The novel β-TrCP protein isoform hidden in circular RNA confers trastuzumab resistance in HER2-positive breast cancer

Trastuzumab notably improves the outcome of human epidermal growth factor receptor 2 (HER2)-positive breast cancer patients, however, resistance to trastuzumab remains a major hurdle to clinical treatment. In the present study, we identify a circular RNA intimately linked to trastuzumab resistance. circ-β-TrCP, derived from the back-splicing of β-TrCP exon 7 and 13, confers trastuzumab resistance by regulating NRF2-mediated antioxidant pathway in a KEAP1-independent manner. Concretely, circ-β-TrCP encodes a novel truncated 343-amino acid peptide located in the nucleus, referred as β-TrCP-343aa, which competitively binds to NRF2, blocks SCFβ-TrCP-mediated NRF2 proteasomal degradation, and this protective effect of β-TrCP-343aa on NRF2 protein requires GSK3 activity. Subsequently, the elevated NRF2 transcriptionally upregulates a cohort of antioxidant genes, giving rise to trastuzumab resistance. Moreover, the translation ability of circ-β-TrCP is inhibited by eIF3j under both basal and oxidative stress conditions, and eIF3j is transcriptionally repressed by NRF2, thus forming a positive feedback circuit between β-TrCP-343aa and NRF2, expediting trastuzumab resistance. Collectively, our data demonstrate that circ-β-TrCP-encoded β-TrCP protein isoform drives HER2-targeted therapy resistance in a NRF2-dependent manner, which provides potential therapeutic targets for overcoming trastuzumab resistance.


Introduction
Breast cancer is the most common malignancy among women, the number of new cases and deaths of breast cancer ranks first in the world, accounting for 25% of the total female cancer cases and 17% of the cancer deaths [1].As a special subtype, human epidermal growth factor receptor 2 (HER2)-positive breast cancer accounts for 15-20% of all breast cancer, which is characterized by aggressive behaviors and poor prognosis [2].Trastuzumab, a humanized monoclonal antibody, exhibits considerable anti-tumor efficacy in both preclinical and clinical trials [3].It is initially approved by the FDA for treatment of HER2-positive breast cancer in 1998, and patients receiving trastuzumab treatment have prolonged overall survival time of nearly five months [4].However, a high proportion of patients eventually develop resistant to trastuzumab [5], which enormously limits its clinical application.Therefore, an improved understanding of the potential mechanism of HER2-targeted therapy resistance will allow identification of new therapeutic strategies to overcome resistance.
The adaptation of tumor cells to oxidative stress is responsible for the insensitivity to anti-tumor drugs [6].NRF2, a well-known basic leucine zipper transcription factor, acts as a master regulator of redox homeostasis via governing the expression of hundreds of genes involved in cell defense against oxidative stress [7].It is clear that NRF2 protein turnover is tightly controlled by the ubiquitin-proteasome system involving KEAP1-Cul3-Rbx1 complex in the cytoplasm and SKP1-Cul1-Rbx1 (SCF β-TRCP ) complex in the nucleus [8].β-TRCP directly binds to NRF2 through its WD domain, while recruits SKP1-Cul1-Rbx1 complex through its F-box domain, resulting in NRF2 ubiquitination and degradation, and the binding of β-TRCP to NRF2 requires the phosphodegron created by GSK-3 [9].Extensive evidence suggests that NRF2 is frequently overexpressed in human cancer and contributes to drug resistance by preventing oxidative stress [10].Recent studies showed that HER2-amplified patients with high NRF2 expression had worse progression-free survival following trastuzumab treatment [11].And brusatol, a novel Nrf2 inhibitor, was capable to synergistically potentiate the anti-tumor activity of trastuzumab in HER2-positive cancers [12], indicating that NRF2 is critical for trastuzumab resistance.Circular RNA is a type of single strand endogenous closed loop RNA that lacks 3'-end poly (A) tail and 5'-end cap structure [13].It is mainly produced by precursor mRNA (pre-mRNA) through back splicing processing, and is highly abundant in eukaryotes, with some circRNAs being expressed at more than 10-fold higher levels than their parental linear mRNAs [14].Due to the limitations of detection technology, circRNA is initially considered as transcriptional "noise" and has long been overlooked by researchers.Nowadays, circRNA is widely recognized as a pivotal regulator in pathological and physiological changes [15,16].The mechanism of action of circRNA is complex, including acting as micro-RNA sponge, regulating gene expression and selective splicing, and binding to various proteins [17].Of note, genome-wide translatomics suggests that circRNA is translatable in human cells [18], and a portion of circRNAs have been proven to be directly translated into the functional proteins, participating in cancer occurrence, development and progression [19,20].For instance, the novel 370-amino acid β-catenin isoform encoded by circ-β-catenin promoted hepatocellular carcinoma growth through stabilizing full-length β-catenin protein [21].The micropeptide E-Cad-254aa translated by circ-E-Cad maintained glioma stem cell tumorigenicity through activation of EGFR-STAT3 signaling [22].In addition, circ-AXIN1-encoded AXIN1-295aa competitively interacted with APC, resulting in activation of Wnt signaling pathway in gastric cancer cells [23].These studies indicate that the previously unrecognized endogenous micropeptides translated by circRNAs harbor important biological functions, but whether they are involved in trastuzumab resistance remains unknown.Herein, we found a circular RNA produced by β-TrCP promoting trastuzumab resistance via encoding a novel β-TrCP isoform, β-TrCP-343aa.Further investigations revealed that β-TrCP-343aa functioned as a stabilizer of NRF2 protein through blunting SCF β-TRCP -mediated NRF2 ubiquitination degradation.Moreover, we found that the translation efficiency of circ-β-TrCP was altered in response to oxidative stress, a process controlled by the NRF2/eIF3j axis.

