617 was initially knocked down with two specific siRNAs, and the position, sequence, and efficiency of the siRNAs are shown in Figure S2. A, B. The mRNA expression levels of its host gene KIAA1617 were also examined using qRT-PCR, which indicated that the outcomes of subsequent experiments did not result from nonspecific knockdown of its host gene (Figure S2. B). MTT and colony formation assays revealed that the knockdown of circKIAA1617 expression significantly decreased the proliferation rate (Fig. 2. A, Figure S2. C). EdU (5-ethynyl-2’-deoxyuridine) and cell cycle assays were subsequently performed and showed that the proliferative activities of MCF7 and T47D cells were impaired following circKIAA1617 silencing (Fig. 2. B, Figure S2. D). Western blot analysis further proved that circKIAA1617 knockdown led to cell cycle arrest in both ER-positive BC cell lines (Fig. 2. C). Given that estrogen can increase stemness to facilitate the progression of ER-positive BC [18], we further assessed the tumor stemness of ER-positive BC cells by performing flow cytometry, tumor sphere formation and immunofluorescence assays and demonstrated that the knockdown of circKIAA1617 markedly attenuated the stemness of ER-positive BC cells (Fig. 2. D, E; Figure S2. E). Previous studies have proven that organoids are in vitro, cell-based models derived from stem cells that possess inherent stemness potential [19]. Therefore, we further verified the association between circKIAA1617 and ER-positive BC stemness using patient-derived organoid (PDO) models, which further exhibited reduced stemness after circKIAA1617 knockdown (Figure S2. F; Fig. 2. F). The levels of proteins related to tumor stemness were also detected, with similar results (Fig. 2. G). Consistent with the in vitro results, we further confirmed that the downregulation of circKIAA1617 suppressed the proliferation and stemness of ER-positive BC cells in vivo (Fig. 2. H-J, Figure S2. G, H).

Given that our findings proved that circKIAA1617 knockdown suppressed the proliferation and stemness of ER-positive BC cells in vitro and in vivo, we further investigated the effects of circKIAA1617 overexpression to comprehensively evaluate its function. A circKIAA1617 overexpression vector (circKIAA1617-OV) was constructed and transfected into MCF7 and T47D cells, and the efficiency was evaluated using qRT-PCR (Figure S3. A). In vitro assays revealed that the overexpression of circKIAA1617 increased the proliferation of MCF7 and T47D cells (Fig. 2. K, L; Figure S3. B-D). In vivo experiments further demonstrated that the overexpression of circKIAA1617 significantly promoted the proliferation of ER-positive BC cells (Figure S3. E-G). Additionally, in vitro assays revealed that the upregulation of circKIAA1617 expression markedly increased the tumor stemness of ER-positive BC cells (Fig. 2. M, Figure S3. H, I). We also overexpressed circKIAA1617 in PDOs derived from ER-positive BC tissues, which further confirmed that circKIAA1617 increased the stemness of ER-positive BC (Fig. 2. N). Western blot analysis of stemness associated markers also proved the stemness-promoting effects of circKIAA1617 (Figure S3. J). Furthermore, rescue experiments were performed and verified that circKIAA1617 did not affect the expression of host gene KIAA1617, and the attenuation of proliferation and stemness after circKIAA1617 knockdown could be reversed by circKIAA1617 overexpression (Figure S3. K-N). We subsequently performed xenograft experiments using limiting dilution assays of MCF7 cells, which indicated that compared with control mice, mice injected with circKIAA1617-overexpressing MCF7 cells exhibited a significantly higher incidence of tumor initiation (Fig. 2. O, P). The expression of the stemness-related proteins CD44 and CD133 was detected using IHC in tumors obtained from mice, and the results were consistent with those described above (Fig. 2. Q). In conclusion, our results showed that circKIAA1617 could increase the proliferation and stemness of ER-positive BC cells both in vitro and in vivo.

CircKIAA1617 promoted proliferation and stemness by inducing autophagy in ER-positive BC cells
Since circKIAA1617 has been shown to be a cancer-promoting gene in ER-positive BC, we further identified the critical downstream pathways correlated with the oncogenic effects of circKIAA1617. RNA-seq was first performed in circKIAA1617 knockdown and control MCF7 cells to identify significantly dysregulated genes (Fig. 3. A), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis indicated that circKIAA1617 expression was significantly correlated with autophagy in ER-positive BC cells (Fig. 3. B). Monodansylcadaverine (MDC) staining was first performed to verify the correlation between circKIAA1617 expression and autophagy in ER-positive BC cells, which proved that circKIAA1617 knockdown decreased the formation of intracellular autophagosomes (Figure S4. A). Moreover, the mCherry-GFP-LC3B vector was co-transfected with circKIAA1617 siRNAs to evaluate the effects of circKIAA1617 knockdown on autophagy flux in ER-positive BC, and the results revealed that autophagy was significantly inhibited after circKIAA1617 was silenced (Fig. 3. C). Additionally, western blotting further confirmed that circKIAA1617 knockdown attenuated autophagy flux in ER-positive BC (Fig. 3. D). Furthermore, electron microscopy revealed decreased cellular autolysosome formation in circKIAA1617-knockdown cells (Fig. 3. E). In addition, we also examined the effects of circKIAA1617 overexpression on ER-positive BC autophagy. As shown in Fig. 3. F-H and Figure S4. B, C, our results proved that circKIAA1617 overexpression increased the level of autophagy flux in ER-positive BC. Taken together, our results proved that circKIAA1617 increased autophagy in ER-positive BC cells.