Tissue samples
The fresh surgically excised specimens were obtained from The Affiliated Shaanxi Fourth People Hospital of Peihua University.HER2positive breast cancer was determined by immunohistochemistry (IHC) staining, and fluorescence in situ hybridization (FISH) was required when IHC result was intermediate staining intensity.We collected a total of 80 tissues, of which 30 were sensitive to trastuzumab and 50 were insensitive.The detailed clinicopathological information is described in Table S1.Resistance to trastuzumab was defined as progression at the first radiological reassessment within 8-12 weeks or 3 months after first-line trastuzumab with or without chemotherapy, or as a new diagnosed recurrence within adjuvant trastuzumab therapy for 12 months [24].In addition, 15 trastuzumab-sensitive and 18 trastuzumab-resistant plasma specimens were also obtained from above patients.All patients were routinely followed up and provided written informed consent.And this study was conducted in line with the Declaration of Helsinki, and approved by the Ethics Committee of The Affiliated Shaanxi Fourth People Hospital of Peihua University.

Cell lines
HEK-293T and HER2-positive breast cancer cell lines BT474 and SKBR3 were obtained from Cell Bank of Chinese Academy of Sciences.Cells were maintained in DMEM medium supplemented with 10% fetal bovine serum (Gibco, USA) and tested for mycoplasma contamination every 3 months.Establishment of cell lines resistant to trastuzumab was conducted by gradually increasing trastuzumab concentrations as described in our previous study [25].Trastuzumab-resistant BT474 (BT474-TR) and SKBR3 (SKBR3-TR) cells were cultured in DMEM medium supplemented with 15 μg/mL trastuzumab.For the functional assays, trastuzumab was removed for at least one week to avoid acute effects.

Quantitative reverse-transcription PCR (qRT-PCR)
Total RNA was isolated by 1 mL Trizol reagent (Invitrogen, USA), followed by measurement of RNA concentration using NanoDrop™ spectrophotometers (Thermo Scientific, USA). 1 μg RNA was reverse transcribed to cDNA using PrimeScript RT Master Mix (Takara Bio, Japan) as per manufacturer's instructions.Then, cDNA was diluted 5fold, nucleic acid amplification and quantification were conducted using SYBR Premix Ex Taq Kit (Takara Bio) and ABI7900 fast real-time PCR system (Applied Biosystems, USA).Gene expression was determined by 2 − ΔΔCt formula, and ACTB gene was used as the reference control.The primer sequences are listed in Table S2.

Northern blot
The assay was conducted by using NorthernMax™ Kit (Thermo Scientific) according to manufacturer's instructions with minor modifications.In brief, 10 μg total RNA was treated with 5 U/μg RNase R (Epicentre Technologies, USA) at 37 • C for 20 min, separated on 1.5% denaturing agarose gel and transferred to hybond-N + membrane.Then, the membrane was hybridized with biotin-labeled probe targeting circβ-TrCP junction site at 58 • C overnight with gentle rotation.The next day, the membrane was rigorously washed and incubated with HRPlinked streptavidin for 30 min at room temperature.The signal on the membrane was detected using Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Scientific).The oligonucleotide probe sequences used in northern blot assay are listed in Table S2.

Detection of circ-β-TrCP subcellular localization
Isolation of cytoplasmic and nuclear RNA was carried out using Ambion® PARIS™ Kit (Thermo Scientific) as per manufacturer's instructions.ACTB and U6 were used as cytoplasmic and nuclear control fragments, respectively.For FISH assay, the Cy3-labeled probe against circ-β-TrCP junction site was designed and synthesized by GenePharma (Shanghai, China), followed by incubation with cell lysates.The assay was conducted using RNA FISH Kit (GenePharma) as per manufacturer's instructions.Cell nucleus was stained by DAPI dye, and the fluorescent signals were observed under a fluorescent microscope.The oligonucleotide probe sequences used in FISH assay are listed in Table S2.

Vectors, siRNAs and transfection
The CRISPR/Cas13d system [26] was used to establish circ-β-TrCP knockdown cells in vivo.Three sgRNAs targeting circ-β-TrCP junction site were designed and inserted into pLKO.1 vector containing direct repeats of RfxCas13d, followed by lentivirus packaging using psPAX2 and pMD2.G vectors.The stable cell lines were screened by 1.5 μg/mL puromycin.To overexpress circ-β-TrCP, the full-length of circ-β-TrCP was synthesized and inserted into pLV-circ-Puro plasmid with two reverse complementary sequences, followed by lentivirus packaging and infection into cells.To overexpress genes, the full-length coding sequences of human eIF3j, β-TrCP, NRF2, β-TrCP-343aa were cloned into pcDNA 3.1, pFlag-CMV-5a and pCMV-HA plasmids as appropriate.The various mutated vectors were prepared using Q5® Site-Directed Mutagenesis Kit (New England Biolabs, UK) as per manufacturer's protocols.To construct NRF2 knockout cells, the validated sgRNA targeting NRF2 was inserted into lenti-CRISPR v2 plasmid, followed by lentivirus packaging using psPAX2 and pMD2.G vectors and infection into cells.The stable cell lines were screened by 1.5 μg/mL puromycin.The knockout efficiency was verified by Sanger sequencing and Western blot assay.To generate endogenous eIF3j antioxidant response element (ARE)-mutated cells, the designed sgRNA with the lowest rate of off-target score was synthesized and ligated into pSpCas9(BB)-2A-Puro plasmid, followed by electroporation along with single-stranded oligodeoxynucleotide (ssODN) into SKBR3 cells using Lonza Amaxa 4D-Nucleofector as per the manufacturer's instructions.The limited dilution method was used to obtain individual colony in 96-well plates.The mutated efficiency was verified by Sanger sequencing.All siRNAs were designed and synthesized by Sigma-Aldrich, followed by transfection into cells using Ribo-Juice siRNA Transfection Reagent (Sigma-Aldrich, USA).And plasmid transfection was conducted using Lipofectamine 3000 (Invitrogen).The sgRNA, siRNA and ssODN sequences used in this study are listed in Table S2.

CCK-8, colony formation and flow cytometry
Cell viability was determined by CCK-8 assay.Equal amounts of cells were seeded into 96-well plates, and when grown to 70-80% confluence, 10 μL CCK-8 reagent was added into the medium.After incubation for 2 h at 37 • C, the absorbance at 450 nm in each well was detected by an automatic microplate spectrophotometer.For colony formation assay, 3000 cells were seeded into 6-well plates, and cultured for 2 weeks.Then, cells were fixed by methanol and stained by crystalline violet.ROS content was detected by H2DCFDA probe (MedChemExpress, USA) and CellROX Deep Red reagent (Invitrogen).In brief, equal amounts of cells were seeded into 6-well plates, and when grown to 70-80% confluence, cells were washed by PBS and incubated with 20 μM H2DCFDA probe at 37 • C for 1 h.Then, cells were collected and resuspended in 300 μL PBS, followed by analysis using Accuri™ C6 Plus flow cytometer (BD Biosciences, USA).For CellROX staining, cells were plated into 96-well plates, followed by incubation with 5 μM CellROX Deep Red reagent at 37 • C for 30 min.After washing twice with PBS, the fluorescence signals were detected by using the CLARIOstar® Plus microplate reader (BMG LABTECH, Germany).

Orthotopic transplantation tumor model
A total of 1 × 10 7 circ-β-TrCP-silenced BT474-TR cells or circ-β-TrCPoverexpressed BT474 cells mixed with 1:1 matrigel (Corning, USA) were orthotopically injected into the abdominal mammary fat pad of NOD/ SCID mice grown under specific-pathogen-free condition.When the tumor grew to 0.05-0.1 cm 3 in size, mice were treated with vehicle or trastuzumab (20 mg/kg, intraperitoneal administration) once a week.Tumor volume was recorded using a vernier caliper and calculated using the formula: 1/2 × length × width 2 , no mice died during the experiment.At the end of the fourth week, all mice were sacrificed and tumor tissues were collected, weighed and photographed.The animal study was approved by the Animal Care and Use Committee of Peihua university.

Western blot, co-immunoprecipitation (Co-IP) and IHC staining
The protocols were as previously described [27].Briefly, cells were trypsinized and washed twice with ice-cold PBS, followed by treatment with lysis buffer containing 0.5 M Tris-HCl, 1.5 M NaCl, 10% NP-40, 10 mM EDTA, 2.5% deoxycholic acid, and protease inhibitor cocktail (Roche, USA).Total protein was quantified using Pierce Rapid Gold BCA Protein Quantitative Kit (Thermo Fisher Scientific) and loaded to SDS-PAGE gel and transferred to PVDF membrane.After blocking with 5% nonfat milk, the membrane was incubated with the corresponding primary and secondary antibodies, and developed using SuperSignal West Atto reagent (Thermo Fisher Scientific).For Co-IP assay, the corresponding affinity gels or antibodies and protein A/G magnetic beads (Thermo Fisher Scientific) were added into cell lysates and incubated for 5 h at 4 • C with gentle rotation.The beads were extensively washed and resuspended in Laemmli buffer, followed by Western blot assay.For in vivo ubiquitylation assay, cells were treated with 20 μM MG132 (Med-ChemExpress) before collection of cell lysates, followed by Co-IP assay.
IHC staining was carried out using SP Rabbit & Mouse HRP Kit (DAB) (Cowin, China) according to manufacturer's instructions.Assessment of IHC staining was conducted by using H-score method as previously described [28].The antibodies used in this study are listed in Table S3.

RNA pull-down and immunoprecipitation (RIP)
For in vitro pull-down assay, the linear circ-β-TrCP was in vitro transcribed and labeled with biotin, followed by circularization using T4 RNA ligase I (New England Biolabs).Then, the purified eIF3j protein (#Ag0678, Proteintech, USA) was incubated with circ-β-TrCP probe, followed by Western blot assay.For in vivo pull-down assay, the biotinconjugated probe against the junction site of circ-β-TrCP was synthesized and incubated with cell lysates at 4 • C for 5 h.Then, the streptavidin magnetic beads (Invitrogen) were added and incubated for another 1 h.The beads were extensively washed and resuspended in Laemmli buffer, followed by Western blot assay.The oligonucleotide probe sequences are listed in Table S2.RIP assay was conducted using Magna RIP Kit (Millipore, USA) according to manufacturer's instructions, followed by qRT-PCR analysis of circ-β-TrCP enrichment.

Luciferase reporter assay
BT474-TR and SKBR3-TR cells were transfected with 500 ng ARE-Firefly luciferase or pGL3-basic-eIF3j-promoter plasmid along with 100 ng pRL-CMV Renilla luciferase reporter using Lipofectamine 3000.After 48 h, the luciferase activity was detected by the dual-luciferase reporter system (Promega) according to manufacturer's instructions.

DNA pull-down assay
BT474-TR and SKBR3-TR cells were trypsinized and washed twice with ice-cold PBS, the nuclear protein was isolated using NE-PER Nuclear and Cytoplasmic Extraction reagents (Thermo Fisher Scientific).Then, the human eIF3j promoter oligonucleotides containing wild-type or mutant ARE were synthesized and labeled with biotin, followed by annealing to form double-stranded oligonucleotides.The above nuclear extracts were incubated with biotinylated DNA probes at 4 • C overnight with gentle rotation, followed by incubation with streptavidin magnetic beads (Invitrogen) for another 1 h.Then, the beads were extensively washed and resuspended in Laemmli buffer, followed by Western blot assay.