Bafilomycin A1 (Baf-A1, an autophagy inhibitor [20]) was used to inhibit circKIAA1617-induced autophagy, and in vitro experiments examining the proliferation and stemness of ER-positive BC cells were performed to verify the regulatory role of circKIAA1617-induced autophagy in the proliferation and stemness of ER-positive BC cells. As shown in Fig. 3. I and Figure S4. D, the suppression of autophagy by Baf-A1 antagonized the acceleration of proliferation induced by circKIAA1617. Moreover, tumor sphere formation and flow cytometry assays further demonstrated that blocking autophagy decreased the stemness of ER-positive BC cells (Fig. 3. J, K and Figure S4. E). In addition, the levels of p62 in xenograft tissues with circKIAA1617 knockdown or overexpression were also evaluated, which further confirmed the positive correlation between circKIAA1617 expression and autophagy in vivo (Fig. 3. L). In conclusion, our results confirmed that circKIAA1617 promoted proliferation and stemness via the induction of autophagy in ER-positive BC.

CircKIAA1617 was activated by classical Estrogen signaling and cyclized by EIF4A3
Based on the results described above, we determined that circKIAA1617 is an estrogen-responsive circRNA, and we further evaluated how estrogen regulates circKIAA1617 expression in ER-positive BC. The classical estrogen signaling pathway involves direct binding of ERα to the promoter to activate gene transcription [21]; thus, we hypothesized that circKIAA1617 might be transcriptionally activated by estrogen in ER-positive BC. As shown in Fig. 4. A, ChIP-seq data from Cistrome (http://cistrome.org/) revealed that ERα was enriched at the promoter region of circKIAA1617 in both the MCF7 and T47D cell lines, indicating that ERα has the potential to activate circKIAA1617 transcription. In addition, the sequence of the circKIAA1617 promoter was analyzed, and two binding motifs for ERα were predicted by the JASPAR database (https://jaspar.elixir.no/) (Fig. 4. B). pGL4.26 vectors expressing the wild-type circKIAA1617 promoter were constructed and transfected into ER-positive BC cells to evaluate the effects of estrogen on the transcriptional activity of circKIAA1617 in ER-positive BC cells. Dual-luciferase assays proved that E2 treatment increased the transcriptional activity of circKIAA1617 in a concentration- and time-dependent manner (Fig. 4. C, D), while ESR1 knockdown suppressed the transcriptional activity of circKIAA1617 (Figure S5. A), proving that estrogen could increase the transcriptional activity of the circKIAA1617 promoter. Furthermore, we constructed pGL4.26 vectors expressing the circKIAA1617 promoter with mutant motifs, and the results indicated that both motifs were crucial for the transcription of circKIAA1617 (Figure S5. B). ChIP assays were subsequently performed, which revealed that ERα could bind to both of the predicted motifs (Fig. 4. E). Moreover, the binding of ERα to both motifs could be enhanced by treatment with estrogen (Fig. 4. F). Taken together, these results indicated that estrogen induced the expression of circKIAA1617 via transcriptional activation.

Previous studies have demonstrated that some RNA-binding proteins (RBPs) can bind to circRNA flanking intron sequences and play important roles in the cyclization and biogenesis of circRNAs [13]. Given the well-documented role of ERα as an RNA binding protein [22], we first investigated whether ERα was also involved in this process and played a dual role in circKIAA1617 modulation. As shown in Figure S5. C, RNA Immunoprecipitation (RIP) assays revealed that ERα could not bind to the flanking introns of circKIAA1617. We further explored the underlying mechanism of circKIAA1617 biogenesis in ER-positive BC by predicting the potential RBPs that bind to the flanking regions of circKIAA1617 through a search of the CircInteractome database [23]. As shown in Figure S5. D, EIF4A3 and U2AF65 were suggested to have the potential to bind with the flanking regions of circKIAA1617. Specific siRNAs and overexpression vectors targeting EIF4A3 and U2AF65 were designed and transfected into ER-positive BC cells to verify the effects of EIF4A3 and U2AF65 on the expression of circKIAA1617 (Fig. 4. G, Figure S5. E), which proved that the expression of EIF4A3 but not U2AF65 was positively correlated with the expression of circKIAA1617 (Fig. 4. H, I). RIP assays were also conducted, which confirmed that endogenous EIF4A3 could bind to both upstream and downstream flanking intron sequences of circKIAA1617 (Fig. 4. J), further indicating that EIF4A3 is involved in the cyclization and biogenesis of circKIAA1617. Moreover, a total of four motifs were predicted within the flanking introns of circKIAA1617 and were individually mutated (Figure S5. F, upper panel). RIP assays showed that motif 1, motif 2 and motif 4 were responsible for the binding of EIF4A3 to the flanking introns (Figure S5. F, lower panel). An analysis of the METABRIC and TCGA databases revealed that EIF4A3 was overexpressed in ER-positive BC tissues and correlated with a poor prognosis for ER-positive BC patients, which indicated the potential functions of EIF4A3 in ER-positive BC (Fig. 4. K, L). In addition, we found that the expression of EIF4A3 could be induced by estrogen and inhibited by ESR1 knockdown in ER-positive BC cells (Fig. 4. M, N), indicating the regulatory effects of EIF4A3 on circKIAA1617 expression upon estrogen treatment. Furthermore, EIF4A3 knockdown partly reversed the effects of estrogen on upregulating circKIAA1617 expression in ER-positive BC cells, proving that EIF4A3 was crucial for estrogen-induced circKIAA1617 expression (Fig. 4. O). In conclusion, these results suggested that EIF4A3 could bind to the flanking introns of circKIAA1617 to facilitate its cyclization and biogenesis.