Chromatin immunoprecipitation (ChIP)
ChIP assay was performed by using the commercialized Sim-pleChIP® Plus Sonication Chromatin IP Kit (Cell Signaling Technology) following manufacturer's protocols.Briefly, BT474-TR and SKBR3-TR cells were cross-linked using 1% formaldehyde, and glycine was added to quench formaldehyde.Cells were then sonicated to obtain chromatin with an average size of 200~1000 bp, followed by centrifugation at 20,000×g for 15min.The supernatant was collected and incubated with anti-NRF2 antibody and protein A/G magnetic beads (Thermo Fisher Scientific) at 4 • C overnight with gentle rotation.The next day, the beads were washed and the precipitated DNA fragments were eluted and analyzed by PCR assay.The primer sequences are listed in Table S2.

Statistical analysis
Data were mean ± standard deviation (SD) of at least three independent experiments carried out in triplicate.Comparisons between the two groups were tested by unpaired t-test.Differences among three or more groups were tested by 1-way ANOVA or 2-way ANOVA, followed by Tukey's post hoc test.The survival curve was plotted by Kaplan-Meier curve and tested by log-rank test.The diagnostic efficiency of plasma circ-β-TrCP was tested by receiver operating characteristic (ROC) curve and calculated by area under curve (AUC) value.The production of charts and statistics was completed by using GraphPad Prism 8. P < 0.05 was considered statistically significant.All P values were two-sided unless otherwise specified.

circ-β-TrCP is upregulated in trastuzumab-resistant HER2-positive breast cancer
In our previous study [25], we screened for differential expression of circular RNAs between BT474-TR and BT474 cells.The top 10 upregulated circular RNAs (Fig. 1A) in BT474-TR cells were validated using qRT-PCR assay, the results showed that circ-β-TrCP had the largest up-regulation ratio (Fig. 1B), thus we chose it for follow-up study.Consistently, circ-β-TrCP was overexpressed in human trastuzumab-resistant breast cancer tissues (Fig. 1C).Patients expressing high circ-β-TrCP had shorter overall survival time than those expressing low circ-β-TrCP (Fig. 1D).And high circ-β-TrCP was also observed in SKBR3-TR cells compared to the parental SKBR3 cells (Fig. 1E).Sequence alignment results showed that circ-β-TrCP is derived from the back-splicing of β-TrCP pre-mRNA exon 7 and 13, the mature full-length is 913 bp, which was further verified by Sanger sequencing (Fig. 1F) and northern blot (Fig. 1G).After treatment with 5 U/μg RNase R, β-TrCP mRNA expression was dramatically reduced, while circ-β-TrCP expression was almost unchanged in both trastuzumab-sensitive and -resistant cells (Fig. 1H-K).The half-life of circ-β-TrCP exceeded 24 h, while that of β-TrCP mRNA was less than 8 h (Fig. 1L and M), suggesting that circ-β-TrCP is highly stable.Next, we explored the subcellular localization of circ-β-TrCP, the results of qRT-PCR and FISH showed that circ-β-TrCP was mainly located in the cytoplasm of both trastuzumab-sensitive and -resistant cells (Fig. 1N-R), indicating that the localization of circ-β-TrCP was not altered during trastuzumab resistance.Moreover, circ-β-TrCP was also detected in the plasma, and its expression was unaffected by acid-base imbalance, room temperature placement and repeated freeze-thaw (Fig. 1S-U).And plasma circ-β-TrCP expression was approximately 4-fold higher in trastuzumab-resistant patients than in sensitive patients (Fig. 1V), with an AUC value of 0.9037 (95% confidence interval (CI): 0.7998-1.000)(Fig. 1W), implying that plasma circ-β-TrCP may be an excellent indicator for evaluating patients' response to trastuzumab.Taken together, these data demonstrate that circ-β-TrCP is a cytoplasmic circular RNA that may be essential for trastuzumab resistance.
The orthotopic transplantation tumor model was established to assess the in vivo effect of circ-β-TrCP on trastuzumab resistance (Fig. 2L).BT474-TR cells were orthotopically injected into the abdominal mammary fat pad of NOD/SCID mice, followed by a weekly dosing of trastuzumab (Fig. 2L).All mice survived at the end of the treatment, and no mouse exhibited severe loss of body weight (>15%) or evidence of infections or wounds (Fig. 2M).The results showed that BT474-TR-derived tumors were highly resistant to trastuzumab (Fig. 2N-P), however, tumor was evidently smaller after knockdown of circ-β-TrCP (Fig. 2N-P).In contrast, BT474-derived tumor size was reduced following trastuzumab administration (Fig. 2Q-U), while overexpression of circ-β-TrCP significantly attenuated the therapeutic effect of trastuzumab (Fig. 2Q-U).In all, these in vitro and in vivo results suggest that circ-β-TrCP is a driver of trastuzumab resistance.
Studies have shown that some translation initiation factors and N6methyladenosine (m 6 A) regulators are responsible for circRNA translation [30].We then used RNAi screening to identify the factors governing the translation process of circ-β-TrCP.The results of qRT-PCR and Western blot showed that knockdown of eIF3 subunits evidently affected β-TrCP-343aa production, while had little effects on the expression of circ-β-TrCP (Fig. 3P), hinting that eIF3 is involved in circ-β-TrCP translation, but not circ-β-TrCP biogenesis.Specifically, silencing of eIF3a and eIF3b decreased, while silencing of eIF3j increased β-TrCP-343aa levels (Fig. 3P), which is consistent with the findings showing that eIF3j repressed circRNA translation through inhibiting translation initiation via blocking the binding of eIF3a and eIF3b to circRNA [31].And the increased β-TrCP-343aa caused by circ-β-TrCP overexpression was effectively blocked by eIF3j overexpression in both BT474-TR and SKBR3-TR cells (Fig. 3Q).The in vitro binding assay was conducted using synthetic full-length circ-β-TrCP and purified eIF3j protein, the results showed that circ-β-TrCP directly binds to eIF3j (Fig. 3R and S).Furthermore, the endogenous interaction between circ-β-TrCP and eIF3j protein was verified by RIP and RNA pull-down assays (Fig. 3T and U).As shown in Fig. 3V, eIF3a,eIF3b and eIF3j proteins were all pulled by circ-β-TrCP probes in both BT474-TR and SKBR3-TR cells, whereas less or almost no eIF3a and eIF3b proteins were observed to be interacted with circ-β-TrCP in the presence of eIF3j overexpression.These results indicate that circ-β-TrCP translation is negatively regulated by eIF3j.

(caption on next page)
S. Wang et al.
Through analyzing the NRF2 ChIP-seq data in ENCODE datasets [32], we found that numerous NRF2 peaks are located in the promoter region of eIF3j (Fig. 5K).Further, one ARE is found at − 21 ~ − 11 region in eIF3j promoter using JASPAR online tool [33](Fig.5L).We then mutated the ARE and conducted luciferase reporter assay, the results showed that NRF2 knockout significantly increased wild-type eIF3j promoter activity, while did not affect the mutant one (Fig. 5M).Likewise, NRF2 protein in nucleus extracts of BT474-TR and SKBR3-TR cells was enriched by wild-type eIF3j promoter probe, but not by the mutated probe (Fig. 5N).Next, we designed three pairs of primers targeting the indicated regions in eIF3j promoter (P1 and P2 were used as negative qRT-PCR analysis of circ-β-TrCP expression in plasma after being left at room temperature for different times, acid and alkali treatment and repeated freeze-thawing.V. qRT-PCR analysis of circ-β-TrCP expression in plasma from trastuzumab-sensitive (n = 15) and -resistant (n = 18) HER2-positive breast cancer patients.W. ROC curve testing the diagnostic efficiency of plasma circ-β-TrCP, analyzed by AUC value.Two-tailed ***P < 0.001, ****P < 0.0001.Unpaired t-test was used for B, C, E, H-K, V. 1-way ANOVA followed by Tukey's post hoc test was used for L, M, S-U.Log-rank test was used for D. TR = trastuzumab resistance, ctrl = control, BSJ = back splicing junction.
S. Wang et al. controls, P3 contained the ARE) (Fig. 5L), followed by ChIP assay to test the binding of NRF2 to eIF3j promoter in vivo.As shown in Fig. 5O, the fragments pulled down by NRF2 were only amplified by P3, but not by P1 and P2.And the qRT-PCR results showed that NRF2 enrichment on ARE motif in eIF3j promoter was nearly 4-fold higher than IgG (Fig. 5P).To further confirm that this ARE is critical for NRF2 binding to eIF3j promoter, we employed CRISPR/Cas9 genome editing and generated endogenous eIF3j ARE-mutated SKBR3 cell lines by mutating six nucleotides from GTGACTCACCT to GATGATCATAT (Fig. S3A).As shown in Fig. S3B, mutation of eIF3j ARE abolished the binding of NRF2 to eIF3j promoter, even when NRF2 was overexpressed.And NRF2 was not able to reduce eIF3j expression in eIF3j ARE-mutated cells (Fig. S3C).Moreover, trastuzumab treatment remarkably increased the occupation of NRF2 on eIF3j promoter (Fig. 5Q and R).Functionally, overexpression of circ-β-TrCP did not affect cell viability following trastuzumab treatment in NRF2 − /− cells (Fig. 5S), and the enhanced cell viability and  volume and weight in the indicated four groups.P. HEK-293T cells were transfected with siRNA targeting the indicated genes, followed by qRT-PCR and Western blot detecting circ-β-TrCP and β-TrCP-343aa expression, respectively.Q.Western blot detecting β-TrCP-343aa and eIF3j protein levels in circ-β-TrCP-overexpressed BT474-TR and SKBR3-TR cells transfected with eIF3j expressing vector.R. The diagram showing the in vitro synthesis of full-length of circ-β-TrCP.S. RNA pull-down coupled with Western blot testing the in vitro binding of circ-β-TrCP to eIF3j.T. RIP assay using anti-eIF3j antibody, followed by qRT-PCR analysis of circ-β-TrCP enrichment.U.The diagram showing the design of circ-β-TrCP probe targeting the junction site of circ-β-TrCP, followed by qRT-PCR analysis verifying the enrichment efficiency of circ-β-TrCP probe.V. RNA pull-down assay in eIF3j-overexpressed BT474-TR and SKBR3-TR cells using circ-β-TrCP probe, followed by Western blot analysis of protein expression of eIF3j, eIF3a and eIF3b.Two-tailed *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.1-way ANOVA followed by Tukey's post hoc test was used for C, K-M, N (right panel), O (right panel), P. Unpaired t-test was used for I, T, U. 2-way ANOVA followed by Tukey's post hoc test was used for N (left panel), O (left panel).BSJ = back splicing junction, OE = overexpression, Mut = mutation, NC = negative control, TR = trastuzumab resistance, ctrl = control, Tras = trastuzumab.colony formation caused by circ-β-TrCP overexpression in the presence of trastuzumab were significantly abolished by eIF3j introduction (Fig. 5T and U), indicating that circ-β-TrCP promotes trastuzumab resistance via the NRF2/eIF3j axis.
Altogether, these data suggest that NRF2 accelerates the translation process of circ-β-TrCP via transcriptionally repressing of eIF3j, thus forming a positive regulatory loop between β-TrCP-343aa and NRF2, expediting trastuzumab resistance.