CircKIAA1617 directly interacted with PGRMC1
We further explored the potential molecular mechanisms by which circKIAA1617 enhances autophagy to promote the proliferation and stemness of ER-positive BC by first performing an RNA pull-down assay to explore potential proteins that bind to circKIAA1617 in MCF7 cells. As shown in Fig. 5. A, a protein band at approximately 25 kDa was markedly enriched with the circKIAA1617 sense sequence compared with the antisense sequence. A liquid chromatography-tandem mass spectrometry (LC-MS) assay was then performed to identify specific proteins of approximately 25 kDa, and 5 potential downstream targets of circKIAA1617 that were significantly enriched in the sense group were identified (Figure S6. A). Among the five proteins, progesterone receptor membrane component 1 (PGRMC1), a protein with a predicted molecular weight of 22 kDa, has been widely reported to participate in autophagy regulation [24, 25], and this PGRMC1 could be the functional target of circKIAA1617 and was selected for further analyses. Previous studies have reported that PGRCM1 can regulate autophagy in various cancers [14, 26], and we hypothesized that PGRCM1 might be the downstream functional target of circKIAA1617. The identified peptide sequences of PGRMC1 are shown in Fig. 5. B. AlphaFold3 (https://alphafoldserver.com/) was further used to predict the molecular docking between circKIAA1617 and the PGRMC1 protein, providing further evidence of the potential interaction between the two molecules (Fig. 5. C).

The subcellular localization of both circKIAA1617 and PGRMC1 was first examined by performing FISH and immunofluorescence (IF) assays to further verify the potential binding between circKIAA1617 and PGRMC1, and the results suggested that circKIAA1617 and PGRMC1 were localized in the cytoplasm of ER-positive BC cells, indicating that circKIAA1617 and PGRMC1 can bind to each other (Fig. 5. D). RNA pull-down assays were subsequently performed in both the MCF7 and T47D cell lines, which proved that PGRMC1 could interact with circKIAA1617 (Fig. 5. E). Moreover, RIP analyses further confirmed that circKIAA1617 was significantly enriched in the immunoprecipitate of PGRMC1 (Fig. 5. F). We generated two truncated isoforms of circKIAA1617, guided by the stem-loop structure predicted using RNAfold (http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi), to identify the specific regions of circKIAA1617 that are essential for the PGRMC1 interaction (Fig. 5. G). RNA pull-down assays were then employed to detect the affinity between the two isoforms and circKIAA1617, and the results showed that both isoforms had the potential to interact with PGRMC1, with the fragment 2 isoform exhibiting a higher binding affinity (Fig. 5. H). Furthermore, three truncated vectors of the PGRMC1 protein were constructed based on the special domain to identify the specific domain of PGRMC1 that is crucial for the circKIAA1617 interaction (Fig. 5. I), and the expression efficiency of the vectors was determined by western blotting (Fig. 5. J). RIP assays were subsequently performed in MCF7 and T47D cell lines transfected with full-length and truncated PGRMC1, confirming that the N-terminal domain spanning residues 1–72 of PGRMC1 constitutes the principal region responsible for mediating its interaction with circKIAA1617 (Fig. 5. K). In conclusion, our results proved that circKIAA1617 could directly interact with the PGRMC1 protein in ER-positive BC cells.
Building upon the established direct interaction between circKIAA1617 and PGRMC1 protein, we investigated whether the expression of PGRMC1 could be influenced by circKIAA1617. Knockdown or overexpression assays showed that circKIAA1617 expression was positively correlated with PGRMC1 protein expression but did not affect its mRNA level (Figure S6. B, C; Fig. 5. L, M), indicating that circKIAA1617 might influence the posttranslational modification of PGRMC1. Moreover, rescue assays were performed in which circKIAA1617 was overexpressed and PGRMC1 was knocked down in ER-positive BC cells to verify whether PGRMC1 is a functional downstream target of circKIAA1617. As shown in Fig. 5. N-P, PGRMC1 knockdown inhibited the induction of autophagy by circKIAA1617. Moreover, in vitro assays further proved that the oncogenic effects of circKIAA1617 on ER-positive BC proliferation and stemness could be inhibited by the PGRMC1 siRNA (Fig. 5. Q, R; Figure S6. D, E). Additionally, PGRMC1 was inhibited by a specific inhibitor (AG-205), further demonstrating that PGRMC1 is functional downstream of circKIAA1617 in ER-positive BC (Figure S6. F-J). In conclusion, the above results confirmed that circKIAA1617 could interact with and modulate the expression of the PGRMC1 protein, a functional target that was responsible for the circKIAA1617-induced autophagy and malignant behaviors of ER-positive BC cells.