Discussion
The emergence of drug resistance has greatly limited the clinical use of trastuzumab.In this study, we found that circ-β-TrCP plays an essential role in trastuzumab resistance.Genetic intervention of circβ-TrCP expression significantly altered the sensitivity of breast cancer cells to trastuzumab in vitro and in vivo.343aa, that competes with β-TrCP to bind to NRF2 in the nucleus, thereby blocking the ubiquitination degradation of NRF2 mediated by SCF β-TrCP E3 complex (Fig. 6).Moreover, circ-β-TrCP translating into β-TrCP-343aa is inhibited by eIF3j under both normal and oxidative stress conditions, and NRF2 is able to directly bind to eIF3j promoter and inhibit its transcription, thus a positive loop is formed between β-TrCP-343aa and NRF2, ultimately accelerating trastuzumab resistance (Fig. 6).Therefore, our work provides new evidence for the importance of the coding function of circular RNA in cell biology, as well as new insights into the mechanisms of trastuzumab resistance in HER2-positive breast cancer patients.
Biomarkers are indicators of disease changes or treatment effects through the detection of proteins and genes contained in human blood, urine and other body fluids or tissues [34].It can be used not only to select therapeutic drugs for tumors, but also to determine the presence or absence of tumors or metastases.Due to the cell-or tissue-specific expression pattern, high conservatism and stability, circular RNA is considered to be an excellent biomarker to monitor tumor occurrence, development and progression [35,36].For example, circ-NEIL3, a TGFβ-repressive circular RNA, was identified as a potential biomarker for colorectal carcinoma liver metastasis [37].Hepatocellular carcinoma patients with high circ-SORE had poor recurrence-free survival, and its tissues content was proposed as a good predictor of sorafenib efficacy [38].Also, circular RNA is demonstrated to be enriched and stable in human peripheral blood [39,40].F-circEA [41], circ-CCAC1 [42], circ-N4BP2L2 [43] and circ-RNF220 [44] were abundantly expressed in human peripheral blood, which could be used as the biomarkers for diagnosis of EML4-ALK-positive non-small cell lung cancer, cholangiocarcinoma, epithelial ovarian cancer and pediatric acute myeloid leukemia, respectively.Herein, we found that HER2-positive breast cancer patients harboring high circ-β-TrCP displayed worse overall survival than those harboring low circ-β-TrCP, indicating that circ-β-TrCP may be a prognostic biomarker for HER2-positive breast cancer patients.More importantly, circ-β-TrCP was also significantly upregulated in plasma from trastuzumab-resistant patients compared to trastuzumab-sensitive patients, with an AUC value of 0.9037, suggesting that plasma circ-β-TrCP has great potential to be used as a non-invasive indicator of trastuzumab response.
NRF2 is a well-known antioxidant that confers cellular resistance to anti-cancer drugs, including trastuzumab [45].Elucidating the regulatory network of NRF2 will shed new light on drug resistance.Under normal physiological conditions, the protein abundance of NRF2 is low, due to rapid degradation by KEAP1 [46,47].However, in response to oxidative stress, KEAP1 undergoes conformational changes and loses NRF2 binding capacity, causing NRF2 accumulation [48].Several factors have been reported to play key roles in disease progression and drug resistance by regulating NRF2 protein level in a KEAP1-dependent or -independent manner, such as ATDC [49], TRIM22 [50] and REDD1 [51].These findings mainly focused on studying the regulation of NRF2 protein turnover in the cytoplasm, however, little is known about its stability regulation in the nucleus, that is where NRF2 exerts the antioxidant effects.It is well documented that nuclear NRF2 protein is degraded by SCF β-TrCP E3 complex [52], but how this process is regulated and whether it involves substantial nuclear accumulation of NRF2 under oxidative stress is not yet clear.In this study, we found a novel β-TrCP isoform, β-TrCP-343aa, translated by circ-β-TrCP, which has the WD domain that binds to NRF2, but has no F-box domain of β-TrCP that binds to SKP1-Cul1-Rbx1 complex and ubiquitinates NRF2.Thus, it was speculated that β-TrCP-343aa inhibits β-TrCP-mediated NRF2 degradation via competitively binding to NRF2.Protein docking model, IP and Western blot assays confirmed this hypothesis that β-TrCP-343aa is a previously unrecognized stabilizer of nuclear NRF2.And under oxidative stress, β-TrCP-343aa is substantially elevated dependent of NRF2, thus a positive feedback loop is formed between them, which can be used to explain why NRF2 rapidly and substantially accumulates in the nucleus in response to oxidative stress.Of note, the phosphodegron created by GSK-3 on NRF2 Neh6 domain is necessary for the binding of β-TrCP to NRF2 [53], a condition that also applies to the interaction between β-TrCP-343aa and NRF2, as illustrated by the loss of β-TrCP-343aa regulation of NRF2 protein stability when the activity of GSK3 was inhibited.Hence, GSK3 activity is a prerequisite for the protective effect of β-TrCP-343aa on nuclear NRF2 protein.
NRF2 is a transcription factor that forms heterodimers with sMaf proteins in the nucleus to recognize the ARE, followed by recruitment of chromatin remodeling complexes and coactivators to transcriptionally active a battery of genes involved in anti-stress responses [54].In addition to regulation of redox homeostasis, NRF2 is also involved in various cellular processes, including metabolic reprogramming, proteostasis, immunity, inflammation, and more [55].Thus, the target gene network of NRF2 is far more complex than we currently know and requires in-depth dissection, which may lead to new insights into disease onset and progression.Here, we identified a novel target of NRF2.NRF2 was capable to directly bind to the ARE located at − 21 ~ − 11 region in eIF3j promoter, thereby inhibiting eIF3j transcription.Mutation of this ARE abolished the transcriptional repressive effect of NRF2 on eIF3j, and more NRF2 occupied the promoter of eIF3j during oxidative stress, suggesting that eIF3j is a bona fide target gene of NRF2 under both basal and oxidative stress conditions.It is worth noting that the NRF2-sMaf heterodimers typically activates gene transcription, thus, this model is not applicable to the regulation of eIF3j by NRF2.A study showed that NRF2-replication protein A1 element (NRE) adjacent to the 3′-end of the ARE recognized by NRF2-RPA1 heterodimers is critical for the transcriptional repressive effect of NRF2 [56], however, no NRE was found flanking ARE in eIF3j promoter.The precise mechanism responsible for BT474-TR and SKBR3-TR cells after treatment with 100 μg/mL cycloheximide for the indicated time.F. Western blot testing NRF2 protein expression in circ-β-TrCPsilenced cells treated with 20 μM sulforaphane, 10 μM chloroquine, 20 μM MG132.G, H. IP assay using anti-NRF2 antibody, followed by Western blot analysis of ubiquitin expression in circ-β-TrCP-silenced BT474-TR and SKBR3-TR cells transfected with the indicated vectors.I.Western blot testing HA protein expression in NRF2 − /− cells transfected with the indicated vectors.J.The cartoon showing the specific protein domains of β-TrCP and β-TrCP-343aa.K. IP assay using anti-SKP1 antibody in BT474-TR and SKBR3-TR cells, followed by Western blot detecting SKP1, β-TrCP and β-TrCP-343aa protein levels.L. The protein docking models of β-TrCP-NRF2 and β-TrCP-343aa-NRF2.M. IP assay using anti-NRF2 antibody in BT474-TR and SKBR3-TR cells, followed by Western blot detecting NRF2 and β-TrCP-343aa protein levels.N. IP assay using anti-Flag affinity gels in BT474-TR and SKBR3-TR cells transfected with the indicated plasmids, followed by Western blot analysis of Flag and HA levels.O, P. IP assay using anti-NRF2 antibody in circ-β-TrCP-silenced cells transfected with the indicated plasmids, followed by Western blot analysis of NRF2, β-TrCP and Flag levels.Q.Cells were transfected with the indicated vectors, followed by Western blot analysis of NRF2, β-TrCP and Flag levels.R. IP assay using anti-NRF2 antibody, followed by Western blot analysis of ubiquitin expression in β-TrCP-overexpressed BT474-TR and SKBR3-TR cells transfected with the indicated vectors.S. NRF2 − /− cells were transfected with the indicated vectors or treated with 10 μM SB216763, followed by Western blot analysis of HA expression.T. IP assay using anti-HA affinity gels in NRF2 − /− cells transfected with HA-NRF2-WT or HA-NRF2-S4A4 (S344A/S347A/S351A/S356A), followed by Western blot analysis of β-TrCP, β-TrCP-343aa and HA levels.Two-tailed *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.1-way ANOVA followed by Tukey's post hoc test was used for B, C. 2-way ANOVA followed by Tukey's post hoc test was used for D, E. OE = overexpression, Mut = mutation, NC = negative control, TR = trastuzumab resistance, CHX = cycloheximide, SFN = sulforaphane, CQ = chloroquine, WT = wild-type.promoter activity in NRF2 − /− BT474-TR and SKBR3-TR cells transfected with wild-type or mutant eIF3j promoter reporters.N. DNA pull-down assay using wild-type or mutant eIF3j promoter probe in BT474-TR and SKBR3-TR cell nucleus extracts, followed by Western blot testing NRF2 protein levels.O, P. ChIP assay using anti-NRF2 antibody, followed by PCR analysis using the above designed primers.Q, R. ChIP assay using anti-NRF2 antibody in BT474 and SKBR3 cells treated with 15 μg/ mL trastuzumab, followed by PCR analysis of ARE enrichment.S. CCK-8 testing cell viability in NRF2 − /− BT474 and SKBR3 cells transfected with circ-β-TrCP expressing plasmid.T, U. Cell viability and colony formation were tested in circ-β-TrCP-overexpressed BT474 and SKBR3 cells transfected with eIF3j expressing plasmid.Two-tailed **P < 0.01, ***P < 0.001, ****P < 0.0001.Unpaired t-test was used for B, D, P, S. 1-way ANOVA followed by Tukey's post hoc test was used for G, J, T, U. 2-way ANOVA followed by Tukey's post hoc test was used for M, Q, R. Ctrl = control, OE = overexpression, NC = negative control, TR = trastuzumab resistance, Mut = mutation, WT = wild-type.NRF2-mediated eIF3j repression remains to be defined and may involve inhibition of RNA Pol II recruitment [57]or cooperation with chromatin-modifying enzymes such as EZH2 [58].
There are several limitations in this work, with the major drawback being that the in vitro and in vivo studies were performed using cell lines and do not truly reflect clinical tumor heterogeneity and the interactions between tumor cells and microenvironment, the use of patient-derived xenograft and organoids models will be helpful.In addition, the sample size we studied is relatively small, which may result in potential bias, further large-scale multi-center studies are needed to determine the diagnostic and prognostic efficacy of circ-β-TrCP.Meanwhile, whether circ-β-TrCP exists in other human fluids such as sweat, urine, saliva, as well as whether it is transported by some vesicles like exosomes, are worthy of in-depth investigation.
Collectively, our data uncover a novel mechanism of trastuzumab resistance driven by circular RNA-encoded peptides, meanwhile provide a proof-of-concept demonstration for a potential strategy to resensitize trastuzumab-resistant HER2-positive breast cancer patients to trastuzumab.

Fig. 1 .
Fig. 1. circ-β-TrCP is highly expressed in trastuzumab-resistant HER2-positive breast cancer.A. CircRNA sequencing showing the top 10 upregulated circRNAs in BT474-TR cells compared to BT474 cells.B. qRT-PCR analysis of these 10 circRNA levels in BT474-TR and BT474 cells.C. qRT-PCR analysis of circ-β-TrCP expression in trastuzumab-sensitive (n = 30) and -resistant (n = 50) HER2-positive breast cancer tissues.D. The survival curve of HER2-positive breast cancer patients with low and high circ-β-TrCP based on median circ-β-TrCP expression.E. qRT-PCR analysis of circ-β-TrCP expression in SKBR3-TR and SKBR3 cells.F. The diagram showing the source of circ-β-TrCP while being verified by Sanger sequencing.G. Northern blot testing the full-length of circ-β-TrCP.H-K.Cells were treated with or without 5 U/μg RNase R, followed by qRT-PCR analysis of circ-β-TrCP and β-TrCP mRNA expression.L, M. Cells were treated with 100 μM Actinomycin D for the indicated time, followed by qRT-PCR analysis of circ-β-TrCP and β-TrCP mRNA expression.N-Q.qRT-PCR analysis of the location of circ-β-TrCP, ACTB and U6 were used as the cytoplasmic and nuclear control fragments, respectively.R. FISH assay testing the location of circ-β-TrCP, DAPI was used to stain nucleus.Scale bar = 25 μm.S-U.