CircKIAA1617 served as a scaffold to enhance the interaction between PGRMC1 and USP14
Based on the results described, we further evaluated how circKIAA1617 affects the expression of the PGRMC1 protein. Since numerous studies have demonstrated that circRNAs can serve as scaffolds to enhance protein binding and further modulate the stabilization and function of proteins [27], we hypothesized that circKIAA1617 might regulate the interaction between PGRMC1 and its regulatory protein. The RNA pull-down results are shown in Fig. 5. A, and ubiquitin-specific proteinase 14 (USP14), a deubiquitinating enzyme, was selected because of its ability to regulate the ubiquitin-proteasome system [28]. The peptide sequences of USP14 identified by LC-MS are shown, and the molecular docking between circKIAA1617 and the USP14 protein was predicted by AlphaFold3 (Fig. 6. A, B). FISH and IF assays proved that circKIAA1617 and USP14 could colocalize in the cytoplasm of MCF7 and T47D cells (Fig. 6. C). RNA pull-down and RIP assays further showed the interaction between circKIAA1617 and USP14 (Fig. 6. D, E). Moreover, two truncation vectors based on the structural domains of USP14 were constructed (Fig. 6. F, Figure S7. A), and RIP assays were performed to determine the specific domain of USP14 that could bind to circKIAA1617, revealing that the C-terminal domain of USP14 was responsible for the interaction between circKIAA1617 and USP14 in ER-positive BC cells (Figure S7. B).

Since we have proved circKIAA1617 could interact with both the PGRMC1 and USP14 proteins, the binding between PGRMC1 and USP14 was further evaluated. As shown in Fig. 6. G and Figure S7. C, immunoprecipitation (IP) assays were performed with specific antibodies against PGRMC1 and USP14, and the results revealed that PGRMC1 and USP14 could bind to each other. The evidence obtained above suggested that circKIAA1617 might function as a central scaffold to enhance the binding between PGRMC1 and USP14. The effects of circKIAA1617 expression on the binding affinity between PGRMC1 and USP14 were further detected via IP assays to validate our hypothesis, and the results revealed a decreased PGRMC1-USP14 binding affinity following circKIAA1617 knockdown, whereas circKIAA1617 overexpression enhanced this interaction (Fig. 6. H, I). Full-length and truncated USP14 vectors were overexpressed with PGRMC1 to further explore the binding domains between PGRMC1 and USP14, and IP assays revealed that the C-terminal catalytic domain (CAT, 105–494 aa) of USP14 directly binds to PGRMC1 (Fig. 6. J). Moreover, the full-length and truncated PGRMC1 vectors mentioned in Fig. 5 together with the USP14 overexpression vector were transfected into HEK-293T cells, and USP14 specifically bound to the N-terminal domain (1–72 aa) of PGRMC1, as evidenced by IP assays (Fig. 6. K).
USP14 was silenced after circKIAA1617 overexpression in MCF7 and T47D cells to further investigate whether USP14 is a functional downstream target of circKIAA1617 in ER-positive BC cells. As shown in Figure S7. D, E, circKIAA1617 overexpression did not affect the mRNA or protein levels of USP14, whereas USP14 knockdown decreased the PGRMC1 protein level induced by circKIAA1617, indicating that USP14 was crucial for circKIAA1617-regulated PGRMC1 expression. In vitro assays demonstrated that the downregulation of USP14 inhibited the effects of circKIAA1617 on the autophagy, proliferation and stemness of ER-positive BC cells (Fig. 6. L-P, Figure S7. F, G). In conclusion, our results proved that circKIAA1617 served as a scaffold to enhance the interaction between PGRMC1 and USP14, which promoted autophagy, proliferation and stemness in ER-positive BC cells.

The circKIAA1617/USP14 axis increased the stabilization of PGRMC1 by catalyzing K48-linked deubiquitination at lysine 105
We explored the potential mechanism by which the circKIAA1617/USP14 axis influences PGRMC1 by assessing the PGRMC1 protein and mRNA levels after the knockdown and overexpression of USP14. As shown in Fig. 7. A and Figure S8. A, B, the protein level but not the mRNA level of PGRMC1 was positively correlated with the expression of USP14. Additionally, CHX (cycloheximide) chase assays were performed, which revealed that the knockdown of USP14 decreased the stabilization of the PGRMC1 protein (Fig. 7. B). The proteasome inhibitor MG132 and the lysosome inhibitor CQ (chloroquine) were used to verify whether USP14 influenced the stabilization of PGRMC1 via the ubiquitin-proteasome system or lysosome, and the results revealed that inhibition of the ubiquitin-proteasome pathway blocked the effect of USP14 on PGRMC1 protein expression (Fig. 7. C).

Based on these results, the effects of USP14 and circKIAA1617 on the level of PGRMC1 ubiquitination were evaluated, and we found that both USP14 and circKIAA1617 decreased the ubiquitination level of PGRMC1 (Fig. 7. D, E). Moreover, USP14 overexpression vectors with a deletion of the UBL domain (ΔUBL) or carrying a C114A inactive mutation [29] were constructed (Figure S8. C), and the results showed that the overexpression of the ΔUBL or C114A mutant vector failed to decrease the level of PGRMC1 ubiquitination (Fig. 7. F, G), further confirming the regulatory effect of USP14 on the PGRMC1 protein. Moreover, our results demonstrated that USP14 and circKIAA1617 selectively catalyzed the removal of K48-linked polyubiquitin chains from PGRMC1 but exhibited no activity toward K6, K11, K27, K29, K33 or K63-linked chains (Fig. 7. H, I). In addition, the overexpression of USP14 mutant vectors or the K48-resistant (K48R) form of ubiquitin further proved that USP14 and circKIAA1617 regulated the K48-linked polyubiquitination of the PGRMC1 protein (Figs. 7. J and S8. D).
PGRMC1 was partitioned into three regions (residues 1–71, 72–171, and 172–195), and all lysine residues in each region were mutated (K1R, K2R, K3R) to identify the specific lysine site in the PGRMC1 protein that is regulated by the circKIAA1617/USP14 axis, and we found that USP14 affects mainly the K48-linked polyubiquitination of the cytochrome b5-like domain of PGRMC1 (72–171 aa) (Figure S8. E). Furthermore, the PGRMC1 protein sequence was predicted using GPS-UBER [30] (Figure S8. F), and four high-confidence lysine ubiquitination sites (K96, K102, K105 and K132) clustered within the cytochrome b5-like domain were predicted to be possibly regulated by USP14. Vectors containing mutations of the four lysine sites were constructed (K-R) and cotransfected with the USP14 overexpression vector; the results revealed that lysine 105 of PGRMC1 was the specific site that was deubiquitinated by USP14 (Figure S8. G; Fig. 7. K). Moreover, the ubiquitination levels of the mutant PGRMC1 protein (K105R) were further examined after the overexpression of circKIAA1617 or silencing of circKIAA1617 and USP14, further proving that lysine 105 of the PGRMC1 protein was regulated by the circKIAA1617/USP14 axis (Figure S8. H; Fig. 7. L). In addition, a K48-Ub-specific antibody was used, confirming that USP14 modulated the K48-linked ubiquitination of PGRMC1 at lysine 105 (Figure S8. I). A conservation analysis revealed that lysine 105 was conserved across species, indicating the potential functional role of lysine 105 in the PGRMC1 protein (Figure S8. J). Furthermore, we generated two siRNA-resistant PGRMC1 constructs encoding wild-type PGRMC1 (PGRMC1res-Flag WT) or the K105R mutant (PGRMC1res-Flag K105R) by synonymous mutation (Figure S8. K). The constructs were transfected into cells with endogenous PGRMC1 knockdown, and the effects of circKIAA1617 silencing on PGRMC1res-Flag WT/PGRMC1res-Flag K105R mediated proliferation and stemness were examined. In vitro assays demonstrated that the circKIAA1617 phenotype depended on PGRMC1 stabilization via the deubiquitination of K105 (Figure S8. L-N). Taken together, our results demonstrated that the circKIAA1617/USP14 axis increased the stabilization of the PGRMC1 protein by catalyzing its K48-linked deubiquitination at lysine 105.

CircKIAA1617-induced autophagy modulated lipophagy to facilitate proliferation and stemness
Since we have proven that circKIAA1617 promoted the proliferation and stemness of ER-positive BC by inducing autophagy, we further evaluated how circKIAA1617/USP14/PGRMC1-induced autophagy facilitated the malignant behaviors of ER-positive BC. As shown in Fig. 3. B, in addition to autophagy, we found that the genes whose expression was dysregulated after circKIAA1617 knockdown were also enriched in pathways associated with fatty acid degradation and fatty acid metabolism, indicating that circKIAA1617 might also play crucial roles in regulating lipid metabolism in ER-positive BC. The number of lipid droplets (LDs), which store free fatty acids (FFAs) and total cholesterol (TC) as triglycerides (TGs) in cells, was first determined by staining with BODIPY 493/503 or Oil red O in ER-positive BC cells to verify our hypothesis. As shown in Fig. 8. A-C, the number of LDs in ER-positive BC cells increased significantly after circKIAA1617 was silenced. Moreover, circKIAA1617 knockdown increased the level of TGs but decreased the levels of TC and FFAs, indicating that circKIAA1617 might promote the degradation of LDs (Fig. 8. D). Furthermore, the detection of fatty acid oxidation and carnitine palmitoyl transferase 1 A (CPT1A) levels in ER-positive BC cells also proved that circKIAA1617 knockdown blocked FAO in ER-positive BC cells (Fig. 8. D, Figure S9. A). Since we demonstrated the correlations between circKIAA1617, autophagy and lipid metabolism, we further evaluated whether lipophagy, a novel type of autophagy that selectively degrades LDs into FFAs and cholesterol through the lysosome-dependent lipolytic pathway, accounted for these results. As shown in Fig. 8. E, BODIPY was used to stain LDs, and LysoTracker was used to stain lysosomes; circKIAA1617 knockdown inhibited the colocalization of LDs and lysosomes. Furthermore, circKIAA1617 silencing suppressed the colocalization of LDs and LC3, indicating that circKIAA1617 knockdown inhibited the degradation of LDs by lipophagy (Fig. 8. F). Moreover, the effects of circKIAA1617 overexpression on the levels of lipophagy and fatty acid oxidation (FAO) were evaluated, which further confirmed that circKIAA1617 enhanced lipophagy and FAO in ER-positive BC cells (Fig. 8. G, Figure S9. B-G). Additionally, the knockdown of adipose triacylglyceride lipase (ATGL) further proved the effects of circKIAA1617 on lipophagy (Figure S9. H, I). Furthermore, circKIAA1617-overexpressing cells were treated with Baf-A1 to verify whether circKIAA1617-induced autophagy was responsible for lipophagy in ER-positive BC cells, and the results proved that the inhibition of autophagy suppressed LD degradation in ER-positive BC cells (Fig. 8. H, I).