Fig. 2 .
Fig. 2. circ-β-TrCP drives trastuzumab resistance. A. The diagram showing the use of CRISPR/Cas13d system to silence circ-β-TrCP.Scale bar = 25 μm.B. qRT-PCR verifying the silencing efficiency of these designed sgRNAs on circ-β-TrCP in BT474-TR and SKBR3-TR cells.C. CCK-8 testing cell viability after circ-β-TrCP knockdown following 15 μg/mL trastuzumab treatment.D. Colony formation assay testing cell cloning ability after circ-β-TrCP knockdown following 15 μg/mL trastuzumab treatment.E, F. Flow cytometry testing ROS content in circ-β-TrCP-silenced BT474-TR and SKBR3-TR cells treated with 15 μg/mL trastuzumab.G.The diagram showing the vector used for circ-β-TrCP overexpression.H-K. Cloning ability, cell viability and ROS levels were detected in BT474 and SKBR3 cells transfected with circ-β-TrCP overexpressing vector in the presence of 15 μg/mL trastuzumab.L, Q.The diagram showing the establishment of orthotopic transplantation tumor model via mammary fat pad injection of BT474-TR or BT474 cells, followed by intraperitoneal administration of 20 mg/kg trastuzumab once a week for a total of 4 times.M, R. The weight change of mice in the indicated four groups.N-P, S-U.The tumor image, volume and weight in the indicated four groups.Two-tailed *P < 0.05, ***P < 0.001, ****P < 0.0001.1-way ANOVA followed by Tukey's post hoc test was used for B-F, H-K, P, U. 2-way ANOVA followed by Tukey's post hoc test was used for M, O, R, T. BSJ = back splicing junction, NC = negative control, TR = trastuzumab resistance, ctrl = control, denotes BT474 and SKBR3 cells that have not received any treatment, OE = overexpression, Tras = trastuzumab.

Fig. 3 .
Fig. 3. circ-β-TrCP encodes β-TrCP-343aa, which is repressed by eIF3j.A. The diagram showing the full-length sequence of circ-β-TrCP and its predicted encoding peptide.B. The diagram showing the construction of OE-circ-β-TrCP, OE-circ-β-TrCP-Flag and OE-circ-β-TrCP-Flag with mutation of start codon.C. qRT-PCR analysis of circ-β-TrCP expression in HEK-293T cells transfected with the indicated vectors.D. Western blot testing the protein levels of Flag and β-TrCP-343aa in HEK-293T cells transfected with the indicated vectors.E. Coomassie brilliant blue staining of circ-β-TrCP-overexpressed cell lysates, followed by mass spectrometry identification.The black arrow indicates the location where the gel was cut for mass spectrometry.F. The fluorescent signals showing the subcellular localization of circβ-TrCP and β-TrCP-343aa, nucleus was stained by DAPI.Scale bar = 25 μm.G.Western blot testing the protein expression of β-TrCP-343aa in the indicated cells.H.The representative images of IHC staining of β-TrCP-343aa in trastuzumab-resistant and -sensitive HER2-positive breast cancer tissues.The red arrows denote the representative staining.Scale bar = 50 μm.I. Semi-quantitative analysis of β-TrCP-343aa IHC staining using H-score method.J.Western blot testing β-TrCP-343aa protein expression in circ-β-TrCP-silenced BT474-TR and SKBR3-TR cells transfected with the indicated vectors.K-M.Detection of cell viability, colony formation and ROS content in circ-β-TrCP-silenced BT474-TR and SKBR3-TR cells transfected with the indicated vectors following 15 μg/mL trastuzumab treatment.N, O. Tumor Further studies revealed that circ-β-TrCP contributes to trastuzumab resistance in a NRF2-dependent manner.circ-β-TrCP encodes a novel β-TrCP protein isoform, β-TrCP-(caption on next page) S. Wang et al.