Since we demonstrated that circKIAA1617-induced autophagy enhanced lipophagy and subsequent FAO in ER-positive BC cells, we explored whether lipophagy-induced FAO was required for the proliferation and stemness of ER-positive BC cells. As shown in Figure S9. J, etomoxir was used to inhibit circKIAA1617-induced FAO. The proliferation and stemness of ER-positive BC cells were evaluated, and the results confirmed that the attenuation of FAO could suppress the oncogenic effects of circKIAA1617 (Fig. 8. J, K, Figure S9. K-N), indicating that lipophagy-mediated fatty acid metabolism is necessary for the circKIAA1617-driven phenotype. In conclusion, our results proved that circKIAA1617-induced autophagy enhanced lipophagy and fatty acid oxidation to facilitate the proliferation and stemness of ER-positive BC cells.

Discussion
Estrogen receptor-positive breast cancer consists of the luminal A and luminal B molecular subtypes, accounting for approximately 70–80% of all newly diagnosed breast cancers globally, which makes it the most prevalent and clinically heterogeneous form of the disease [1]. Estrogen overexposure and aberrant activation of the ER signaling pathway are thought to be the main drivers of ER-positive BC [31], influencing key malignant biological behaviors such as proliferation and stemness in ER-positive BC [32]. For instance, Kumar et al. reported that the stabilization of DLL1 by estrogen signaling mediated Notch signaling in breast cancer to promote tumor cell progression [33]. In another study, researchers reported that estrogen promoted cancer stemness in ER-positive BC cells via Gli1 [18]. Accumulating evidence indicates that estrogen regulates tumor progression through effects on noncoding RNAs, notably circRNAs. For example, estrogen-regulated circPGR drives ER-positive breast cancer progression by acting as a competing endogenous RNA to sponge miR-301a-5p and regulate the expression of multiple cell cycle genes [10]. In addition, estrogen increases circFAM171A1 levels in myoblasts, which in turn modulates the oar-miR-485-5p/MAPK15 axis to increase myoblast proliferation and limit apoptosis [34]. However, the regulatory roles and mechanisms of estrogen-induced circRNAs have not been fully elucidated and need further investigation.
In this study, using circRNA-seq screening, bioinformatics analyses and qRT-PCR validation, we identified circKIAA1617 as an estrogen-responsive circRNA whose expression is upregulated in ER-positive BC cells and tissues. In vitro and in vivo studies verified that circKIAA1617 knockdown could inhibit the proliferation and stemness of ER-positive BC cells and that circKIAA1617 overexpression had oncogenic effects, which confirmed the vital role of circKIAA1617 in ER-positive BC. Moreover, circKIAA1617 expression is correlated with the clinicopathological characteristics and poor prognosis of patients with ER-positive BC. Notably, considering that patients with ER-positive breast cancer routinely receive adjuvant endocrine therapy, the association between high circKIAA1617 expression and shorter overall survival suggests that elevated circKIAA1617 levels might indicate potential endocrine resistance. Further functional assays and multicenter validation should be performed to verify the roles and predictive values of circKIAA1617 in endocrine resistance of ER-positive BC. RNA-seq was subsequently performed in MCF7 cells with circKIAA1617 knockdown to explore the downstream pathway regulated by circKIAA1617 and responsible for the oncogenic effects of circKIAA1617. The enrichment analysis indicated that the expression of circKIAA1617 was closely associated with the progression of autophagy in ER-positive BC cells. As a cellular process involved in the degradation and recycling of cellular components, autophagy has been implicated in multiple aspects of cancer development [35]. Previous studies have shown the oncogenic roles of autophagy in various cancers, including the survival, proliferation and stemness of cancer cells. For instance, Kao et al. reported that TFEB- and TFE3-dependent autophagy activation supported cancer cell proliferation in the absence of centrosomes [36]; hypoxia-induced galectin-8 expression maintained stemness in glioma stem cells by regulating autophagy [37]. In the present study, our in vitro and in vivo results confirmed the effect of circKIAA1617 on driving the autophagy of ER-positive BC, and the inhibition of autophagy by Baf-A1 reversed the oncogenic effects of circKIAA1617, indicating that circKIAA1617 promoted ER-positive BC proliferation and stemness via the induction of autophagy.
Given that we have proven that circKIAA1617 is an estrogen-responsive circRNA, we further evaluated the mechanisms underlying estrogen-induced circKIAA1617 biogenesis in ER-positive BC. The binding of estrogen to ERα can recruit coactivators to initiate the transcription of oncogenes [38], further promoting the progression of ER-positive BC. For example, estrogen increases the transcription of ERINA to facilitate the progression of breast cancer [39], and the estrogen-inducible kinase SGK3 promotes the survival of breast cancer cells [40]. We thus hypothesized that circKIAA1617 might be an oncogene that is also transcriptionally activated by estrogen in ER-positive BC, which was further confirmed in our ChIP and luciferase assays. The back-splicing of circRNAs is catalyzed by the canonical spliceosomal machinery and modulated by both intronic complementary sequences (ICSs) and RNA binding proteins (RBPs) [7]; thus, we further explored potential RBPs responsible for the biogenesis of circKIAA1617. EIF4A3 is the core component of the exon junction complex (EJC), which is considered an important regulator of posttranscriptional regulatory processes, including mRNA splicing, transport, translation, and surveillance [41], and has been reported to regulate the biogenesis of oncogenic circRNAs such as circSIPA1L3 and circSEPT9 to facilitate the progression of cancers [42, 43]. Through bioinformatics analysis and experimental verification, we found that EIF4A3 could bind to both the upstream and downstream flanking sequences of circKIAA1617 pre-mRNA and was positively correlated with the expression of circKIAA1617, which proved that EIF4A3 could regulate the biogenesis and cyclization of circKIAA1617. Given that circKIAA1617 is identified as a direct transcriptional target of ERα, the circKIAA1617-mediated oncogenic axis should be broadly applicable across ER-positive breast cancer subtypes, including both Luminal A and Luminal B tumors. It was important to note that the intensity of ER signaling and the landscape of transcriptional co-regulators could vary significantly between subtypes [44], and differences in ER signaling strength might influence circKIAA1617 regulation.
Recently, the ability of circRNAs to bind with proteins has attracted increasing attention, as circRNAs can act as scaffolds that promote the binding of different proteins, further regulating the expression, functions and subcellular locations of downstream proteins [45, 46]. We explored the molecular mechanism by which circKIAA1617 enhances autophagy to promote proliferation and stemness in ER-positive BC by performing RNA pull-down and LC-MS to identify downstream targets of circKIAA1617, which revealed that circKIAA1617 could directly bind to the PGRMC1 protein. PGRMC1 has emerged as a heme-binding protein that mechanistically underpins various cellular and tissue functions, including cytochrome P450 activity, heme homeostasis, protein quality control, female reproduction, and cancer [47]. Previous studies have indicated that PGRMC1 converges as a pancancer modulator that promotes proliferation, stemness and chemoresistance in malignant tumors and that its expression is correlated with shorter overall survival of individuals with cancers [48, 49]. For instance, Zhao et al. reported that PGRMC1 potentiated cell proliferation through the suppression of ferroptosis in triple-negative breast cancer [50], and Guan et al. found that cancer stem cell phenotypes and chemotherapy resistance are increased by PGRMC1 through the modulation of AHR ubiquitination in non-small cell lung cancer [51]. Moreover, studies have indicated that PGRMC1 plays regulatory roles in cancer autophagy. For example, Shakeel et al. reported that PGRMC1 is required for the degradative activity of autophagy [24] and that PGRMC1-mediated autophagy increases the resistance of glioblastoma to radiotherapy [14]. We identified PGRMC1 as a direct functional target of circKIAA1617, which modulates PGRMC1 protein levels without altering its mRNA expression. Notably, PGRMC1 knockdown abolished circKIAA1617-mediated autophagy, proliferation, and stemness, validating the circKIAA1617/PGRMC1 oncogenic axis. We elucidated the potential molecular mechanism by which circKIAA1617 regulates PGRMC1 expression by further analyzing the results of RNA pull-down assays, and USP14 was also shown to be enriched by circKIAA1617 probes. We hypothesized that circKIAA1617 might act as a scaffold to enhance the interaction between PGRCM1 and USP14, as demonstrated in our study. USP14 is a member of the ubiquitin-specific protease (USP) family, which is the largest family of deubiquitylating enzymes and orchestrates various cellular processes through the removal of ubiquitins from target proteins to modulate the stabilization, localization and function of proteins [52, 53]. USP14 plays a critical role in the posttranslational regulation of proteins related to tumorigenesis and development. For instance, Shi et al. reported that USP14 deubiquitinates and upregulates IDO1, enhancing immune suppression to promote colorectal cancer progression [28]; Li et al. revealed a mechanism underlying USP14-mediated stabilization of SDC2 in gastric cancer tissue, which promotes the growth and invasive capability of gastric cancer cells [54]. In the present study, our results proved that the knockdown or overexpression of USP14 could affect the expression and stabilization of PGRMC1, which was mediated by the ubiquitin-proteasome pathway, indicating that circKIAA1617 increased PGRMC1 protein expression via USP14-mediated deubiquitination. Ubiquitination is among the most common and important posttranslational modifications of proteins, and different types of ubiquitination mediate distinct fates of substrate proteins [55]. For instance, K48-linked polyubiquitination usually leads to the proteasome-dependent degradation of proteins, whereas K63-linked polyubiquitination is implicated in signal transduction [55]. In our study, we demonstrated that USP14 increased the stabilization of the PGRMC1 protein by catalyzing the K48-linked deubiquitination of PGRMC1 at lysine 105, which accounted for the role of circKIAA1617 in regulating autophagy in ER-positive BC cells.
Autophagy is a complex process that can be divided into several distinct types depending on the specific cellular components targeted for degradation, including mitophagy, lipophagy, and ER-phagy, each of which plays crucial roles in maintaining cellular homeostasis and cancer progression [56]. In this study, we further explored which aspect of cellular progression was influenced by circKIAA1617/USP14/PGRMC1-mediated autophagy to facilitate the proliferation and stemness of ER-positive BC. Based on the KEGG enrichment results from circKIAA1617 knockdown cells, we noticed that pathways correlated with lipid metabolism reprogramming were also significantly dysregulated in circKIAA1617-silenced cells. Notably, lipid metabolic reprogramming refers to adaptive changes in lipid synthesis, uptake, storage, and breakdown in cells, and it plays crucial roles in the regulation of tumor cell fate [57]. The detection of LDs, TG, TC, FFAs and FAO indicated that circKIAA1617-induced autophagy might be associated with the degradation of LDs and increased fatty acid metabolism. Recent studies have shown that autophagy intersects with lipid metabolism in tumors by driving lipid droplet catabolism and fatty acid release, playing a pivotal role in orchestrating the metabolic mechanisms underlying tumor progression [58, 59]. Lipophagy, an autophagy-lysosome-dependent process that mobilizes intracellular lipid droplets into cholesterol and FFAs for β-oxidation, plays critical roles in the pathogenesis and progression of various diseases [60–63]. In tumor cells, lipophagy is a dynamically regulated process governed by complicated molecular networks to alter the malignant biological behavior of cancers. For example, lipophagy facilitates the generation of FFAs to promote senescence in prostate cancer [64], and lipophagy mediated by the AURKA/DDX5/TMEM147-AS1/let-7 signaling axis promotes cisplatin resistance in epithelial ovarian cancer [65]. Moreover, PGRMC1, the downstream functional target of circKIAA1617, has been reported to be associated with lipophagy in cancer cells [66]. In our study, in vitro and in vivo experiments were performed, and we confirmed that circKIAA1617-induced autophagy enhanced lipophagy to facilitate the proliferation and stemness of ER-positive BC cells. Moreover, etomoxir was used to specific inhibite carnitine palmitoyltransferase 1 (CPT1), which serves as the rate-limiting enzyme for mitochondrial β-oxidation [67], further proving that circKIAA1617 associated lipophagy facilitated proliferation and stemness of ER-positive BC cells via enhancing FAO. Crucially, this process established a hierarchical axis where lipophagy-dependent mobilization of lipid droplets provided essential substrates to fuel fatty acid oxidation, thereby positioning FAO as a downstream metabolic effector that sustained the energetic demands of tumor proliferation and stemness.
Our findings highlighted a mechanistic paradigm that was distinct from the majority of previously reported autophagy-regulating circRNAs [68–70]. Firstly, circKIAA1617 was an estrogen responsive circRNA that facilitated autophagy in ER-positive BC, which could partly explicate the autophagy-promoting roles of estrogen reported in tumors [71, 72]. Moreover, our study demonstrated that circKIAA1617 enhanced autophagy via USP14/PGRMC1 axis, which specifically targeted LDs and promoted lipid metabolism reprogramming. It is important to note that our study establishes the necessity of the FAO pathway, as its inhibition effectively reversed circKIAA1617-induced malignant phenotypes. However, whether lipophagy activation is sufficient on its own to drive these behaviors warrants further investigation, and other circKIAA1617-mediated mechanisms may also contribute to tumor progression.

Conclusions
In summary, we demonstrated that circKIAA1617 promoted the proliferation and stemness of ER-positive BC cells by inducing autophagy. Mechanistically, circKIAA1617 transcription was upregulated by estrogen and it was cyclized by EIF4A3. Moreover, circKIAA1617 functioned as a scaffold to enhance the interaction between PGRMC1 and USP14, enabling the K48-linked deubiquitination of PGRMC1 at lysine 105 to stabilize its expression, which accounted for circKIAA1617-induced autophagy, proliferation, and stemness of ER-positive BC cells. In addition, we demonstrated that autophagy induced by the circKIAA1617/USP14/PGRMC1 axis modulates lipid metabolic reprogramming in ER-positive BC by enhancing lipophagy. Clinical analyses revealed that circKIAA1617 is an independent prognostic biomarker correlated with a poor prognosis for patients with ER-positive BC. Collectively, our findings provide mechanistic insights into ER-positive BC proliferation and stemness through the circKIAA1617/USP14/PGRMC1 axis, positioning it as a potential diagnostic biomarker for the treatment of patients with ER-positive BC (Fig. 9).

Supplementary Information